Audience Builder is a GUI that lets you define the full audience for a segment.

You can use Audience Builder to define the full audience for a segment. The full audience is the intersection, union or exclusion between datasets. Once you have defined your full audience, you can use Segment Builder to define your target audience, called a segment. A segment is the full audience filtered by category and attribute-based criteria, for example  “in market for car” and “earning more than 30k”.

In order to query and activate datasets, you need to use Audience Builder first and then Segment Builder. Both are a graphical way of analyzing data that can also be done using the Query Tool. However, the Query Tool’s IQL language is more versatile and does not place a limit on the number of datasets or criteria that can be included in a query.

Creating an audience

Adding datasets

1. Go to the Data tab and select Audience Builder.
2. Drag and drop the datasets you are interested in to the Datasets row.
   

• Datasets Total shows the number of datasets you have selected, as well as the maximum number of datasets you can select for an audience (i.e. 7).
• Total Rows shows the number of records that sits in the selected datasets (the row counts shown in the example screenshots may differ from those returned by the Platform).
• The datasets that appear under Data Sources are the list of insight datasets that you have permission to query (note that you cannot list activation datasets in Audience Builder).

By default, Audience Builder uses the Platform’s automatic choice of best key for matching and counting based on the highest fill rate and unique records for any connection between datasets. Users can override this behavior using Key Override. 

To find out the key being used, go to the Connections List and look at details for the selected key and intersection. An Intersection must exist between all selected datasets.

Use the following buttons to build your audience:

You can define your audience as follows:

• Union - to combine the audience from both datasets.
• Intersect - to only use people that appear in both datasets.
• Exclude - to only use people that appear in one dataset but not in another.

To combine datasets with other datasets click  to create another row. The new row can be used to intersect or exclude another row of datasets.

Use the Statistics pane to show the categories in a selected dataset. If only one dataset has the category you are interested in, you can filter the full audience on that category. This is because matching is done by key, but filtering is done by category.

Enriching your datasets

If the category you are interested in is not in either dataset, you can work with a partner who holds this information. You can then use their dataset to enrich your filters with additional categories that are not in your defined audience. For example, to use a Personal Income category to find customer records in the intersection with a salary over 30k if none of your selected datasets hold this category (as shown in the screen excerpts below).  

Enriching your datasets with a data partner that holds Personal Income for the intersection, as shown:

Now returns 38.9k customer records containing Personal Income for the intersection between the datasets, as shown. 

Note: An enrichment dataset does not change the total rows of your defined audience. This is because the enrichment dataset is applied to the defined audience only. 

Linking datasets

The same individuals may appear in different datasets, but will not be matched unless their records share a common key in all datasets. The LINK WITH row allows you to specify linking datasets that hold additional keys that are used to match records common to all datasets. You can work with identity partners to get the permissions to use their dataset. 

In the example below, the ACME and Autosports datasets do not share some uncommon identifiers for an individual (ACME has Cookie and Autosports has Mobile Phone Number). But the IdentityX dataset holds keys that can 'glue' the ACME and Autosports datasets together, so during a query the total rows can be improved.

This example shows how adding a linked dataset increased the total rows from 38.9k to 88.3K. 

In the example below, the ACME and AutosportsS datasets do not share any common identifiers for an individual. For this reason, the total rows is 0.

But the IdentityX dataset holds keys that can 'glue' these two datasets together. Linking with IdentityX increased the total rows from 0 to 82.9K.

Setting key overrides

Key overrides let you override the Platform’s automatic choice of the best key for matching based on the highest fill rate and unique records for connected datasets. To specify a key override:

1. Add a new row to the Key Overrides section.
2. In the Connect fields, select the source and destination path between datasets, and the path direction.
3. From the using prioritized keys field, select the prioritized key to match on. 

The Platform recalculates the total rows. For example, if the Platform selected key Email returned total rows of 38.9k, then specifying a key override using the Mobile Advertising ID prioritized key returns a reduced total rows of 19.8k, as shown:

Overriding the Platform’s automatically selected key creates a lower match rate, but adding multiple common keys in the prioritized key field can improve the match rate. If you choose to match on multiple keys, a match on any key counts toward the total rows (with multiple matches for a record counted only once) because it uses the OR function. If the datasets contain duplicates, changing the order of the keys can change the count total.

For example, adding both Email and Mobile Advertising ID prioritized keys to the key override returns matches for both keys and increases the match rate from 38.9k to 50.9k, as shown:

And changing the prioritized key order returns the same result:

Setting multiple key overrides

You can set more than one key override by adding more rows, as shown:

Saving your audience

When you have finished defining your audience, give it a name and select the Save button to make it available to Segment Builder, where you can drill down to the target audience. Once you have defined the target audience, it can be activated to a desired destination.

To revisit a saved audience, go to the Data tab and select it from Audiences.