Z-Dashboard

Basic Functions:

• Click-through Rate (CTR)
• Pageviews
• Content Popularity

Advanced Functions:

• User Segmentation
• Latent Factors

When you visualize user segmentation, each point represents one user, and each color represents similar interests that those users share. Examples of segmentation analysis can be performed using:

• similar content segmentation: "product X is related to product Y”
• user segmentation: “people in your group bought product Z”

Users are then automatically segmented in profiles using latent factors, such as age, clicked content or any other relevant information you have available.

From the moment you know the user profile it is possible to offer personalized items, evaluate content performance and many other actions (see more details in Personalize).

While content segmentation is reasonably simple, profile segmentation is complex and offers better results, being one of the strong features of zunnit products. In profile segmentation each user is unique in his set of group preferences.

Segmentation information is available not only in the Z-Dashboard, but as an API with detailed documentation as well. With this API you can integrate user segmentation in your business model automatically.

Entity Recognition API

The Entity Recognition API selects and categorizes the relevant words that exist in your content, such as persons, organizations and places. These categories can be customized to your needs.

This API can be used to automatically tag your content and create word clouds.

Topic Page Creation API

The Topic Page Creation API selects content based in specific topics prioritizing relevance automatically.

The relevance parameters are customized for you, with options such as photo gallery, timeline, similar news and relevant products.

Recommendation Case: E-commerce Demo Store

In order to showcase Zunnit recommender systems, an e-commerce demo store is setup at http://magento.playground.zunnit.com.

Beneath each product, four recommended products are shown using algorithms that take into account:

• Item popularity


• User behavior, such as "Users also liked" or "Users also bought"


• User segmentation in profiles, such as "young woman", "business man", "mother", etc.

How does personalized recommendation work?

The goal of Zunnit recommender systems is to offer the users of your site a personalized experience, improving conversion rates and engagement.

In order to achieve this objective it is vital to know your users, predict their behavior and offer highly attractive options.

Zunnit Systems do all that in real time. When a user clicks in a product or news page, this information is sent to Zunnit Systems. This user is classified in a profile based on his behavior and of similar users. Then, the most relevant product or news is selected and offered to this user automatically in milliseconds. In the user's point of view, the recommended item was there all the time.

Each user has a relevant and unique experience.

The next step: Personalized Actions

Products and news recommendations are just the first step.

Zunnit Systems allow you to integrate personalization and user profiling in your decision-making processes.

Use cases include:

• Personalized ads, campaigns or a complete personalization of the site experience.
• Predictive analysis based on profiles combined with action suggestions. For example, analysis of which user profiles are more likely to upgrade subscription plans, or which are more likely to cancel subscription.
• Individual predictive risk analysis to prevent undesirable behavior. For example, anticipate if a user might file complaints with regulatory organs and suggest preventive actions. Another case is risk analysis of hospital patients with suggestion of actions to health professionals.

To personalize your e-commerce or receive more information, contact us.

To personalize your blog or news site, start now by clicking the button.

---

- Big Data Capture: The first step is to collect the available data that may be pageviews and clicks on a webpage, BI internal data, CRM data or customized databases. These data is then structured and stored.

- Deep Learning Engines: The next step applies Deep Learning Engines to find patterns and clusters. Clustering is an automatic classification based on latent factors subjacent to the data and Zunnit provides natural clusters (unsupervised learning) or clusters to achieve a specific objective (supervised learning).

- Inference & Analysis Engine: Inference and Analysis tools are then applied to the discovered clusters and decision support tools are used to achieve specific goals such as product and actions recommendation (Recommendation System), suggestion of sales leads with high acceptability probabilities (Sales Leads), detection and prevention of risky behaviour (Decision Support) and others.

- Data Visualization: The detected clusters and the performance of the decision support system are visualized in a Dashboard developed for each client that shows all the available metrics and their evolution in time.

This platform is being successfully used in several clients to evaluate and understand user profiles generating sales leads, identifying fraudulent behaviour, discovering latent factors, and even automatic grouping and management of multimedia assets.