Skip to main content

Using a Datalab

Step 1: Upload your datasets​

Upload your prepared datasets to the Dataset Portal using one of these methods:

Make sure your data is properly formatted according to the dataset requirements.

Step 2: Provision datasets to the Datalab​

Open the Datalab you created and provision each dataset:

  • Click Provision dataset next to the dataset slot
  • Select Choose from my stored datasets
Provision dataset from stored datasets

If a Datalab dataset has outdated data that needs to be refreshed, deprovision the existing dataset first:

Deprovision outdated dataset
danger

Note that currently Datalabs need to be reprovisioned to Media DCRs for change of data to take effect (we're working on a better UX).

Step 3: Validation​

Click Validate Datalab to start validation. Validation checks that:

  • Each dataset matches its expected format
  • Identifiers match the configured identifier type
  • userId values are consistent across all datasets

Validation may take up to a few hours, depending on data size. Once successful, the status turns green and the Datalab is ready to be provisioned to a Media DCR:

Datalab validation completed - Ready status

Datalab statistics​

After validation, the Datalab page shows aggregated statistics grouped by dataset (Identifiers, Segments, Demographics, Embeddings). Use the filter to display only specific statistics, or download the full report as JSON.

Datalab statistics dashboard
tip

Publishers are strongly advised to review the dashboard with a Decentriq expert to make sure the data is ready for a Media DCR. Contact Decentriq to book a session.

note

Aggregated statistics can be shared with Decentriq admins, depending on your organization's statistics sharing setting. The per-dataset validation reports — which are downloaded as separate files — are never shared. If you hit a validation error, paste the report contents into an email to support@decentriq.com.

Identifiers overview​

Always shown.

Which IDs are available? For each, how many distinct IDs exist? To how many userIds does each ID map to?

Shows if the expected number of IDs have been uploaded.

Identifiers overview

Basic identifiers statistics​

Always shown.

What are the basic data quality statistics for identifiers?

Helps spot anomalies.

Basic identifiers statistics

Remarketing activation potential​

Shown when the Datalab supports the Remarketing audiences collaboration type.

When matching on a given matching ID, how many distinct activation IDs can be reached?

Helps understand the potential of a remarketing campaign for a given matching / activation ID pair.

Remarketing activation potential

Basic segments statistics​

Shown when a Segments dataset is provided.

What are the basic data quality statistics for segments?

Helps spot anomalies before running deeper matchability or activation analysis.

Basic segments statistics

Segments activation potential​

Shown when the Datalab supports the AI lookalike or Rule-based audiences collaboration type and a Segments dataset is provided.

For the users with at least one segment, how many activation IDs can be reached?

Higher activation counts mean greater audience reach for targeted segment-based campaigns.

Segments activation potential

Segment user counts​

Shown when the Datalab supports Insights and a Segments dataset is provided.

How is segment membership distributed across different user groups?

Helps compare segment representation between all users and users you can match via different identifier types.

Segment user counts

Segments-per-user counts​

Shown when the Datalab supports the AI lookalike or Rule-based audiences collaboration type and a Segments dataset is provided.

How many segments does each user belong to?

Shows whether segmentation is evenly distributed or concentrated. Toggle between all users and matchable subsets.

Segments-per-user counts

Basic demographics statistics​

Shown when a Demographics dataset is provided.

What are the basic data quality statistics for demographics?

Helps spot anomalies before running deeper matchability or activation analysis.

Basic demographics statistics

Demographic user counts​

Shown when the Datalab supports the AI lookalike or Rule-based audiences collaboration type and a Demographics dataset is provided.

How does age and gender distribution compare between all users and matched users?

Helps compare demographic representation across the total dataset versus the subset of users you can uniquely match.

Demographic user counts

Basic embeddings statistics​

Shown when an Embeddings dataset is provided.

What are the basic data quality statistics for embeddings?

Helps spot anomalies before running deeper matchability or activation analysis.

Basic embeddings statistics

Embeddings activation potential​

Shown when the Datalab supports the AI lookalike or Rule-based audiences collaboration type and an Embeddings dataset is provided.

For the users with an embedding, how many activation IDs can be reached?

The more users that have both an embedding value and an activation ID, the better the lookalike audience can be activated.

Embeddings activation potential

Matchable embeddings users​

Shown when the Datalab supports the AI lookalike or Rule-based audiences collaboration type and an Embeddings dataset is provided.

How many matchable users have embeddings, enabling AI-driven audience expansion?

The more users that have both a matching ID and an embedding value, the more likely the seed audience users can be found for lookalike modeling.

Matchable embeddings users

Step 4: Provisioning to Media DCRs​

Once your Datalab is validated and shows a Ready status, you can provision it to Media DCRs. See the Media DCR Data tab for the provisioning process and how to monitor results.

A single Datalab can be provisioned to multiple Media DCRs simultaneously. To keep your Datalab current with fresh data, see Refreshing the base audience.