Many companies take their first stab at Web marketing by simply building a basic Web site. Once they've dipped their toes in the water, they eventually realize that they need to promote the site. So in the next wave of budgeting, they design some ad banners, buy several thousand impressions on some leading sites, and sit back to wait for the 2% or so of those who see the ads to click on them and come to the site.
And for a great many online ventures, that remains the extent of "Web marketing."
Now that you, dear reader, have read through to this last chapter, we hope you appreciate that online marketing is more complex and rewarding than simply putting up one Web site and a shower of ad banners. As described in Part I, your site needs to be not ordinary, but extraordinary, representing real strategic goals for the company and true benefits for visitors. And as covered in the preceding four chapters of Part II, there are many other, more cost-effective ways to promote a site and develop a loyal audience than online ads, including newsgroups, mailing lists, affinity site links, contests, media relations, and more.
This is not to say that online advertising plays no part, of course. It can be very important for many sites. USWeb's definition of audience development is made up of two key components: the audience creation techniques described in the previous four chapters of Part II, and paid media, also known as advertising. This two-pronged strategy of paid and nonpaid promotion is directly akin to how public relations and advertising are counterparts in most traditional marketing campaigns.
Internet advertising is definitely coming into its own, experiencing a surge in 1997 that even surprised many in the online industry. At a June 1997 meeting of the Internet Advertising Bureau, whose members include many of the biggest sites and online advertisers, a membership poll said that online ad spending for the year would be in the range of $500-700 million, up from around $300 million in 1996. By the end of 1997, however, it appeared more likely that advertisers had spent close to $1 billion. And this in only the third year that any ad dollars had been spent on the Web at all!
However, it is strategically important to understand where the real benefits of online advertising lie and how to use ads in harmony with the free media placement techniques we've been describing.
Most sites use banners to drive inbound traffic. Although there are clearly cases where this makes sense, the kinds of audience creation techniques discussed in previous chapters are often more cost-effective for generating traffic. Furthermore, the strategic importance of banners for brand building is often overlooked by many Web marketers, as discussed later in this chapter.
One reason more sites don't question the cost-effectiveness of using banners to drive traffic is a confusing incongruity between the two key measurements popularly used in Web advertising. For whatever reason, online advertisers have accepted the CPM (cost per thousand) measure as the standard for banner pricing. CPM comes primarily out of the broadcast and print worlds, where branding is the prevailing advertising objective. Yet at the same time these advertisers have deemed clickthroughs to be the most common measure of advertising success, which is a clear extension of direct marketing principles.
Does IBM communicate much more about its brand message in its similar TV commercials? Seems like this animated Web banner ad more or less says it all.
If you're paying for branding but you're measuring for direct results, it's hard to know if you're getting what you're paying for. To the extent that you're interested in paying for branding, it makes sense to quantify the costs in terms of sheer numbers of impressions registered with desired audience segments, as in other media. Measuring the success of branding is complicated, as we'll discuss later in the chapter. Clickthroughs are a poor measure of branding success, and this measurement shouldn't be forced to fit this objective. If driving traffic is indeed the goal, on the other hand, it is useful to translate the CPM rate into a cost-per-click measure to make sure you're getting a fair price.
For example, say an advertiser pays a typical $20 CPM and receives an approximate industry average 2% clickthrough rate (CTR). It makes for easy math: $1 per clickthrough (see "The Math" sidebars later in this chapter). At this fairly typical ad cost, a company would pay $10,000 to generate 10,000 visits to its homepage.
However, $10,000 spent on the type of nonpaid audience creation techniques described earlier in Part II should render a lot more than 10,000 page views. In the Women's Wire case study, you read that Ramona Ambrozic said that such audience creation techniques were 50% more cost-effective than the company's typical banner advertising campaigns. USWeb's promotions for clients such as ichat, Paramount, Macromedia, Jefferson Starship, and others generated tens to hundreds of thousands of page views for substantially less than $1 per click.
Amazon.com's remarkable promotion with John Updike, discussed in Chapter 7, "Era of Innovation: Contests, Sweepstakes, and Other Special Promotions," cost the company a bit over $140,000 in prize money. Yet the site received nearly 400,000 story submissions, to say nothing of the multitudes more who came daily over the six weeks of the contest to follow the interactive story without submitting entries. A fair percentage of those visitors likely bought books while at the site, offsetting marketing costs, to say nothing of the resulting press coverage in almost 300 prestigious outlets worldwide. According to Kay Dangaard, director of media relations, although the site did do a small amount of banner advertising for the promotion, it worked out to pennies per visitor because the real volume of traffic came in response to word of mouth, grassroots, and PR promotions.
One way in which banner ads are clearly more effective than nonpaid audience creation techniques for driving traffic is their ability to scale their results proportionally to a company's spending. Marketers can try to target audience segments in only so many newsgroups without resorting to spamming. Companies such as Microsoft, Intel, and Procter & Gamble are committing multimillion-dollar budgets to online advertising. It would be hard to spend such amounts effectively just on search engine registration, sweepstakes, or even public relations, along with the other nonpaid techniques discussed in earlier chapters. However, successful banner campaigns delivering better-than-average CTRs can easily be expanded to the purchase of more impressions across more sites, and, theoretically, can continue to generate the same proportion of clickthrough traffic indefinitely.
Microsoft and Visa can afford to spend whatever it takes to drive traffic. Presumably their crack marketing teams have taken the time to thoroughly analyze the cost of acquiring new customers and have found banners an effective means.
Web advertising is also an excellent strategic complement to the unpaid promotional strategies of audience creation. Advertising with a publisher opens the door for additional levels of cooperation, including exclusive sponsorships of select content, joint promotions, affinity links with top sites, and so on.
Furthermore, the $1-per-clickthrough cost depicted previously is an oversimplification of the deeply nuanced strategies and rewards available in Web advertising today. That's what we'll cover in the rest of this chapter.
Since 1938, REI has been selling quality outdoor gear and clothing in retail outlets around the country and through its mail-order catalogue. The company ventured into online retail in 1997, and by the end of the year, Web sales were booming.
A few months after launching its site, the company retained USWeb to help build traffic. Through our usual tactics of promoting REI's site and doing online sales promotions in recreational newsgroups, mailing lists, and the like, we were able to effect a significant rise in traffic immediately. REI expanded its promotional budget five months later, and USWeb launched a banner advertising campaign, after which traffic and sales surged dramatically again.
The sporting set has reacted well to REI's online store. After the company opened up an online advertising budget, outdoor- oriented sites were more accommodating than ever to link to REI and recommend the store to readers.
The jump was not entirely due to the success of the banners alone, which experienced average CTRs and ran mostly on medium-sized sites focused on biking, hiking, climbing, camping, environmentalism, and the like. Rather, the paid media gave us new bargaining power with the publishing sites to expand our free media-placement tactics.
Although publishers might be loath to admit it, the policies at many sites are much looser than the traditional church-and-state division between editorial and advertising departments at established print publications. As the medium matures, a little advertising grease can generally loosen editorial favoritism.
"I worked closely with Guy Hill, our media planner, to add value to the REI media plan," said Molly Parsley, the online marketing specialist who led REI's audience creation campaign.
Ms. Parsley wrote emails to the Webmasters of all the sites where Mr. Hill bought advertisements, explaining to them, "Part of my job as Online Marketing Specialist is to get links established for REI, and I would like to know if you could offer any added value to our client by providing a link to REI."
To make the point clear that future advertising may hinge on their cooperation in giving REI some free exposure, she added, "We will be reporting back to REI with a list of links from our current media buy, so please let me know if I can add your site to this list."
The results were powerful. Nearly every site contacted added links to REI, and several added a few lines about REI's services to their member newsletters and other editorial services.
As in other media, there are two principal objectives in advertising online: branding and direct response. Both can be had online in a wide array of flavors and give you much greater control than other media over market segmentation, performance review, speed to execution, and other advantages. Most online marketers can benefit from a combination of the branding and direct-response characteristics of Web advertising, but it pays to appreciate the distinctions between the two.
By and large, marketers have been slow to embrace the brand-building potential of Web ads. It's not hard to see where some of that reluctance comes from. For one thing, page banners, the most common form of online advertising, are small. The most popular size, 468x60 pixels, fills about 10% of a Web browser window on a 14-inch or 15-inch monitor, and proportionally less on a larger monitor. For advertisers used to full-page newspaper ads, 17-foot-tall billboards, and 30-second TV commercials to achieve their branding objectives, a one-by-six-inch rectangle doesn't seem like much to work with.
Two of the most popular banner sizes, standardized by the Internet Advertising Bureau (IAB) and the Coalition for Advertising Supported Information and Entertainment (CASIE), don't take up much real estate on a 14-inch monitor.
Add to this the easy-to-measure clickthrough ability inherent in the ads, and the appeal of a direct-response model has generally overshadowed branding as the key online advertising objective for many sites.
"Measuring the success of Web advertising can be complex, taking into account many different factors," said Guy Hill, media planner at USWeb. "We try to communicate some of that in our reports to our clients, but often the top bosses don't have the time for such subtleties. They want to be able to glance at one number and make a decision about whether a campaign is working or not. So, by the time our contacts present our reports to their bosses, everything gets reduced to a clickthrough rate."
This is regrettable. Not to take anything away from the power of direct response online, but there's also real potential to achieve strong branding on the Web. Without a doubt, the best evidence of this comes from the Internet Advertising Bureau (IAB) and Millward Brown Interactive's 1997 Advertising Effectiveness Study.
Millward Brown, a 20-year-old international advertising research group, meticulously designed a study to demonstrate how giving users a single exposure to a Web banner could have a significant positive impact on their recollection of the ad, affinity with the brand message, and even their intent to purchase the product. After surveying nearly 17,000 Web surfers' reactions to 12 real ad banners on 12 leading Web sites, the report concluded that after only one viewing of a banner ad, Web surfers were on average 5% more aware of the brand and 4% more likely to purchase the product over a competitor's product.
The concept of measuring branding may seem a bit like herding cats--abstract at best. Yet Millward Brown has established itself as an expert of brand measurement, with a third of the Fortune 100 companies on its client list. The methodology of the study is too complex to explain in detail here, but we'll sketch it out briefly. (Copies of the report are available free on Millward Brown's site. See the "Resources" section at the end of the chapter for the URL.)
Collaborating with the IAB's prestigious roster of members, Millward Brown's online division, MBinteractive, was able to experiment with visitors to sites including CNN, CompuServe, ESPN SportsZone, Excite, Geocities, HotWired, and Pathfinder's People magazine, running real advertisements for companies such as Schick, Toyota, Toshiba, Deja News, Kenwood, Delta Business Class, Volvo, and MasterCard.
MBinteractive picked one link at each site, such as the Celebrity News link on People's homepage, and randomly intercepted a certain percentage of folks who clicked it. Instead of taking them to the intended page, the system invited them to partake in a survey. Nearly half agreed to do so. At this stage, the survey just requested basic demographic information "to help the site better understand those who use its service." After submitting the completed survey, the participants were thanked and sent to the page they wanted.
Unbeknownst to the participants, when they submitted the demographic information they were randomly split into two equal groups--equal both in terms of absolute numbers and demographic balance. On the following page, one group was served a test ad and the other was served a control ad, with all other conditions exactly the same.
Then, between a day and a week later, the participants were emailed and invited to come back and answer further questions. This time the survey focused on their recollection of the test banner, its brand message, their opinion of the product, and so on. Again, roughly half participated in the second phase of the survey, a total of 16,758 respondents.
Some of the results were remarkable. Bear in mind that all of the test banners used were in normal circulation throughout the test sites and elsewhere on the Web, so both the test and control participants may have already seen the test banners before or after the tested exposure. With all other things equal between the test and control groups, the survey precisely measured the impact of at least one additional banner viewing.
One showing of this Schick banner ad caused a dramatic brand benefit in the perceptions of those who saw it. (Data courtesy of Millward Brown Interactive and the Internet Advertising Bureau.)
In the case of Schick's Tracer FX razor, members of the test group who were shown the banner ad were 31% more likely to agree that it "meets your need for a men's cartridge razor" compared to the control group. And the test group was 25% more inclined to believe that the product "is an acceptable price" compared to the control group.
Rex Briggs, the MBInteractive analyst who authored the report, had this to say about the responses to those two questions: "These shifts represent increases in perceived relevance and price performance--two critical dimensions for a packaged good."
In the case of the Toshiba Video Board ad banner, 11% of the test group felt the product was "better than other laptop computer accessories," versus 5% of the control group, a 120% increase in positive perception. Of those who saw Volvo's test banner, 11% had "a higher opinion of Volvo than other automobiles," versus 9% of the control group, a 44% increase. Those who viewed the Kenwood ad were 23% more inclined to agree that the brand "appeals to you more than other brands of home stereo systems."
MBinteractive crunches the numbers in myriad ways in the 88-page report, using a variety of complex, proprietary methodologies for quantifying "consumer loyalty," "brand awareness," and other generalized brand concepts. Their methods are carefully explained and based on strong assumptions, though, so the numbers can be taken at face value with relative confidence.
Among the 12 ads, average awareness of the brands increased 5%, from 61% to 64%, after just one additional exposure, the authors concluded. "Consumer loyalty" was up 26% for Volvo, 20% for Delta Business Class Air Travel, and more than 10% each for the Apple QuickTake digital camera, Deja News, and Toshiba.
The news wasn't all positive, though. The Strong Funds investment group lost more than 20% in consumer loyalty after one ad exposure, and two others also slipped slightly. But across all 12 test banners, consumer loyalty rose by an average of 4%.
The report's authors are cognizant of Mark Twain's warning about statistics: "People commonly use statistics like a drunk uses a lamppost: for support rather than for illumination." The IAB and MBinteractive both offer interpretations of the numbers and provide explanations for why the branding impact of a single ad banner appears to be so surprisingly high. (MBinteractive, using Millward Brown's proprietary "FORCE" measure for balancing across different media, goes so far as to argue that Web ad banners are more effective for branding than television commercials.)
The most salient argument in favor of the Web's branding power, which many had made even before the IAB/MBinteractive study, is that Web users are deeply engrossed in the medium and sit only inches from the monitor, unlike TV viewers who lounge halfway across the room and take bathroom breaks and snack runs during most commercials.
In fact, it's little wonder that banners get our attention as much as they do. They're often the only animation on otherwise static pages, and often the first elements to load. While a viewer waits for a page to finish downloading, she has little choice but to watch as the banner fills the top position of the page and runs through its animated message.
RealAudio's ad on this CBS SportsLine page has the user's undivided attention as other page elements gradually load.
For some broad consumer brands, it may make more sense to devote budgetary resources to branding through banners than to try to drive traffic to a large destination Web site at all. In a March 1997 report titled "What Advertising Works," Forrester Research suggested, "Instead of building a big site, a brand like P&G's Tide should be looking for a way to sponsor the online schedules of every Little League in the country." (When we spoke to Tide's Clothesline.com site designers later that summer, they hadn't yet heard Forrester's suggestion, although they admitted it was interesting.)
For destination sites that use banners primarily to acquire visitors, branding still should not be overlooked.
"Even when a site is trying to drive traffic with its banners, there is a branding value there for them," said USWeb's Mr. Hill. "I can see a banner once and not click on it, but it sets the agenda in my mind for me to click on it later. I haven't forgotten it. If I know the brand and occasionally visit the site, the exposure reinforces my behavior. It tells me I'm right. If I haven't been to the site before, at least it puts it in my mind for the next time I have a free moment and am interested in that category of site."
Because the Web is as new as it is, Mr. Hill makes the point that having an online presence at all continues to have its own branding cachet.
"In the offline world, you can choose from among more than a dozen different brands of jeans," he said. "Online, Levi's has got a big presence. I go to their site or see one of their ads and I happen to be wearing Levi's, and it's a strong positive reinforcement. I think, `They're the online jeans brand, and I'm an online kind of a guy. It's hip to wear Levi's.'"
MBinteractive's study supports this point. Sixty-three percent of respondents agreed with the statement, "Brands that advertise on the Web are more forward-thinking than other brands."
Levi's uses Web advertising to reinforce its hip, free-spirited image to the upscale digerati audience.
Another obvious benefit of online branding is the superb demographics of Web users. The latest data from the Web's largest demographic survey, conducted by the Georgia Institute of Technology, indicates that as more of the general population comes online, the average user's income and education levels are coming down while the average age goes up. But it's still the upper end of society leading the way.
In the eighth survey, conducted in October and November of 1997, the mean average household income was $53,000 a year, dropping from $58,000 in the seventh survey six months earlier. That's still far above the U.S. median annual household income, which was $35,500 in 1996 according to the U.S. Census Bureau. Forty-seven percent of respondents to the eighth survey had basic college degrees or higher, which is down from 54% in the seventh survey but still more than double than the 1995 U.S. national average of 23%.
The point as regards branding is clear: Although companies may not yet be able to reach the majority of their consumer audience online, they can reach the potentially most lucrative segment of that audience.
One way to view the online advertising objectives of branding versus direct response is in terms of long-term versus short-term rewards. Although branding may translate into sales eventually, it's generally not an immediate cause and effect. When you want to take advantage of the Web's capacity for instant results, clicks are what count.
Online advertising offers direct marketers substantial advantages over traditional telemarketing and mailing campaigns, chiefly in terms of keen targeting and cost savings.
As we'll discuss in more detail later in the chapter, you can increasingly target banner ads to Web surfers by using the same demographic data used in traditional direct marketing--ZIP code, age, gender, income, and education, as well as shopping history and other known factors. You can further layer other targeting criteria on top of these demographics, such as editorial affiliation, time of day, and immediate user behaviors (for example, intercepting users as they search topical keywords or browse shopping directories).
"The cost of generating a lead on the Internet for us is substantially cheaper than it would be in the offline world," said Ken Neibaur, VP of marketing at the Internet Shopping Network (ISN). Owned by the Home Shopping Network cable TV marketing company, ISN has two online properties--the Computer Super Store, online since 1994 and claiming to be the Web's first commerce site, and First Auction, which began conducting real-time auctions of merchandise in 1997. According to Mr. Neibaur, the two stores combined were doing $2 million a month in sales by the beginning of 1998, with revenues rising by 15-20% a month.
The savings from direct marketing online versus the traditional approach are numerous, he explains, including dramatically lower expenses for acquiring customers, virtually nonexistent costs for repeat email marketing to existing customers, and great savings in campaign testing. Before joining ISN, Mr. Neibaur held management positions at several telemarketing and direct mail ventures, including as a VP at Lintas: Marketing Communications.
"Offline direct-mailing lists cost $60 to $125 per thousand names. The average CPM on banner ads is around $30," he said, noting that a direct-mail envelope is fairly comparable to a banner ad. Most direct-mail recipients throw away the envelopes unopened, and similarly most Web surfers won't click on banners to learn more about their offers.
"But the cost of buying the list is just one aspect of direct marketing costs offline. You've got postage, creative, printing, mailing services, so you're way up there just to acquire your customer base," he said.
On the Web, however, there are virtually no similar production and distribution expenses. Once customers have gone to ISN's online stores the first time and registered as members, ISN adds them to an email marketing list with their permission. A custom software program then automatically generates email messages with new promotions to send to them.
"It costs almost nothing to send to that list twice a week, and we will get anywhere from a 10% to 25% response," Mr. Neibaur continued. "I've never seen an offline direct mail house list get anything above a 10% response, I don't care what it is. And if I were offline and I had to send out direct mail or do telemarketing calls to keep those customers interested, it would cost me a lot of money" in terms of paper, postage, and phone expenses.
Online savings are even greater for direct marketing in terms of testing campaigns, he said.
"In the offline world, you have a tremendous amount of fixed costs going into testing. If you're doing broadcast, you have to produce a commercial and buy a certain amount of media time. For print advertising, you have to buy the creative [that is, ad industry jargon for the art and written copy] and get the insertions. And if you're doing direct mail, it's even worse because you have to print a lot of stuff and mail it. If you buy the wrong list, if your creative doesn't work, if your product is priced wrong, or whatever, you won't recoup that investment. If you spent a dollar on a mail package, which is very common, you're out of luck.
"On the Internet, you can put a test program out for a couple of days or an hour using some models. You can put out five or eight different ad banners in the same media buy and find out that afternoon or the next morning what isn't working and pull it back before you're exposed. On the Web, the up-front cost is basically you, the media buyer, and an art director. That's obviously not the case in the offline world," Mr. Neibaur said.
The Internet Shopping Network's First Auction site, together with its sister site the Computer Super Store, is bringing in $2 million a month and growing fast. ISN's marketers calculate their customer acquisition costs from advertising down to the penny, and they know they're getting a bargain.
More and more sites are giving in to pressure to charge for advertising not only on a CPM basis of impressions shown to visitors but also on a cost-per-click or cost-per-transaction basis. Our favorite jargon for this type of pricing, be it cost-per-click, cost-per-lead, or cost-per-sale, is CPW: cost-per-whatever.
Publishers are frequently reluctant to fix the pricing of a banner to its performance, which is understandable. They argue that many factors that might impact the CTR are out of their control, such as the quality of the banner's design and copy writing or the promotion on offer. It's a bit like a department store paying a newspaper for advertising based on how many sales the ad generates.
The dark secret, however, is that few Web sites are actually selling anywhere near their total inventory of available impressions. House ads, barter, and freebies frequently fill up the majority of ad impressions served at all but the top handful of sites. For this reason, many sites will negotiate CPW ad deals (although most try to keep the availability of such pricing models quiet, as they may be perceived to weaken the site's ability to ask high CPM rates).
2% CTR of 1,000 impressions = 20 visitors.
20 visitors divided by the expense of $20 per thousand ad impressions = $1 per visitor.
5% of those visitors spend $50 a year.
5%, or 1 in 20 = $20 acquisition cost.
Bottom line = $30 net revenue per acquisition.
Ad-supported sites ultimately recognize that getting something for that excess ad inventory (even if at fire-sale prices) is better than nothing. A few networks have recently introduced performance-based pricing over the table, including DoubleClick Direct, Petry, and PointCast Direct.
When you're measuring the success of advertising by a CTR or other CPW criteria, it is critical to do the math on the bottom-line worth of the "whatever" in "cost per whatever" in order to calculate the return on the investment of marketing expenses.
An online commerce site, for example, may be paying the equivalent of $1 per clickthrough, either on a straight CPW basis or as $20 CPM with a 2% CTR. By tracking visitors over time with cookies, the site may calculate that, on average, 5% of those who hit the homepage make a purchase worth $50 within a year. That means it costs the site $20 to acquire a $50 purchase. Considering that the customer may become a regular shopper, and balancing that against other overhead, marketers at such a site may decide that's a profitable acquisition cost.
It may be harder to make the same argument for using banner ads if a site survives off ad revenue. Using the same $1-per-clickthrough ad cost, a site would spend $10,000 to attract 10,000 visitors. If it sold its own banner spaces at the $20-per-thousand rate, it would immediately earn back only $200 in ad sales from those initial clickthroughs. (A site charging $20 CPM effectively earns 2 cents per banner shown.) In order for the site to break even on its advertising expenses, every one of those 10,000 visitors would have to click an additional 49 pages beyond the homepage to earn back the cost of the original ad campaign.
2% CTR of 1,000 impressions = 20 visitors.
20 visitors divided by the expense of $20 per thousand ad impressions = $1 per visitor.
$10,000 spent divided by $1 per visitor = 10,000 visitors.
$20 CPM earned = $0.02 per impression.
10,000 initial homepage impressions x $0.02 per impression = $200 earned.
$10,000 recouped divided by $0.02 per impression = 500,000 impressions must be shown.
500,000 impressions divided by 10,000 visitors = 50 impressions per visitor.
Bottom line = Every last visitor of the 2% who originally click through the ad banner must eventually view another 49 banners at the destination site for site to break even on the ad expense.
For many sites, justifying this math may be a tall order. Put yourself in the Web surfer's place: When you click on a banner, how likely are you to become a loyal user of that site and to visit it again 50 times over the course of several months? Probably not that likely. Most of the time you click a banner, take a quick look at the homepage it leads to, decide you're not interested, and never go back again. Right?
Of course, the preceding equation oversimplifies matters. Although the $20 CPM and 2% CTR may be typical, lots of sites are doing 10% CTR and more, and some are paying $10 CPM or less, at which point the acquisition cost gets much more favorable. However, at $30 CPM and 1.5% CTR, the pendulum swings the other way. The main point is that advertisers who measure success by clickthrough need to do this math.
A picture may be worth a thousand words, but is it worth the price of CPM when generating traffic is the goal? Marketers who are measuring results by clickthroughs had better do the math.
On the other hand, this purely direct-response measure of success also ignores the inherent branding benefit of banners discussed earlier in the chapter. Recall Mr. Hill's argument that although a visitor might not click a banner the first time he sees it, he may recall the URL and visit later. With this in mind, the acquisition costs seem somewhat more reasonable.
But this works only when advertisers anticipate the latent branding benefit and promote the brand.
"It astounds me how many ad banners don't even have a company's name on them," said Kent Valandra, executive VP and director of New Media at Western International Media. An online advertising veteran, Mr. Valandra started working with Prodigy's ad department in the late '80s, before which he had spent 20 years in the print magazine world.
"It's obvious that banners have a branding value," he continued. "A salesman from Time magazine doesn't say to an advertiser, `I can give you 50 million readers, of whom maybe 3 million are going to care about your ad.' If it's worth money for someone to put their name on a basketball court or a rodeo fence or the side of a bus, it's worth putting a name on a Web banner. You never see a billboard for cigarettes at a rodeo without the name of the cigarette."
The IAB/MBinteractive study further makes the point that a CTR is not an effective measure of the branding power of an ad. On average there was a less than 1% difference between people who did not click on an ad but who still recalled its message (43.7%) and those who had seen the ad, including those who had clicked on it, and who recalled its message (44.1%).
In the case of the Schick razor ad, a single exposure of the ad banner boosted by 31% the number of people who agreed that it "meets your needs for a men's cartridge razor." Yet the ad had a lowly 0.5% CTR. Clearly, the CTR alone is a poor measure of that ad's branding success.
The moral of the story is that although a cryptic banner that doesn't reveal the name of the company or product may marginally raise the CTR, the sacrifice of branding is usually a net loss. As Forrester Re-search noted in its report "What Advertising Works," MCI's memorable "Shop Naked!" banner campaign for its online shopping mall generated a lot of clicks, but not many shoppers. The mall is now defunct. (You might wonder what a telecommunications company was doing in the consumer-product retail business in the first place, but that's a different question.)
In cases where clickthroughs are an important and well-considered objective, there are a few rules of thumb for design and ad management, in addition to the most critical factor of targeting the right audience:
Most of these fall in line with the classic direct-marketing mnemonic of AIDA: A promotion should grab attention, rouse interest, stimulate desire, and call to action.
This SmartMoney banner demonstrates the AIDA direct response advertising principle. Its animation grabs one's attention, its changing message rouses interest, its advice for making money stimulates desire, and its offer of a free product to those who "click here" calls to action.
The one online-specific rule of thumb worth elaborating on is the last one: limited frequency of exposures. Contrary to the accepted wisdom in traditional brand advertising, that a prospective customer becomes more likely to buy a product the more he's exposed to the same advertisement (ideally somewhere around 7-10 viewings), the law of diminishing returns seems to apply more quickly on the Web.
In mid-1996, the DoubleClick network conducted an oft-cited study regarding the effects of ad frequency on CTRs. It concluded that after a typical Web surfer had seen the same ad more than three times, the likelihood he would click on it dropped below 1%. DoubleClick dubbed this phenomenon "banner burnout."
Because the Internet is an amalgamation of several media experiences--text, images, sound, video, 3D, chat, and so on--online ads also come in many varieties. Other than scratch-and-sniff magazine ads, advertising on the Web can come close to matching the experience of promotions in every other media, and then some.
Before we get "beyond the banner," an almost meaningless buzz phrase so many online advertisers love to talk about (but seemingly rarely act upon), there is much for you to consider within the realm of those lovable, ubiquitous hyperlinked rectangles.
When HotWired introduced the first ad banner for AT&T in 1994, it was simply a static graphic image fixed at the top of the page. No particular targeting, no rotation, no animation, nada. Pretty crude by today's standards. Now banners do more than act as static, two-dimensional logos fixed on every page. In addition to advanced technical targeting techniques discussed later in the chapter, banners have become dramatically more flashy and functional, thanks to Shockwave, Java, compressed audio and video formats, and other hot technologies.
One drawback to using the "rich-media" formats (that is, anything much more exciting than static graphics) in ad banners is that some users and even more sites don't support them. The news for end users is more encouraging. According to the browser statistics survey maintained by University of Illinois student Ed Kubaitis, discussed in Chapter 3, "Design Optimization," as of December 1997 more than 80% of Internet users surfed on version 3.0 or higher of Netscape and Microsoft's browsers, which support Java by default and accept the more sophisticated multimedia plug-ins, such as Shockwave.
Less encouraging for advertisers were the sites themselves. AdKnowledge, an advertising services company that maintains a database of the 1,000 or so top Internet publishers, reported in August 1997 that only 17% of sites they track support Java in ads, and 14% support Shockwave. Although Java is by now a fairly reliable technology on most browsers, Shockwave animations are limited by whether Web surfers have downloaded and installed the required plug-in.
The most common type of banner animation uses the GIF89 technique. Its popularly stems from its near-universal browser support (including the 2.0 versions of the "big two" browsers), smaller file sizes than other animation types, and the relative simplicity of creating the animations. The name derives from a 1989 adaptation of the graphic interchange format (GIF) that CompuServe developed a few years earlier.
Basically, GIF89s aren't much different than the penny-arcade mutoscopes that so charmed folks 100 years ago: a series of images that flip past each other in quick succession. Most of the banner ads that endlessly cycle through the same four frames of clever text ending in "Click Here!" are GIF89s.
A technology named for the year 1989 is ancient history in Web terms, of course, and innovative companies have since introduced more compelling animation techniques. A company called InterVU, for example, has developed a technology to put video in ad banners for such clients as Budweiser, Warner Bros., and Volvo. InterVU claims that 90% of Web surfers can view video clips with its technology, which first figures out which video format a Web surfer can support, be it MPEG, Vivo, QuickTime, or plain old GIF89, and then serves the appropriate one.
InterVU's video banner technology has brought ads like Budweiser's "Frank and Louie" series to life on the Web.
USWeb worked with Silicon Graphics's virtual reality division, Cosmo, to distribute the first virtual reality modeling language (VRML) ad banner on the Web. During the media hype surrounding NASA's Mars probe in the summer of 1997, Silicon Grahpics developed an ad banner for Pepsi Cola that featured the Sojourner buggy driving around a 3D simulation of the Martian surface and bumping into a giant can of Pepsi. At the time, relatively few Web surfers had a VRML extension installed on their browsers to experience the ad properly, but it made quite a sensation in cutting-edge circles and was a powerful proof of the concept. Silicon Graphics Cosmo Player is now a standard component of the Netscape 4.0 browser, so VRML ads may well become more common in the near future.
It's just not the same in print. The first 3D Web banner, promoting Pepsi during the Mars Sojourner hype, using Silicon Graphics's Cosmo VRML technology.
Realizing that about 98% of the time Web surfers don't want to click ad banners because they're happy with the site they're on, some marketers have begun creating ads in which Web surfers can perform functions within the banner without needing to click through to another destination.
Narrative Enlivens Ad Interactions A company called Narrative Communications makes Enliven, a software product that specializes in interactive banners. The ads are Java files, so they don't require any special plug-ins on standard browsers, yet they deliver smooth animation and sound effects sure to grab a Web surfer's attention. Using streaming technology, the ads download additional data from the server only when the user interacts, creating an experience of seemingly endless new content without excessive delays.
Hewlett-Packard's sensational Pong banner, designed by Red Sky Interactive with Shockwave technology, set new standards for creativity in banner design. The banner let netizens relive the '70s by playing the world's first video game again.
This banner for Diamond Multimedia uses Narrative's Enliven technology to rousing effect. The ad starts with a baseball bat swinging at a ball and making an audible crack, followed by a crowd's cheers. It then offers some true/false questions and a chance to win prizes.
The result is a more powerful ad experience than possible on TV or in print. The user truly interacts with the brand as she clicks on icons, selects choices from pull-down menus, drags and drops items across the screen, and otherwise immerses herself in the ad. Although still a novelty, the ads are fun and can typically absorb users for five minutes or more.
Advertisers such as Citibank, New Balance, and L.L. Bean have used Enliven in ads to demonstrate product features, collect sweepstakes entries, provide useful consumer information, and more, all within standard-size banner ads.
"Our advertisers regularly see anywhere between 16% and 35% of users interacting with the ads," said Jamie Bertasi, director of business development at Narrative. "Our server, meanwhile, collects a tremendous amount of data about all the users' interactions, down to how long they hold their mouse over a particular item, whether more surfers click on the left or the right side, and so on. It gives the advertisers a tremendous amount of information to work with to redesign their ads more effectively, change the promotion, whatever."
A competitor, Thinking Media, makes a similar product with an advantage to publishers in that it doesn't require them to purchase a special server for the streaming effect, as Narrative's Enliven does. Which company, if either, will become the standard for interactive ads is anyone's guess at this point.
Another company, First Virtual, which specializes in secure transaction technology, has developed a type of ad banner in which Web surfers can actually make secure purchases within the banner (in addition to a newer, unrelated line of software for email commerce). Customers of the banner technology, called VirtualTAG, include PC Zone, Casio, and the U.S. Post Office, who have used it to sell computers, watches, and stamps.
Ms. Bertasi says that in 1998 Narrative plans to enhance Enliven so it can also transact secure commerce within banners.
First Virtual used its VirtualTAG electronic- commerce banner technology to create this ad where a user can donate money to the United Cerebral Palsy charity in an animated, interactive game-like environment.
One downside to interactive banners is that many users have become wary of clicking on them due to the trend of faux interactive ads. These deceptive ads invite the user to click on a search field or scrollbar within the banner, only to find that it's a static graphic that automatically transfers the user to a new page.
Porno sites, always at the vanguard of new techniques to attract users, favor this ruse. Try searching on the word "butt" on Yahoo!, Excite, or another search engine. The resulting banners for hardcore porno sites frequently indicate that they're in the process of automatically downloading a video clip to your desktop. They prominently feature a Cancel button, but when you press it, it merely takes you to the site you're trying to avoid. The search engines are unapologetic about displaying these all-round unsavory ads.
What appears to be the first search box on this AltaVista page is actually a fake interactive ad banner that loads DoubleClick's site when clicked. A few experiences like that and Web surfers may be conditioned against trying real interactive ads.
Searching Yahoo! for "butt" brings up this banner. Although it's really just a GIF89 animation, users may believe they're downloading a porno video. Hitting the Cancel button calls up a smut site.
If ever a piece of computer jargon deserved to be hung up by its thumbs, it has to be the abominable interstitials. These are the full-screen ads that occasionally pop up between pages on cutting-edge sites. Unfortunately, interstitial is indeed a genuine, if somewhat obscure dictionary word meaning "situated in the space that intervenes between things."
Interstitials are what TV advertisers have been waiting for on the Web. They get in your face, take up the full browser screen, and don't go away until you click on them or they're done with their animation. The term has been applied to splash screen ads, which are simply static ad pages that the user has to click to go to the next page, but we prefer to think of true interstitials as animated ones lasting 5-15 seconds.
"Until now, online advertising has lacked the kind of impact possible with TV commercials, where a well-made AT&T ad can make viewers weepy," said Evan Neufeld, an analyst with Jupiter Communications. "You just can't get that from a banner. Interstitials, however, promise to bring that power to the Web."
In a June 1997 report titled "Banners and Beyond," Jupiter analysts predicted that by the year 2001, a quarter of all online ad spending would go to interstitials. We certainly hope that doesn't mean a quarter of all ads will be interstitials, or anywhere close. The same report notes that although Net users haven't objected to the still-uncommon phenomenon of interstitials, if they became so widespread that Web-surfing meant constant 10-second interruptions by animated commercials, users would almost certainly revolt.
In addition to users' annoyance at having their machines hijacked by intrusive ads they can't avoid--flying in the face of the innate freedom the Web offers users--there's also the matter of delays while the interstitials download.
The "interstitial" aspect of the ads is that they appear between Web pages. Theoretically, while the user views page A, the interstitial file should download in the background, ready to run instantly when the user clicks to go to page B. But anyone who has waited for a 15-second Shockwave animation to download can guess that the experience might not be so seamless.
Berkeley Systems' frenetic online trivia game, You Don't Know Jack, makes great use of interstitial ads. Three times during play, the game pauses for a pair of full-screen animated ads. Here, James Bond walks through his classic logo to promote a marathon 007 week on TBS.
A pioneer of interstitial-style animated ads is the push-technology leader PointCast. Using Macromedia Director (the same application used to design Shockwave animations), ad agencies have created hundreds of animated PointCast ads for companies including Toyota, Birds Eye, and MCI. Because it's a push application (see Chapter 7 for more discussion on push), PointCast gets around the delay of downloading large animation files. Its users are generally connected to the Web through corporate T1 lines, so the server can feed the files to the desktop without waiting for users to "pull" the data off a Web site, as browsers must do. As a result, the animations are preloaded for instant play while the user is on the phone, writing memos, or otherwise occupied.
Pulling off this smooth effect on a Web site is more challenging. One software company with interstitial technology, Adletts, has attempted to work around the problem by keeping its Java animations files very small, around 15KB. Instead of taking over the entire browser screen, the Adletts files open a small window where the quick animations run in the foreground and cannot be canceled until they're finished.
Because interstitials are still in the early stages, it remains to be seen whether they'll have the impact some analysts predict without being condemned by users as an annoyance, scaring sites away from using them.
In addition to Jupiter's prediction that interstitials will make up a quarter of all ad spending by 2001, the analyst group thinks that half of all ad spending will remain with banners, particularly interactive ones. The remaining quarter, they believe, will be applied to sponsorships.
We suspect that sponsorships will end up with a larger piece of the pie than interstitials.
Already, sponsorships are popular across the Web in a variety of forms. With the boundaries between online editorial and marketing still maturing, it's frequently difficult to tell where the content ends and the advertising begins. Regardless, a lot of marketers are feeling confident in exploring sponsorships.
"There is definitely a potential danger there for both sites and advertisers that users may feel cheated, that what they thought was editorial content they later realize is paid advertising," said Christopher Theodoros, VP and director of creative sponsorship strategies at DoubleClick. "When approached carefully, however, sponsorships can be an extremely powerful form of advertising online. It's a question of preserving the credibility of the content while enhancing it with the brand association."
Although in most forms of sponsorship Web surfers are encouraged to click through to the sponsor's site, the goal of this model is principally brand-building. Advertisers and publishers are still playing with the definitions of sponsorship on the Web, and so far the results have taken many different forms. In our view, based on current trends, there are five basic types of sponsorships:
Branded content most closely follows the "brought to you by" model of early television and today's public television. Think of yesteryear's "Mutual of Omaha's Wild Kingdom."
In the branded content model, the advertiser has no hand in creating or shaping the content. Editorial control is left entirely to the publisher, while the brand benefits from the association with the quality content. These kinds of deals are frequently negotiated as long-term contracts, normally in an exclusive arrangement so that no other advertiser, particularly a competitor, is associated with the content.
In such cases an advertiser may sponsor a regular section of a site, or the entire site in some cases. Oldsmobile's sponsorship of CBS SportsLine's Auto Racing section is an example of a site section sponsorship. The relationship has lasted more than a year, precluding any other auto brand from being associated with this editorial positioning. Toyota, meanwhile, has gone a step further and underwritten the entire ad budget of Time Warner's Cars and Culture site.
CBS SportsLine's Auto Racing section is "presented by Aurora from Oldsmobile," a branded content sponsorship that has lasted more than a year.
Event promotions involve a closer integration of publisher and advertiser content. These promotions last only a short period, normally from one to several weeks, and highlight a special offer, contest, or event. They're frequently given great prominence by the publisher, such as at the top of the homepage.
Yahoo! seems to run this type of contest sponsorship perpetually, with the promotion frequently integrated into the Yahoo! logo. The Dilbert contest discussed in Chapter 7 is an example of this type of sponsorship, in which both United Media and Nynex's Big Yellow business directory were promoted. Later Yahoo! hosted a virtual road rally contest sponsored by Volvo, with a grand prize of a trip to Scandinavia.
Yahoo! regularly hosts event promotions, such as this contest sponsored by Volvo.
Movies frequently engage in this type of sponsorship. For the U.S. launch of the film Mr. Bean, Rowan Atkinson's goofy character was spotted loping incongruously across HotWired's pages. Disney, meanwhile, commemorated the opening of 101 Dalmatians by adding black spots to the backgrounds of several prominent sites.
Further blurring the line between ad and editorial content online are advertorials. They're a familiar tradition in print journalism, where publishers normally make it abundantly clear that they're paid advertisements. Typically they do this by setting the type in a distinct font, surrounding the item in a border, and prominently labeling it "Paid Advertisement." Web publishers aren't always so fastidious.
Many smaller publishers struggling to stay afloat are willing to cut corners on such formalities and will display material in whatever way makes advertisers happy. Surprisingly, even established media companies have shown a willingness to blur this line in their online ventures.
For example, one wonders whether Forbes is the "capitalist tool" or merely the tool of capitalists. The magazine's Web site was quick to embrace the advertorial model, sharing columns evenly in its various content sections, with staff articles on the left and advertiser text on the right. In the first year or so, the site did label this sponsorship "advertisement" in bold type at the top. More recently, however, that label has disappeared, replaced by a small disclaimer at the bottom of the column.
The whole right-hand column of this page is an advertorial, although Forbes doesn't make this clear until the bottom of the column.
Microsites expand the idea of the advertorial to multiple ad/content pages, along the lines of special ad supplements to magazines and newspapers. Rather than drawing the visitor away from the editorial site to the advertiser's corporate site, these microsites usually maintain a thematic congruity with the publisher's content while allowing the advertiser to immerse the visitor in the brand. For users, microsites can be an attractive way to explore a brand without a complete shift of context and the annoyance of waiting for a whole new set of server connections to load a new site in the browser.
A good example of a microsite is Levi's "They Go On" section within NBC.com. The ad feature was highlighted prominently on NBC's homepage alongside microsites for the Homicide series, Jay Leno's Tonight Show, and other network programming, as if "They Go On" were one of NBC's shows. However, the microsite turns out to be an enigmatic, artsy account of a young couple's cross-country road trip in the impressionistic style of Levi's striking "They Go On" television commercials.
It's hard to tell whether this microsite for Levi's on NBC.com is advertising or content. On dozens of stylish pages, it recounts the travels of a young couple around the country and includes discussion forums for visitors. At heart, however, it's an ad in the hip, youthful cinematic style of Levi's "They Go On" TV commercials.
What we refer to as portals are the murkiest category of sponsorship. Here, the distinctions between advertising and syndication and revenue-sharing and other online promotional and money-making strategies begin to meld completely.
A portal sponsorship, by our definition, is where one site agrees to integrate the content of another site as a service to Web surfers and a branding value to the content provider. Examples include a news site incorporating the functions of a leading search engine, or a content aggregator like AOL linking to only one shopping site in a particular category, or a business magazine offering up-to-the-minute stock quotes from a specialized stock service.
Amazon.com and Barnes and Noble have been in pitched battle to secure exclusive linking arrangements with leading sites as the preferred online bookstore. Amazon.com Associates Program encourages smaller sites to post an Amazon.com logo with recommendations of specific books, rewarding the sites with up to a 15% revenue share on any subsequent sales via the link. Is that an ad or a sales commission? The distinction probably doesn't matter to the sites getting paid to host the link. Amazon.com wins either way, getting free branding even if the link generates no sales.
In many portal arrangements, no money ever changes hands. ABCNews.com, for example, links to Mr. Showbiz for its entertainment news, and likewise Mr. Showbiz links to ABCNews.com on its home-page. Because both sites are properties of ABC, it's a pure cross- promotion deal.
USA Today's site features several portals to content from other publishers, such as investment advice from the Motley Fool. One suspects that USA Today gets the content for free from Motley Fool in exchange for the great brand exposure. When asked, executives from both USA Today and Motley Fool declined to explain the terms of the arrangement, citing a confidentiality agreement. However, USA Today marketing director Allegra Young acknowledged that the site does indeed get content for free from some partners in exchange for the publicity it offers them.
DLJdirect pays the New York Times for the publicity it receives on NYTimes.com and provides the news giant with a stock quote service for free. A classic portal sponsorship.
Although the Web generally gets star billing among Internet media, there are of course other platforms, such as email, push technology, and chat, each with significant numbers of users and advertising possibilities.
One advantage these services have over Web advertising is that they provide strong demographic targeting opportunities. Chat, push, and email subscription services generally break through the Web's barrier of anonymity by offering users personalized services that require them to define their age, gender, geographic location, and other demographic data, along with hobbies and other special interests. Whereas many Web sites encounter reluctance from users to fill out registration forms, these alternative platforms can often entice valuable demographics from users in exchange for value-added services.
Electronic mail or email is used by significantly more people worldwide than the Web. And although unsolicited commercial emails, called spam, will likely remain a brand-damaging anathema to most Internet users indefinitely, there are several viable channels available for email advertising.
Services such as Juno, Hotmail, and RocketMail have signed up millions who have agreed to receive ads in exchange for free email service. Juno supplies users with its own software that dials directly to a local access number for the email account, thereby eliminating the need for users to have their own Internet access.
Hotmail and RocketMail, on the other hand, offer email via the Web, raising the seeming contradiction that users must first have access to the Internet and the Web before they can get free email. Yet these services have found a sizable, if diverse, base of users, including students and others with limited institutional Web access, library surfers, business travelers who may find Web access on the road, users seeking secondary private email accounts to avoid snooping bosses or conduct secret romances, and employees in risky start-up companies who want a consistent online address in case their corporate handle vanishes in bankruptcy.
Juno lets advertisers send direct email promotions to users, in addition to serving banner ads within its proprietary email software. Hotmail and RocketMail serve only traditional Web banner ads when users log on to those Web sites to check their email.
Hotmail gives users email for free on the Web in exchange for banner ads, such as this one for Visa.
BonusMail presents another email advertising model. Users agree to receive direct email ads in exchange for points they accumulate, redeemable for free gifts at a variety of participating sponsors. To collect points, users must demonstrate that they have read the ads by replying to the messages with code words embedded in the ads. When they act on special promotional offers in the ads, they receive additional credits. BonusMail has targeted users of HTML-enhanced email software packages for its service, so nearly 70% of subscribers get graphics, hyperlinks, and specially formatted text in the direct email ad messages.
BonusMail's 27-year-old president and CEO, Steven Markowitz, described the opportunity he envisioned in founding the company: "When I came up with the idea, I saw it as a way to solve a problem direct marketers faced online. On the one hand, spam has a terrible reputation and is still damaging for the image of advertisers who engage in it. On the other hand, there is a tremendous appeal in correlating the simplicity of email with computer-managed consumer demographics. The answer, as I saw it, was to give users an incentive to agree to receive ad messages, while making sure those messages were closely targeted to their interests. After all, the difference in perception between what is junk mail and what is valuable information depends entirely on how interesting the offer is to the recipient."
Real-time chat is one of the fastest-growing applications in online culture. Since the early popularization of cyberspace, netizens on commercial services like CompuServe and AOL, as well as those on the geekier Internet Relay Chat (IRC) protocol, have enjoyed the magical connectedness of typing text into their computer screens and seeing instant responses from others who are hundreds or thousands of miles away. More recently, businesses have begun to embrace chat for use in customer service, virtual business meetings, and other applications. The driving force behind the medium, however, remains flirtation and other social interaction.
In early 1997, AOL led the charge in chat advertising, enabling marketers to display ads to the millions of users of its more than 15,000 chat rooms. In the same year Web-based chat software hit its stride, and now many sites offer chat communities as exclusive site services or as enhancements to other content.
WebChat Broadcasting, for example, one of the largest Internet chat communities, has integrated graphics into its chat software, allowing members to append pictures of themselves to their posts. The same technology allows ads to float amidst the scrolling chat text, in addition to ads fixed elsewhere in the page frames.
In WebChat Broadcast's forum on pro football, Atlanta Internet Bank's banner gets some great exposure amidst sports fans' chat about getting drunk.
A hopeful software firm, Blaxxun (formerly known as Black Sun), has created a product called Community Server that allows advertisers to create robot advertisements that draw users into private chats about products. The chatting ads can be sent into action when users type key phrases during public chat sessions.
The company went so far as to develop a 3D version of the software for the small-but-growing world of VRML chat rooms. As a proof of concept, it developed a character named Dusty, a chatting vacuum cleaner, to promote Black & Decker's Dustbuster product. When visitors wandered into a 3D world on Blaxxun's site and happened to use a word such as "dirt" or "housework" during a chat (or even if they just hung around in the virtual plaza long enough), they were approached by the 3D likeness of a handheld vacuum cleaner with Groucho Marx glasses, a mustache, and bushy eyebrows. This avatar (VRML-speak for a 3D virtual character), Dusty, politely invited them into a preprogrammed conversation about the virtues of the Dustbuster. As far as we know, this 3D ad robot technology never found paying customers, and Blaxxun, a Munich-based company, has humbly closed its U.S. offices and kept a low profile about the product since its launch.
Chat presents both benefits and liabilities for advertisers. On the upside, the audience is generally very loyal, checking back with the same communities frequently and staying logged in to the services longer than visitors to typical Web sites, which presents opportunities for strong brand associations. Also, most chat sites collect basic demographic information about users that can be used in ad targeting.
On the downside, chat doesn't generally drive much traffic because users come for the conversation, not for ideas on where to surf next. Also, although specific chat rooms may have narrow themes, such as professional hockey, mountain climbing, or Jimi Hendrix, there's no telling what path the actual conversations will take. Brands may end up being positioned next to off-color or otherwise inappropriate discussions, possibly even bashing of the advertiser in question.
As we discussed in Chapter 7 push technology has yet to come close to the Internet-rocking impact promised in the hype of early 1997. As of this writing, it's too early to tell whether the release of Netscape's Netcaster and Microsoft's Active Desktop push platforms, bundled with the 4.0 versions of their browsers, will revitalize this trend.
Jupiter, in its June 1997 report "Beyond the Browser," certainly believed it will, predicting that "push technology solutions will become ubiquitous among leading Web sites, and that a significant portion of publishers' revenue will be derived from these services." We remain to be convinced.
Meanwhile, PointCast, the granddaddy of push applications, does give some indication of the powerful potential of push advertising. Since its launch in early 1996, PointCast claims to have attracted more than 2 million loyal users. The service began targeting mainly office workers in the high-tech industry who have fast, constant connections to the Internet. Using proprietary end-user software, the service kicks in as a screen saver when users' computers are idle, scrolling news stories and animated ads.
Building on its immediate success with users, PointCast has signed syndication arrangements with more than 1,000 content providers. It has also introduced targeted news services including a Canadian edition, a college network, and specialized vertical professional news networks serving such niche industries as banking, health care, law, real estate, and state and federal government agencies.
PointCast's Business Network channel, supported by advertisers like MCI, reaches the valuable demographic of corporate workers with fast Internet connections.
Anna Zornosa, senior vice president of advertising and affiliate development at PointCast, explained that in late 1995, selling advertisers on the idea of TV-like animated ads on the Internet wasn't a piece of cake.
"Everyone's initial reaction was, `Oh, come on, how are you going to make an ad impression with a screen saver?' None of the clients or ad agencies had ever made a 30-second animated computer commercial before.
"In the magazine industry, a typical ad sales call lasts about 40 minutes. Our early sales calls lasted 3-4 hours. An hour into the meeting, we would mention you can click on the ad. They were amazed. Until then, they hadn't realized these were interactive ads we were talking about."
Since those early days, the network has proved its worth to many of the dozens of advertisers who've taken advantage of it, including Polaroid, Oldsmobile, and Procter & Gamble.
The battle cry of Internet marketing has long been "one-to-one marketing." Although the phrase has fast become a cyber-cliche, the message behind it is real. In the world of online advertising, the operative word is targeting.
Imagine if television sets were two-way communication devices that could stare back at their viewers and relay information about them to the broadcast station. For example, one TV set could see that its viewer was a woman in her mid-40s. In fact, it knew her name was Mary Wilson. She was fairly affluent, judging by the furnishings in her living room, yet she was single. According to her viewing habits, the TV surmised that she was interested in sports, cooking, and celebrity entertainment news. It knew that Mary lived on the lower-east side of Manhattan. And it also had a sense of self, aware that it was a six-year-old Sony and that soon she would be in the market for a replacement.
The TV dutifully sent all this information back to the broadcaster, who accordingly targeted her with ads suited just for her: running shoes, New York singles bars, cooking classes, and new TVs. Her neighbor in the apartment next door watched the same shows at the same time, but saw a completely different set of ads.
It sounds like an Orwellian nightmare. But this scenario isn't far off from what's taking place on the Web today, whether the majority of netizens realize it or not.
Rotating advertising is no mean feat for any site, and targeting those ads to individual users based on their unique characteristics, all within a fraction of a second, seems impossibly complicated. Yet more than a dozen specialized companies provide software packages and services to do exactly that, including Accipiter, AdKnowledge, AdSmart, Aptex, ATG, Autonomy, BellCore, DoubleClick, FlyCast, Imgis, Intelligent Interactions, MatchLogic, Narrowline, Nestor, NetGravity, Real Media, and others.
Although these companies take care of the under-the-hood technical aspects of ad targeting, it behooves you to understand the general strategies for audience segmentation online. These include content, technographic, demographic, geographic, and psychographic targeting.
The most basic form of online ad targeting, and also one of the most effective, is content affiliation. Just as a running shoe manufacturer would advertise in Sports Illustrated magazine and an investment service would advertise in Money magazine, advertisers online generally seek their target audiences according to the editorial themes of specialized sites. In fact, the vast majority of ad targeting on the Web today is done this way.
Content targeting generally doesn't necessarily exploit much technical sophistication. Rather, it's done through familiar, manual means: Media buyers analyze which sites are suitable venues for their marketing objectives, and the ads are then placed in general rotation across the site or within particular site sections.
One of the advantages (and difficulties) of targeting by content is the incredible breadth and diversity of Web sites out there. Most of the more than 1 million sites on the Web are run by hobbyists or institutions that don't run any ads. Most of them would probably be more than willing to do so if the opportunity presented itself, though. For advertisers in search of niche markets, no other medium comes close to the Web's minute categorization.
The challenge for the marketer is in qualifying what's available. The Web doesn't yet have the kinds of established directories that ad agencies use to reference media properties in TV, radio, newspapers, and magazines. The few services that come close, notably AdKnowledge's MarketMatch database, include only a few hundred sites among the thousands of potential ad venues.
Even the more professional sites have wildly divergent rate cards and reporting standards, making it sometimes maddening for advertisers to be clear about what they're paying for.
LinkExchange: King of Niche and Reach
For advertisers who want to consolidate large ad buys in a single package, there are several advertising networks that represent dozens or even hundreds of large and medium sites, including DoubleClick, Zulu Media, 24/7 Media, Real Media, AdSmart, and FlyCast.
One network stands out as very different, with the unique value propositions of incredible reach (that is, a large distribution to many Web surfers) and extraordinary niche targeting: the LinkExchange network.
The brainchild of a couple of 23-year-old Harvard computer graduates, Sanjay Madan and Tony Hsieh, LinkExchange brilliantly embodies the original "information wants to be free" spirit of the Internet community. Because of this, many marketers neglected to take the network seriously, assuming the founders were just starry-eyed kids who'd get real jobs eventually. The investment specialists at Sequoia Capital looked a bit deeper and saw the network's financial promise, however, banking the company for $3 million in mid-1997. At that point, many analysts and advertisers finally started to catch on to the distinctive power of this network.
In a nutshell, LinkExchange offers small sites free advertising on a 2:1 basis--for every two ads a member displays on its own site, it gets one ad to promote itself on other member sites. Because each member site displays twice as many ads as it's given to promote itself, the LinkExchange company ends up with a huge available inventory of unused ad credits.
That's where the money comes in: LinkExchange can sell all those excess ads to non-members--that is, paying advertisers.
The idea has spawned many imitators, but none comes remotely close to LinkExchange's reach. The network has caught on like wildfire, counting more than 200,000 members by the end of 1997, less than two years after it launched. To put it in perspective, more than 10% of all sites on the Web belong to the LinkExchange network. The network shows more than five million ads a day, and, according to the Media Metrix research group, its combined sites are seen by more Web surfers than any other Web services except Yahoo! and America Online.
The appeal to advertisers is twofold. First, no other service rivals LinkExchange's impact in making a massive impression across the whole Web.
Consider the case of Universal Studios' promotion of its celebrity chat for The Lost World: Jurassic Park. Like many film studios, Universal regularly schedules chats to coincide with its movie releases. According to LinkExchange's executives, Universal came to them in a panic two days before the premiere of the film. In all the work the moviemakers did to get ready for the blockbuster's debut, they'd overlooked their usual means of promoting the celebrity chat, and now there was no time left to pursue those avenues. Not expecting much, Universal asked LinkExchange to do what it could to get the word out.
Universal's celebrity chats typically had been attracting around 2,000-3,000 participants. After LinkExchange blanketed more than 100,000 sites (the size of the network at the time) for a day, a record 12,000 chatters showed up to type along with stars Jeff Goldblum and Julianne Moore. Universal was impressed enough about setting an Internet chat record to put out a press release.
Second, LinkExchange can deliver incredible content targeting by every imaginable niche topic. It has categorized its member sites into 1,600 different topic affiliations.
Surf Point, LinkExchange's member directory, lists 433 sites in the real estate category alone.
Imagine an advertiser who's interested in targeting football players and fans with a football trivia game. He may go predictably to ESPN SportsZone and CBS SportsLine, as well as the NFL site and perhaps a handful of other top sports sites. LinkExchange, meanwhile, has nearly 10,000 sports sites, including 250 sites devoted to professional football, 100 on college football, 75 on fantasy football, 20 on football trading cards, 10 on high school football, and another 100 or so in six other football categories.
In addition to other major sports categories, LinkExchange's section "Other Ball Sports" includes further subcategories for sites dedicated to billiards and snooker, bowling, cricket, Australian football, handball, lacrosse, racquetball, squash, rugby, table tennis, badminton, volleyball, and more.
"Some advertisers think they should stick to sites with known brands," said Ali Partovi, VP of business development for LinkExchange. "But think of advertising on radio. No advertiser tells their ad agency that they want their ads to run during a certain DJ's show on a particular station in Seattle, or immediately after a specific song on a station in Dallas. They say, `Get me broad reach during drive-time on top 40 stations,' and it's the media buyer's job at the ad agency to know which stations fit the bill. LinkExchange provides that same alternative on the Web."
Also called environmental targeting, aiming ads to users according to the technographic information volunteered by every browser is the most basic form of dynamic ad targeting. In the back-and-forth chatting of the HTTP Web protocol, the browser gives the server various information about the Web surfer's computer and network, such as what kind of browser and operating system he uses, his ISP, what URL he linked to your site from, what time of day he visited, and other data.
With the appropriate server software, sites can target ads to individual users based on all these criteria. Many sites that brag about being able to target ads to users "dynamically" or in "real-time" are targeting according to such technographic variables, mainly because it's the easiest data about users to collect.
Unfortunately, technographic targeting is of limited benefit to most non-technology advertisers. If you're selling beer, mutual funds, or cars, it probably doesn't make much difference whether Web surfers use Netscape or Microsoft browsers on Macs or PCs.
Was it an amazing coincidence that I'm one of the precious few remaining Macintosh users, and I saw this same ad for the Mac version of Internet Explorer all across the Web one week?? No. The sites sniffed out my OS before serving the ad.
Perhaps a product particularly geared towards students could target according to .edu domains. Or you could target Web surfers with AOL accounts on the broad assumption that they represent typical Middle American consumers (and presuming there's some advantage to targeting them on a particular Web site instead of directly within AOL's network).
Time of day is another data point that sites extol as a target for advertisers. They cite examples such as a TV program that's promoted an hour before showtime or a snack maker that runs banners around lunch time. The problem with targeting by time, however, is time zones. Although it may be TV primetime according to the site's California-based servers, it would already be pushing midnight for a New York Web surfer. Judging the users' geographical location by technographic information is a much trickier proposition, as we'll discuss shortly.
On the balance, although technographics can be very important for software vendors and other technological companies, non-tech advertisers need to scrutinize the purported benefits of this kind of ad targeting.
Demographic targeting is "the beef" of the "one-to-one marketing" hype. Demographics weed out the pimple-faced teenage burger-flippers surfing in mom's basement from the males aged 35-45 in the Bay Area who earn over $100,000 a year and are interested in fast cars.
Thankfully, from a privacy standpoint, it's still rather difficult to distinguish the two groups from their browsers alone. Most demographic information valuable to marketers must be gleaned from Web surfers directly. This generally means requesting visitors to fill out site registration data.
Hotmail's free email service relies on users to fill out a brief demographic survey, allowing the site to target ads more effectively to them.
The problem for advertisers is that few sites truly implement demographic targeting effectively. There are two main reasons for this. First, many users won't fill out such forms, either due to privacy concerns or because it's a pain in the posterior to keep entering the same data about themselves at site after site. Second, even when sites do manage to collect lots of user information, managing that data is no mean feat for a site with thousands or millions of visitors.
Nonetheless, there are sites out there that manage to do so, offering advertisers powerfully advanced targeting abilities. The New York Times is one of the few sites that has succeeded in demanding that all users register to gain access. Due to its superb content, this is a tradeoff that more than two million registered users have been willing to make. As a result, the Times can target ads to the demographic "golden triangle" of ZAG (ZIP code, age, and gender), plus income brackets.
Bruce Judson, executive VP and chief marketing officer at CellularVision, a New York-based high-speed wireless ISP, praised the New York Times' demographic targeting as the most effective advertising vehicle in any media that CellularVision has used to promote its services. Mr. Judson, who is the former general manager of Time Inc. New Media and cofounder of Time's Pathfinder site, explained that initially CellularVision wanted to broadly target New York state residents due to geographic licensing restrictions. However, the company soon found that it got significantly higher direct sales results the more it customized the ads to highly local regions.
"After some initial tests with the Times, we started to customize within geographic areas, which was made possible by their unique ZIP code capabilities," he said. "We literally said, `Hey Brooklyn!' in an ad, or `Hey Chelsea!' People coming to the site really knew, `Oh, this is for me! I live in Brooklyn,' or `I live in Chelsea.' We found that this kind of personalization led to a dramatic lift in responses. More importantly, we also found that the kind of audience that we brought in converted very well to customers."
A company called Imgis, which outsources the technical infrastructure involved in serving ads for sites and networks, has announced its intention to take demographic ad targeting to the next level by matching Web surfers against their offline shopping histories. The company has arranged to use demographic information from MetroMail, a 50-year-old giant in the direct marketing business with extensive consumer information on millions of Americans, and match that against the customer lists from ISPs and other sources. This way Imgis could let advertisers target netizens who bought luxury cars five years ago, for example, or those with appetites for expensive electronic gadgets.
CellularVision used the New York Times' ZIP code demographic to target ads to users in particular New York neighborhoods, such as the fashionable downtown Chelsea district. Acquisition rates outperformed the company's ads in all other media on a cost basis.
Not surprisingly, privacy advocates have jumped all over Imgis for its plan. Company officials insist that privacy will be respected to a very high degree. No names, phone numbers, or even email addresses will be available to advertisers or even Imgis employees, due to a method of cross-referencing anonymous serial numbers through a trusted third party.
Nonetheless, even the fact that direct marketers in the offline world regularly use these same sorts of consumer histories for target marketing doesn't ease the anxieties of many Internet users. Netizens hold the new online media to a higher standard of privacy protection, as was discussed in Chapter 2, "Web Value Propositions."
Cookies and OPS
Probably the most misunderstood, maligned, and punned-upon Internet technology is the so-called cookie file.
Originally dubbed magic cookie by its inventors at Netscape, a cookie is nothing more than a text file that resides on the user's hard drive, into which sites can record a short string of data about the user for future reference. Without cookies, the HTTP protocol of the Web doesn't provide any easy way to recognize the same user from one click to the next.
Typically, a site will use a cookie to assign a serial number to an individual user, against which it can cross-reference data stored on the site's own database regarding the user's previous interactions with the site.
Common uses for cookies include the following: enabling a user of a commerce site to carry a virtual shopping basket from page to page, so the site can keep track of what items the user selected for purchase on earlier pages; recognizing a registered member at the homepage so he doesn't have to always enter his password and ID; tracking how many times a Web surfer has seen a particular ad banner to avoid showing him the same one too many times; and matching a user to his demographic profile.
Despite hysterical media reports, cookies cannot "learn" anything about Web surfers. They're simply repositories for data, not agents that sniff anything out. They most certainly can't determine a visitor's email address, bank account number, or any other personal information, unless he has already volunteered that information to the site in the first place. In that case, the cookie simply allows the site to cross-reference the ID of the browser against the user's personal information in the site's database.
Because sites don't actually store much more than a serial number on the user's hard drive, an unrelated site can't use the cookie to gain access to any personal information that the user may have divulged. If sites belong to the same network, however, the network can cross-reference the share data about individual users from one site in the network to the next through cookies.
The idea of third-party cookies bothered some privacy advocates, who in mid-1997 succeeded in lobbying a leading Internet standards body, the Internet Engineering Task Force (IETF), to propose a new set of standards that would do away with third-party cookies.
The online advertising community, which hadn't paid attention to the drafting process of the new standards, woke up in a panic and lobbied Microsoft and Netscape to ignore the proposed standards. Because the IETF only recommends standards without any power to enforce them, the Big Two browser makers sided with the advertising community. They ignored the IETF's recommendations in the release of their 4.0 browsers, leaving cookie management pretty much the same as it had been. Eliminating third-party cookies would have badly shaken the ability of many advertisers to target effectively.
As a compromise, Netscape, Microsoft, and a slew of other companies endorsed a new proposed technology known as the Open Profiling Standard (OPS). Details of the initiative have been scant since its announcement in the summer of 1997, but committees are at work drafting a form of it likely to be adopted with the 5.0 versions of the main browsers.
Basically, OPS would allow users to configure their browsers at setup with a wide range of personal demographic data. This would include most of the information requested on registration forms, such as name, email address, gender, and so on, and even such sensitive information as credit card numbers.
The information would be stored securely within the browser so that snooping colleagues or other users couldn't access it without a password. Then, when the Web surfer encountered a site that asked for personal data, he could release only the amount of data he wanted on a site-by-site basis, and without having to retype it over and over.
Presuming it does come to pass, OPS sounds like it would solve several sticky problems of personal data management, ad targeting, and privacy on the Web.
Many marketers are realizing that the World Wide Web can be used effectively for very local marketing. Local Web services, from city entertainment listings such as Microsoft's Sidewalk to small- and medium-sized businesses who want to market regionally, are a fast-growing Web trend. And as CellularVision found on the New York Times site, targeting ads to users by neighborhood can be highly effective.
So how can you target geographically? The most reliable way is, again, demographic data supplied by the users. As we've seen, however, relatively few sites are prepared to deliver real demographics. Fortunately, there are several alternative ways you can target by geography.
One method that's not always reliable is known as reverse IP look-up. Sites and networks attempt to recognize visitors by their assigned IP address--a unique string of numbers assigned to every host computer connected to the Internet--and determine where that computer resides physically. This works with many corporate Internet users, whose desktop computers are constantly online and are assigned fixed IP addresses that never change.
The drawback is that the majority of Internet users are connecting from home through ISPs that assign different IP addresses to users each time they connect. The problem is particularly acute with AOL's 10 million-plus users, who are all routed through a handful of external gateways to the Internet. So according to their IP addresses, they almost all appear to be residents of Vienna, Virginia. The same may be true of employees of large national corporations, all of whose IP addresses may be routed through one gateway.
At most, sites may be able to geographically locate 10-20% of their traffic within a general region by using the reverse IP look-up method. The technique certainly cannot promise the kind of neighborhood pinpointing that ZIP code-based targeting provides.
A more obvious alternative is to advertise on locally oriented sites, such as a regional newspaper's online service.
Furthermore, many sites can deduce a Web surfer's geographical location through her use of specialized services such as weather or mapping. Why else would someone check the weather in San Francisco unless she lived there or had some intimate interest in the region?
The Weather Channel's site shows a copy of a BayInsider ad to surfers who search on San Francisco weather conditions.
Many sites provide localized service to Web surfers by asking for their ZIP codes and no other demographic data. TV Guide's site, for example, provides local TV listings based on users' ZIP codes. MovieFinder.com does the same for local cinema showtimes.
Advertisers who are determined to target ads geographically need only hunt around a bit to discover many sites that can provide this valuable data point.
Arguably the most promising means of ad targeting is based on reading people's behavior. Psychologists and market researchers agree that in order to reliably predict what people will do in the future, you should pay less attention to what they say they'll do than to what they're already doing.
Already among the most valuable kinds of ad targeting is catching Web surfers as they're searching for somewhere to go. The most obvious form of this is through keywords on search engines. The search engines sell banner ads that come up on the results pages in response to surfers searching on popular keywords. Most prime words associated with big-ticket items, such as "travel," "cars," and "computers," are sold out months or years in advance on the leading search engines.
"We see 25% clickthrough rates on our ads tied to the search word `airline,'" said Anouk Snyder, advertising and promotion manager at Preview Travel. "We buy on Infoseek, Yahoo!, AltaVista, Hotbot, and Excite. There's nothing so remarkable about the creative design of our ads. It's the keywords that do it. It's hard to buy good keywords."
Keywords fetch an even higher premium on "yellow pages" directories, where it's taken for granted that Web surfers are in shopping mode as they search for product categories.
The recent emergence of neural networking technology on the Web is likely to have an important impact on ad targeting in the near future. Aptex, Autonomy, and Nestor are three leading vendors of neural networking software, originally designed for military use in pattern recognition, such as the modus operandi of terrorists. The same software is already providing remarkable results on the Web, recognizing patterns in how users interact with Web pages and serving ads accordingly. In short order, the software can get the gist of what kind of content a Web surfer is browsing and anticipate which ad would best catch his attention.
The Infoseek search engine has gotten rave reviews from advertisers for its Ultramatch service, which is based on Aptex's SelectCast software.
"We've seen clickthrough rates double for clients that have advertised on sites using SelectCast," said Brent Hall, senior media buyer at Anderson Lembke, an ad agency that's active in Web advertising.
Mike Lynch, CEO of Autonomy, describes how analyzing a user's behavior and forming a psychographic profile can be a more accurate predictor of behavior than mere demographic measures.
"Say you knew that I was 40-year-old male, living in Silicon Valley, married, and earning $150,000 a year," he said. "You'd think the ads best suited for me would be for a red sports car or a cell phone or a top-of-the-line hi-fi system. What you don't know, however, is that I hate my job, my marriage is falling apart, and I'm dreaming of back-packing in Nepal. If you could read the pages I'm looking at online, however, like our software can, you'd have a better idea and could target me with ads for graduate schools, travel packages, marriage counselors, or dating services."
Miles Walsh, CEO of the FlyCast ad network, is also convinced that neural networks have a big role in the future of online advertising. He plans to buy one of the neural networking products for his ad network in the coming months. "The combination of editable cookies, along the lines of the proposed Open Profile Standard, and neural networks are going to be huge," he said. "They promise the best of both worlds. Editable cookies will let the user give you the demographic information he wants you to know about him, and the neural networks will let you observe what he's doing and make strong predictions about what he's really looking for. Using the two together is going to take direct marketing online to the next level."
Stay tuned. It promises to be a wild ride.
Also see the Appendix, "Internet Resources," for more online resources related to Web advertising.
The IAB is the leading association for companies concerned with Web advertising, whose members include leading online publishers, advertisers, software makers, ad agencies, and others. The site of a wealth of free resources and even more to paying members.
Best known by its acronym, CASIE represents mostly ad agencies. The site offers some useful resources.
The Direct Marketing Association doesn't have a particularly strong Web site yet, but it's a leading association for direct marketing in the real world, so maybe it will become a better resource with time.
"The Four As," as it's known, is a leading association in the traditional advertising world. Drill down a bit, and the site offers many useful links and services.
The ABC, the reigning body for monitoring publications' circulation figures in the world of print, has branched into Web measurements as well. The site features some useful services.
Ad Age, the bible of the real world ad trade, was quick to get hip to the Net, both with its own site and a dedicated section covering Internet advertising (to which your author is a regular contributor). The Interaction section of the site is a bookmark for most who follow online advertising.
Adweek, Ad Age's historic rival, also has a site featuring online ad news.
This regularly updated site features descriptions of many of the players in the online advertising space and other resources.
The site reviews online ad campaigns and offers other sevices.
One of the many properties of Mecklermedia's Internet.com site, the Internet Advertising Report also covers the online industry closely.
This email newsletter is a valuable insider report of behind-the-scenes deals in the online ad space. It is written with wit by Michael Tchong, founder of MacWEEK magazine and former executive at the Chiat/Day ad agency. The Web site provides subscription information and archives.
Purchased by ClickZ in late 1997, Microscope reviews ad campaigns and invites media buyers to write about their experiences.
The Online Ads email list is generally thought of as the best forum for the wide range of companies trying to make a buck or promote a product with Web advertising. Newbies and old hands are in equal measure. Sign-up is easy at the site. The free list also features the weekly AdBytes digest of many of the week's most important stories about online advertising, with links to complete coverage by other media.
This is a near facsimile of the Online Ads list. We prefer Online Ads, though, when we have time to follow only one.
You can find another fair-quality ad list discussion here.
This list forum is specifically for members of the LinkExchange ad network, but is open to anyone.
This site is a collection of lots of good online ad-related links and content.
Here you can find a strong set of links, including current news titles from other sites, and other online ad resources.
This URL is a direct link to the free, excellent study described in detail earlier in the chapter.
This service is a staple in traditional advertising for providing data on ad outlets ranging from consumer magazines and newspaper to direct marketing lists, radio, and TV. Like everyone else, it's moving online. Although its best services are available only to subscribers, the site does have an excellent list of other online advertising and marketing resources.
This page gives thousands of email lists that accept paying advertising, letting advertisers leverage email without spamming.
The leading ad network offers lots of advice on this page, including its tips for effective banners and the "Banner Burnout" study mentioned earlier in the chapter.
An oddly branded site itself, seemingly the side project of an online real estate service (White Palm), this page has extensive listings to advice, original and collected, about effective banner campaigns.
This page links to lots more sites covering the topic of online advertising.
© Copyright, Macmillan Computer Publishing. All rights reserved.