Utilizing the Audiences feature requires several steps to create and send audiences to downstream systems, e.g. Facebook, Yahoo!, Google Analytics, Nielsen. The high-level steps of how and what data is available for defining attributes, how to utilize the attributes to create audiences, and how to select the audiences to send to the downstream systems are described in the article Building an Audience Workflow.
This article describes the details of the produced results after the second of those steps, Creating Attribute Builder apps, is completed:
- Attribute Table Name
- Attribute result tables
- Attribute Metric Window Length field impact
- Reviewing attribute results
Attribute Table Name
The name of the attribute table(s) is defined within a data layer in the Configuration menu via the configuration gear icon (
). This value is unchangeable after the data layer is created.
While only one value is given, there are three table names created based on the value. Details of these tables are described below.
Attribute result tables
Results table
The results table contains the unique identifier values with each attribute value it matches per partition of when the condition was met. For example, visitor 12345 met the condition for the attribute brand ABC on Jan 1.
The name that is seen in the field, defaulted to syn_attribute, can only be changed at the time of data layer creation. This will be the exact name of the table(s) that hold the results of all attribute builder processes created within this data layer.
The development and production results table(s) will be created and stored in the respective paths and databases noted in the selected event store for the data layer.
Metadata
The metadata table contains the unique attribute values for all attribute builder apps in the data layer. For example, attribute group Brand has attribute value X.
The table(s) created for storing the metadata are named with the name found in the field along with "_metadata" added as a suffix. As with the results table(s), there will be a development and production metadata tables stored in the paths and databases defined in the event store for the data layer.
Metrics
The metrics table contains statistics of the number of unique identifier values matching each attribute value condition per partition, usually per day. For example, 99 visitors met the condition for the attribute brand ABC on Jan 1.
The table(s) created for storing the metrics are named with the name found in the field along with "_metrics" added as a suffix. As with the results and metadata table(s), there will be a development and production metrics tables stored in the paths and databases defined in the event store for the data layer.
Attribute Metric Window Length field impact
This field provides the breadth of a secondary analysis period for meeting attribute conditions. This is in addition to the recency field that is set at the attribute group. Thus, when reviewing attribute metric results there are two distinct time periods displaying counts for the number of visitors matching an attribute.
The attribute metrics table contains two sets of metrics: counts_per_day and counts_last_n_days. The value entered in this field, which is defaulted at 7 days, represents the N found in counts_last_n_days as explained below:
- counts_per_day - As attribute results are generated, this column displays the number of unique identifier values matching each attribute value condition per partition, usually per day, for the period defined by the recency of the attribute group.
- counts_last_n_days - As attribute results are generated, this column displays the number of unique identifier values matching each attribute value condition per partition, usually per day, for the period defined by the Attribute Metric Window Length plus the recency of the attribute group.
For example, we have the below number of visitors that met the condition for the attribute brand ABC per the noted days:
| Event Date | Num IDs |
| Jan 1 | 10 |
| Jan 2 | 30 |
| Jan 3 | 20 |
| Jan 4 | 25 |
| Jan 5 | 15 |
Adding to our example, if we have the recency of the attribute set to 1 and the Attribute Metric Window Length set to 2, then the attribute metrics table would display the following:
| Partition | Attr Group | Attribute | counts_per_day | counts_last_n_days |
| Jan 1 | Brand | ABC | 10 | - |
| Jan 2 | Brand | ABC | 30 | - |
| Jan 3 | Brand | ABC | 20 | - |
| Jan 4 | Brand | ABC | 25 | 60 |
| Jan 5 | Brand | ABC | 15 | 75 |
Since the recency is set to 1 the counts_per_day time period is a single day and therefore matching the counts per event date.
Since the Attribute Metric Window Length is set to 2 the counts_last_n_days time period is 3 days (recency + window length) and therefore summing the previous three days.
The above is a simplified example. In practice, the counts_last_n_days will not be an exact sum of the previous N days, but rather the number of unique visitors across the N days time period. This will eliminate the double-counting of visitors that appear multiple times across days.
Reviewing attribute results
Table summaries
All datasets throughout the platform have several properties, Details, Schema, Preview, Stats, that are available and shown by clicking on the dataset. All properties are viewable by clicking on the dataset from the workflow canvas, including attributes and audiences, but attributes and audiences datasets are also available from within the data layer screen.
Details & Schema
While viewing from within the data layer screen, these two properties, Details and Schema, are combined into a single view. The Details portion, seen at the top of the screen, provides details on several technical aspects of the table; the Schema portion, seen at the bottom of the screen, provides the list of columns that exist for the dataset.
Preview
As the name suggests, this section displays a preview, and thus a limited number of results, of the actual data found within the selected dataset. The data can be sorted and filtered from within the application or can be downloaded into an xlsx file for review outside of the application. The download is limited to the data found within the preview screen.
Stats
This section, also see as State from with the workflow canvas, displays statistical information about the contents of the dataset, i.e. number of rows and size per each partition, usually date, but can be per hour.
Table contents
Results table
As noted above, the results table contains the unique identifier values with each attribute value it matches per partition of when the condition was met. In the first example below, visitor -231830... met the condition for the product attribute Clothes on Jan 1.
Metadata
As noted above, the metadata table contains the unique attribute values for all attribute builder apps in the data layer. In the example below, the attribute group Product has attribute values Accessories, Equipment, and Clothes.
Metrics
As noted above, the metrics table contains statistics of the number of unique identifier values matching each attribute value condition per partition, usually per day. In the example below, 733 visitors met the condition for the product attribute Equipment on Jan 1.
As explained above regarding the Attribute Metric Window Length, the counts_per_day and counts_last_n_days show counts for the attribute across two different time periods.
The column counts_per_day represents the number of unique visitors for the given attribute value that is calculated on the day seen in attribute_partition. The timespan for the summation is based on the value of the recency of the attribute group. For example, the 733 seen above will be the number of visitors seen for Dec 31 - Jan 1 if the recency of the attribute group Product is set to 2.
The column counts_last_n_days represents the number of unique visitors for the given attribute value that is calculated for the prior N days. The value of N is the setting of Attribute Metric Window Length found within the data layer.