It’s crazy how things can go up and down very fast sometimes. My last blog entry was about how Automagram started to work great and generate revenues; not even two month later, here I am writing about its shutdown. I’ll try to explain in details how I got Automagram high, how it all stopped, how I managed to controll the issues related to it and what I’ll remember from this experience.
Few days after my last blog entry, I did a talk at Microsoft Research Asia for Barcamp Beijing about Automagram. The talk title was “How I got a startup by accident” – a little catchy as I never really considered Automagram as a start-up, but more as a passive income. I mainly talked about how I came up with the idea and how things went crazy as soon as I setup the premium membership page. At this time, the statistics were going up, but slowly, same for the income generated by the premium memberships.
After the talk, seeing the interest in the product growing, I decided to take it to another level by working on Automagram SEO. I started to buy a lot of social marketing gigs on Fiverr and other places, in order to get a lot of social signals to www.automagram.com. I also took care to redirect the non-www traffic to the www and optimized the homepage title (from “Automagram • Grow your influence on Instagram” to “Automagram • Free bot for Instagram automation”). I also tweaked the description a little (from “A free service to get targeted followers on Instagram based on your topics and interests” to “A free service to get targeted followers on Instagram. Automate your like, follow and comment interactions to start getting quality followers”). Small but effective changes, the application of what I’ve learned with Mon-Instit, the first project I ever built and now very solid in the SERP. A week later, Automagram was on the first page for many of my main keywords, including “instagram bot”, “instagram auto like” and their variations. That’s when everything started to go out of control.
People like numbers, so here you go: right before it stopped, Automagram was receiving 3.000+ visits a day and had 6.000+ users; the premium memberships were generating between 100$ and 300$ a day; the app was used to send more than 1.000.000 likes to Instagram each day.
I think it’s that last point that made the system collapse.
I reached the inbox zero few days before everything started to go wrong with Automagram, so yeah, I definitly remember that email. I thought it would be yet-another-support email with an easy question or something, but it wasn’t. A user reported an issue that never happened before, so I postponed and decided to take a look at it after my office hours. Through the day, I received bunch of other emails reporting the same issue, making me realize that something serious was going on.
My office hours done, I decided to investigate the issue right away and quickly faced this error message from Instagram API while debugging: “We’ve taken extra measures to protect the Instagram API from abnormal activity. Since you have recently created your API application, please contact firstname.lastname@example.org to receive whitelisting for a higher rate limit.”. At some point, Instagram finally detected the abnormal activity going through Automagram client. Coincidence or not, this happened just few days after Instagram released major security updates targeting the black hat scene and system abuses – followers delivery was sometime leading to the “follow” button getting frozen and unusable for any new user wanting to follow the customer who paid for the delivery. Of course, I was aware that Automagram was anything but respecting Instagram TOS and that it could happen – I assumed all the people that were using Automagram were aware as well, but that’s another thing.
Anyway, being aware of this, I already thought about many scenarios where Automagram would get ban and came up with a workaround quickly: I ordered a new VPS at DigitalOcean to get a new IP, created a new Instagram account to hold the new Instagram API client Automagram will used and backup then restored everything on a fresh new config, with a fresh new client, IP and all. It did the trick. So I decided to redirect all the traffic from www.automagram.com to my new IP – not a transparent redirection, a 301 via htaccess to get rid of the automagram.com domain, just in case Instagram also banned the domain as it was used as a callback in the old client. It was working when I was testing, so I thought that was the final step for a back to normal. I then went to smoke my cigarette, after rushing all the thing described (in a nutshell) above, thinking I handled the situation. But when I came back, I realized the issue occurred again, on the fresh version of Automagram. That’s the moment I started to loose hope.
I saw I started to play a hide’n’seek game with Instagram that I had no time for. I would probably have find a solution and/or an alternative to go through their security measures again (especially by digging the signed_body parameter they’re using in-app to make their API calls) but that would not have been worth it for me in term of personal investment and timing.
So I sat and took some time to think about the situation. I came up with the controversial decision to just shutdown Automagram.
The Bad Buzz
From the moment I decided to shutdown Automagram, I knew a lot of angry users/customers would complain. I’m not a community manager, but that’s the deal when you’re doing things alone: you have to do everything on your own.
Hopefully, I received some support messages from comprehensive users wishing me best of luck and saying thanks for the help provided by the service, but mainly, messages I received was full of anger and almost all were asking for a refund as I didn’t respected the “lifetime” membership they signed up for. Unfortunately for myself, I didn’t took the time to write any TOS for Automagram as I thought everybody understood the nature of the (black hat) tool and that it could be stopped by Instagram anytime. So obviously, “lifetime” meant “lifetime” of the service but,yeah, looks like it was not that obvious for everybody. Of course these kind of messages can easily affect you as a human being, but in my case I just focused and thought I have to move on, keep on working and find solutions instead. To be able to refund as many angry customers as possible – on a first claim, first served basis for customers who really didn’t have time to enjoy their premium membership for long – I decided to sell the domain to a Chinese competitor that was interested in it and who contacted me through Skype. That gave me some more immediate funds that I was able to use to start sending out refunds.
I quickly realized I did a lot of things wrong while trying to manage the bad buzz (redirecting directly the domain to the Chinese competitor without any further more explanations for instance) but, yeah, that’s just one of the many things I learned the hard way with Automagram.
What I’ve Learned
When I first started Automagram, I never thought I’ll learn so much through it.
First, I discovered a whole new aspect of social media marketing I wasn’t aware of. I discovered an active scene of people hacking on about all the existing social medias. It was interesting to exchange with them about their journey and see how some made a living out of this kind of activities. I also learned how to make a service up from the ground, how to monetize it, how to assure customer support, how to grow and convert an user base etc. I don’t say that was a complete success, but at least I tried and had a first glimpse of what entrepreneurship could be.
On the other hand, I realized I did a lot of things wrong among which: trying to build a product or a service based on a third party – especially if you’re not respecting that third party terms; trying to offer many features and not automating as much thing as possible right from the beginning; trying to sell lifetime subscriptions instead of monthly, yearly or so; trying to do everything alone; and many other things I took note of.
Try. Fail. Learn. Repeat.
Last but not least, I would like to present my sincere apologies to all the users that supported the project and who are now disappointed. My thanks go to all the early-users who followed me in the project and who never stopped sending me good feedback or support messages, that’s greatly appreciated, thank you.
I know this entry and apologies won’t be enough for some, but life goes on. I learned a lot with Automagram and I won’t stop working on new projects, but I don’t think that will be related to the blackhat scene anymore. So consider this as a goodbye.
Thanks for everything.