Documentation / Product / Features / Audiences

Audience Discovery

Audience Discovery provides a comparison between two audiences and helps you discover the attributes that influence your customers to move between them. Drive your conversions using these insights to better target or suppress the audience for your campaigns. Audience Discovery reports are updated in real-time enabling you to take action for existing and new marketing campaigns.

An attribute can be thought of as any data associated with the user profile. Affinity for a certain topic, impressions on a particular campaign, or membership in any given audience are all attributes.

You will need the Discovery Insights role to see the Discovery page in your Audience tab. To generate Audience Discovery reports for your account, contact your Lytics Account Manager.

audience discovery search

Navigate to your Audience Discovery reports via Audience > Discovery. You can search for reports by name or description.

Understanding Audience Discovery

Audience Discovery reports use data science to compare two audiences — a source and a target. Our algorithms first observe the attributes of all the people in each audience, and based on those attributes they predict which users are most likely to move between audiences. For example, if most of the people in the target audience have Attribute X, anyone in the source who has Attribute X is considered more likely to enter the target audience. The degree to which they are more likely to enter the target audience is represented by lift.

The Audiences Discovery report surfaces several Audience Insights, each of which shows how likely certain audience members are to convert. For example, you might see that "users with a high affinity for NFL are 4% more likely to convert" or "users in Newsletter Referrer audience are 8% less likely to convert".

Audience Key Performance Indicators

audience discovery kpi

Along the top of your report, you will see a set of metrics. These metrics are helpful for at-a-glance understanding of your audience sizes. The table below defines each metric and how often these numbers are updated in the UI.

MetricDefinitionUpdated
Source sizeThe count of users who have yet to complete a conversion event or behavior (e.g. Unknown users).Every 2 hours
Target sizeThe count of users who have completed a conversion event or behavior (e.g. Known users).Every 2 hours
Likely to convertThe count of users in the source audience that are deemed likely to convert to the target audience.Every 2 hours

Insight Explorer

insight-explorer-3

Use this chart to browse all attributes that make users more or less likely to enter your target audience. You can switch between the positive and negative lift buttons in the top right for Likely and Unlikely. Filter the attributes with the All attributes dropdown.

Each attribute is represented by a circle. The size of a circle reflects how many users have an attribute, and the opacity indicates how much lift is provided. So, in the example above, "Stock" is the content affinity with the most users (therefore the biggest bubble), while "Data-as-content" provides the highest percentage of lift, which you can see in the table view in the section below. Click an attribute to view additional details including how many users have that attribute and how much lift is provided.

Insight Table View

At the bottom of the Insight Explorer, select Show details for a table view of the audience attributes. This shows the Type, Lift percentage, and the count of Users likely to convert (or "unlikely" if you've selected that button above). You can filter the table by any of these to get a better sense of what attributes have the most influence on getting your users to convert from the source audience to the target audience.

insights-table-expanded-2

When thinking about lift, it is helpful to remember that for every attribute that provides positive lift, the absence of that attribute provides negative lift. For example, if users who have Attribute X are 10% more likely to convert, the opposite is also true; users without Attribute X are 10% less likely to convert.

MetricDefinitionUpdated
More likely to convertThe count of users who have an attribute indicating they are more likely to enter the target audience.Every 2 hours
Less likely to convertThe count of users who have an attribute indicating they are less likely to enter the target audience.Every 2 hours

Audience Insights

Screen Shot 2019-02-01 at 4.08.45 PM

Audience Insights surface the attributes that Lytics recommends taking action on. To create an audience based on these attributes, click Target N Users to jump into the audience builder. You can use one of these audiences to improve the targeting on a new or existing experience.

Note: Some insights will suggest an attribute that provides negative lift — that is, users with a particular attribute may be less likely to convert. Audiences created with an attribute that provides negative lift will exclude users with that attribute by default.

Donut chart

Screen Shot 2018-11-28 at 1.19.27 PM

The donut chart is a great way to quickly understand how the sizes of each audience in the report relate to each other. Areas in pink represent the source audience, areas in green represent the target. Darker areas are the portions of that audience likely to move from the source to the target or from the target to the source.

Audience sizes over time

audience discovery time chart

The time series chart tracks the sizes of four groups of users, which are:

  1. People in the source audience (light pink).
  2. People in the source audience who are likely to enter the target audience (dark pink).
  3. People in the target audience (light green).
  4. People in the target audience who are likely to enter the source audience (dark green).

    Note: Even though this may not be possible (for example, a known user cannot regress to unknown), these users can be considered "at risk" or otherwise low value.

As you take action based on the data in this report, you can use this chart to track how your audience sizes are changing in response to your marketing.

MetricDefinitionUpdated
Likely to convert over timeThe count of users in the source audience that are deemed more likely to convert to the target audience over time.Every 2 hours
Likely to regress over timeThe count of users in the target audience that are deemed likely to revert to the source audience over time.Every 2 hours