Topics & Affinities


Topics are the atomic unit of Affinity Engine. A Topic is simply an attribute associated with the documents your users are interacting with. Documents by default are the URLs of your site, but could also be customized to include products from a product catalog. For each document, there will be a set of Topics that represent the content found in the document.

By default, every materialized profile will contain a set of URLs visited. Given the link between URLs visited and topics for URLs, Lytics can algorithmically calculate which topics a user has shown interest in.

This is the core of Lytics content recommendations as well as building audiences based on content affinity.


Since topic affinities are a product of user behavior, conclusions can only be drawn about topic affinities relative to other topics and other users. For these reasons, attempting to calculate topic aversion would be an unsafe extrapolation.

To view your Topics, navigate to the Interest Engine page, and click on the Default tile.


An Affinity groups one or more Topics into a targetable unit. In most cases, Topics are too fine-grained for marketing use cases. More commonly, you'll want to create an Affinity consisting of multiple Topics that share something in common.

For example, let's say you are a car dealership and you want to target users who are interested in luxury cars. You could create a Luxury Car Affinity that contains brands like BMW, Audi, Mercedes, etc. This Affinity would allow you to create an audience of users who have an affinity for any luxury car brand, and export that audience to various ad networks or other marketing tools.


Data-Driven Marketing

Affinity Engine allows your data team to import activity data so that Lytics can fully leverage it, while allowing your marketing team to organize the information in a way that enables their marketing campaigns. By grouping Topics into Affinities, your team can leverage the Lytics Affinity Engine to enable all sorts of data-driven marketing use cases.

Leveraging Affinities

The purpose of the Affinity Engine is to enrich a user's profile based on their behavior. However, the utility of this data is limited if we don't take action based on it. The three main ways that Affinity Engine information can be used in Lytics are through segmentation, recommendation, and Lookalike Models.


Lytics surfaces Affinity Engine scores on a user's profile. You can then use those scores inside of our audience builder to segment a user based on their Affinities. This opens the door to all sorts of interesting use cases. Perhaps, you are a shoe company and launching a new type of boots. You can now target an audience of users who are interested in boots without wasting ad spend on users who show very little interest in boots.

By creating an audience of users with an interest in boots, you can also see how your user's Affinities shift over time. Perhaps Affinities for boots peak during the winter. Or does boot interest peak during the fall as people explore their options before the coldest months hit. Affinities can become a useful trend tracking tool as Affinities shift among your users.

Recommendation and Personalization

Lytics' Recommendation system combines AI and marketing strategy to allow marketers to easily suggest content and products that are relevant to their users. Setting up Recommendations can create more personalized experiences with users, which leads to increased user engagement, and more time on site.

Lytics makes it easy to get up and running with Content Recommendations. Upon configuration, our Recommendation system:

  • Leverages over 500 behavioral signals to provide robust and relevant recommendations.

  • Is completely autonomous. Our models get retrained and optimized weekly so that your recommendations stay fresh.

  • Works well out-of-the-box for new users; little or no data is required to provide users with recommended content.

  • Can handle large loads and is able to rapidly scale to meet your needs.

Lookalike Models

Affinity Engine scores can be very useful for Lookalike Models. Affinity Engine scores serve as a standardized way of scoring users against each other. The data model consistency that Affinity Engine creates across users is extremely helpful when building machine learning models. Due to this consistency, we frequently find affinity scores showing up as important data points in Lookalike Models. If your team plans on executing use cases with predictive audiences, we highly recommend that you leverage Affinities.

Affinity Builder

Affinities allow you to group related Topics into a targetable unit, making it easier to reach your customers based on their interest groups. This document will walk you through how to use the Affinity Builder step-by-step. Once an Affinity is created, you can use it to build Audiences and content collections.

Navigate to Content > Affinities. From the list view, click Create New.

  1. Add a Name for your Affinity. This will be displayed in the Affinity tab of the audience builder.
  2. Add a Description for more context. The name and description fields are required.
  3. Assign Topics to your Affinity. There are a few ways to explore and add Topics.
    • Use the filter to quickly find Common Topics. Lytics surfaces the Topics that are most prevalent in your taxonomy, based on the number of documents with that Topic. The default limit for Common Topics is 500.
    • Click on a row to see related Topics.
    • Select the checkbox next to a Topic to add it to your Affinity.
    • Use the Search to find any Topic in your account.

The search option gives you access to any Topic in your taxonomy, even if it's not included in your list of Common Topics. This means you don't need to allow or block certain Topics to make the ones you care about available in the UI.

As you select Topics, they will be added to the Assigned Topics list on the right. In this example, we created an "Advertising Affinity" that included Topics such as "Google Ads" and "targeted advertising."

There are two ways you can remove selected Topics from your Affinity.

  • Hover over the Topic in the Assigned Topics list and click Remove Topic.
  • Unselect the checkbox next to the Topic within the table view.

Once an Affinity has been created, you can add or remove Topics at a later time, and edit the name or description.

Affinity Summary

Each Affinity has a summary page showing how many users are interested in this set of Topics and how much of your brand's content or product inventory is represented. You will be taken to the summary page after creating an Affinity, and when you click on any row in the Affinity list.

At the top of the Affinity Summary page, you'll see the following information:

  • Owner: Lytics user who created the Affinity.
  • Created: Date the Affinity was originally created.
  • Updated: Date the Affinity was most recently edited.


After you create an Affinity, it may take up to a few days for this chart to populate as Lytics is re-evaluating your users' interest in the set of Topics. To the left of the chart, there are a few key metrics:

  • Total Documents that contain one or more of your selected Topics for this Affinity.
  • Users with Affinity that have shown interest in one or more Topics in this Affinity.

On the chart, the X axis represents affinity scores on a scale of 0-100, with 0 being "no interest" and 100 being "highest interest." The Y axis represents the number of users. You can click on individual bars to see how many users have a specific affinity level.

This information can help your marketing teams decide which users to target based on their interest in particular Topics. For example, you can focus on users with a medium-high affinity for the Topics of a particular ad campaign, as those users are more likely to find your messaging relevant and convert.


This section outlines how your Affinity is configured.

  • Topics: The set of related Topics selected for this Affinity.
  • Configuration ID: The method of calculation for the Affinity. default will be listed unless you have a custom setup configured by the Lytics Data Science Team.
  • Slug: Automatically created based on the name of your Affinity. Lowercase letters are used with underscores between words.

Sample Documents

At the bottom of the summary page, you will see a paginated list of sample documents for this Affinity. Your content must be classified before it will appear in this list.

Affinities List

Affinities are designed to be curated groups of your brand's most important Topics. Lytics recommends you start by creating Affinities that are central to your marketing efforts and add more as needed. Since Affinities are groups of Topics, you should expect to have far fewer Affinities than Topics.

In the Lytics app, navigate to Content > Affinities to see the list of your created Affinities.

At the top of the Affinities List page, you'll see the following information:

  • Inventory Table: The table containing the item inventory
  • Inventory ID: The unique identifier for the document or item.
  • User Inventory ID: The inventory of documents or items that a user has interacted with.
  • Inventory Features: The features or Topics associated with your items

You can search by name, and organize the list based on the name or date columns.