Predictive Audiences Reporting

After you save a Predictive Audiences model, Audience Manager starts training it. Within a couple of hours, the computed model will start analyzing audiences on the Data Collection Servers. Reporting will be available the following day.

To see the results of your Predictive Audiences classification, go to Audience Data > Models, and click your model in the list.

Use the filtering options on the left hand side to search for the model name or filter the results based on model type.

predictive-audiences-filter

The models table shows you the following information:

  • ID: the model ID uniquely identifies each model in your Audience Manager account;

  • Name: the name that you provided in the model creation step;

  • Description: the description that you provided in the model creation step;

  • Model Type: the type of each model (Look-Alike Modeling or Predictive Audiences);

  • Status: the status of each model:

    • Pending: the model is initializing and will start producing results shortly;
    • Active: the model is running successfully and producing results;
    • Warning: the model failed to produce results, due to insufficient data (i.e. low baselines population, user profiles are not rich);
    • Error: the model failed to run. You should contact your Adobe representative.

Model Overview Report :headding-anchor:model-report

Once you choose a model, its reporting page will load. At the top of the page you can see the top 5 largest predictive segments, based on 1 day real-time realization, that the model has classified your target audience by. The Other category includes the rest of the personas, which were not included in the top 5 largest predictive segments.

Audience Manager displays both a color coded donut chart and a timeline graph for your Predictive Audiences.

Clicking the personas tabs at the top of the page adds or removes them from the chart and graph.

The donut chart shows you a persona-based breakdown of your target audience, while the graph shows you the 1-day real time population trend of your predictive segments over the past 6 days.

If the model status is Pending, Warning, or Error, the model status is displayed instead of the graphs.

smart-persona-report

The report table shows you the following information for each Predictive Audiences segment.

  1. SEGMENT ID: the segment ID of the auto-created segment associated to each persona;

  2. NAME: the persona name;

  3. STATUS: the status of the Predictive Audiences segment:

    • Succeeded: users are being classified into this segment;
    • Pending: the segment is still being initialized;
    • Insufficient Training Data: users are not being classified into this segment due to insufficient data. The total baseline population is too low and does not provide enough data to learn from.
  4. 1 DAY REAL TIME POPULATION: The number of segment realizations for each persona, over the past 24 hours.

  5. 1 DAY REAL TIME POPULATION %: The percentage of segment realizations for each persona, over the past 24 hours, out of the total model population.

Influential Traits :headding-anchor:influential-traits

Influential Traits are traits that the Predictive Audiences algorithm discovered to be the strongest predictors for determining the persona classification of a visitor.

Their sign indicates whether the presence of the trait increases(+) or decreases(-) the likelihood of the user belonging to the selected persona.

To view the influential traits for all of your personas, click View All Influential Traits.

The Influential Traits window shows you the following information, for each persona from the selected model:

influential-traits

  1. TRAIT ID: the trait ID of each influential trait for the selected persona;
  2. NAME: the name of each influential trait for the selected persona;
  3. DESCRIPTION: the description of each influential trait for the selected persona;
  4. WEIGHT: the weight of each influential trait for the selected persona. Influential Traits are by default sorted by weight, in descending order. The value of the weights indicate their predictive power. The sign indicates whether the presence of the trait increases(+) or decreases(-) the likelihood of belonging to a persona.
  5. 30 DAY REAL TIME POPULATION: the number of unique trait realizations for each influential trait for the selected persona, over the past 30 days.
recommendation-more-help
de293fbf-b489-49b0-8daa-51ed303af695