Content Interest criterion

Content Interest is a segmentation criterion that enables defining a segment. 

To learn how to create a segment, please refer to this article.

Using the Content Interest segment criterion in the targeting of your campaign enables you to restrict its visibility to visitors based on their interest in certain products or services featured on your website. The more pages your users visit, the more likely they will be recognized as appetent to a type of product or service.

AB Tasty uses NLP (Natural Language Processing) and keyword matching to capture visitor browsing interests based on visited pages and many other metrics. The service is based on page titles and recognized keywords within this element. It’s important to verify that your page titles are well configured and represent your product categories and pages before relying entirely on this criterion. 

☀️ Good to know

Content Interest is compatible only with languages having spaces between words, thus Content Interest will not work with Japanese, Mandarin, Cantonese , Arabic, etc. 

👻 How is the “interest” calculated? 

The interest is calculated from a page title keyword’s number of occurrences we find during a visitor navigation. Every day, we calculate how many keyword occurrences we can find for each visitor specific navigation. From that, we determine interest by taking for each keyword the top 20 percentile. This is fine tuned for each of our customers with an AI service to offer you the best fit! 

Once we have calculated the minimum number of views that are required to be “interested”, we calculate the share of visitors that reached this minimum threshold to be considered interested in a specific keyword. 

Let’s take an example. Imagine the word “pantalon”. We will count the number of views (in the page title) for this word:

visitor_1: 0

visitor_2: 1

visitor_3: 1

visitor_4: 1

visitor_5: 2

visitor_6: 2

visitor_7: 2

visitor_8: 3

visitor_9: 3

visitor_10: 4

Statistically, the top 20 percentile in that example is 3. We will now consider that a visitor needs to see the word 3 times in his navigation to be considered interested. Finally, we will be able to calculate the audience by taking all the visitors that have reached this number of views. 

🥇 Which keywords are selected?

We train algorithms and use Natural Language Processing (NLP) to detect the most valuable keywords within the page titles. Our service detects in priority what we call “items” (products, brands, services, etc.). It does after that a second treatment to detect the “attributes” (colors, adjectives, etc.). We developed this service to remove keywords that do not match a specific interest to your business. 

⚙️ How does it work?

Within the tag, we created an object called “ABTastyData”, that contains information about your visitors journey. Inside this object, we store visited pages data in a location called “VisitedPages”. This is performed every time a user loads a new page. The following information are stored:

• Current page URL
• Current page title
• Current session number (from the “ns” variable of the “ABTasty” cookie)
• Current timestamp

For the Content Interest calculation, our data services are using this “page title” information stored in our global object. As mentioned above, AB Tasty then calculates a number of data associated with each keyword to offer you this service. It can also be enriched with the Product Hits integration.

🤖 Configuration

To use this criterion inside the targeting, simply drag and drop the “Content Interest” label inside the middle column of the segment builder. Once it is done, apply the following steps to start discovering which kind of interests (called “keywords”) we suggest you to use in your experimentation strategy:

• Select a period from the first drop-down list: 
• Over the past week (selected by default) or 
• Over the past two weeks. 
• Click “See results” to unveil the keyword results we have aggregated on your website
• A table is then displayed with every result AB Tasty suggests you to create new audiences on
• Each keyword can be associated with 4 main KPIs:
• Traffic share 
• Number of views 
• Transaction rate of the audience interested by each keywords (that reflect a specific interest on your website)
• Number of transactions of this audience
• Select one or multiple keywords by clicking on the radio button on the left of each keyword 

To be able to display metrics associated with the content interest, you need to integrate what we call the Product Hits. It’s a specific integration that you can proceed with, that helps us receive from you a precise number of product views, and associate those views with transactions on the same product references (through the matching of product name or SKU, with data from our transaction tag collection). These data will help you decide which audience to target according to your business objectives. 

If a keyword does not interest you for any experimentation campaign. 

☀️ Good to know 

When you select multiple keywords inside the same segment to create an audience gathering multiple types of interested visitors, data displayed within the selection zone still correspond to one specific keyword. Indeed, the metrics do not add to one another, because some visitors could belong to multiple groups of interest at the same time. 

Let’s review a basic example: 

A keywords X has a traffic share of 5%

A keyword Y has a traffic share of 3%

→ The audience created with the keywords X + Y will not necessarily represent a traffic share of 8%

We advise you to be careful with the interpretation of those data 😇

💡 Use case

You’ll find below some examples of configuration to get some inspiration. As it focuses mainly on categories and product related keywords, you can imagine all kinds of product recommendations, discounts, and displays according to the keywords you choose in your targeting.