Audience Preview
Audience preview lets you inspect a sample of entity records that match your audience definition before saving or activating. This helps you verify that the right customers are being captured by your filter conditions.
How Preview Works
When you click Preview in the audience builder, SignalSmith:
- Compiles the filter conditions into a SQL query
- Adds a
LIMITclause to return only a sample (up to 100 rows) - Joins entity attributes and trait values to the result
- Executes the query against your warehouse
- Displays the results in a table view
The preview runs the actual audience query (not a sampled estimate), so the members shown are real matches — not approximations. The LIMIT simply caps the number of results returned for performance.
Preview Columns
The preview table displays the following columns:
| Column Type | Description |
|---|---|
| Entity Key | The primary identifier for the entity (e.g., user_id) |
| Entity Attributes | Key attributes from the entity type’s mapped table (e.g., email, name, country) |
| Trait Values | The current materialized values of traits referenced in the audience conditions |
| Additional Traits | Other traits attached to the entity type, for context |
Column Selection
By default, the preview shows:
- The entity key
- Up to 5 entity attributes (configurable in the preview settings)
- All traits referenced in the filter conditions
You can customize which columns are shown by clicking the Columns button above the preview table and toggling columns on or off.
Using Preview Effectively
Validate Filter Logic
After building your conditions, preview the members and ask:
- Are these the types of customers I intended to target?
- Do the trait values look reasonable for the conditions I set?
- Are there any unexpected records that suggest a filter condition is too broad?
Spot Data Issues
Preview can reveal data quality problems:
- Null values in unexpected columns may indicate incomplete data ingestion
- Extreme outliers (e.g., a lifetime value of $1,000,000) may indicate data errors
- Duplicate-looking records may suggest identity resolution is needed
Compare Before and After
When editing an audience, preview both before and after your changes to understand how the membership shifts. This is especially useful when tightening or loosening filter conditions.
Preview vs. Estimation
| Feature | Estimation | Preview |
|---|---|---|
| Returns | Approximate count | Sample records |
| Speed | 2-10 seconds | 3-15 seconds |
| Accuracy | Approximate | Exact (for the sample shown) |
| Use for | ”How big is this audience?" | "Who is in this audience?” |
| Sampling | Table-level sampling | Full query with LIMIT |
Use estimation for quick size checks and preview for qualitative validation of membership.
Preview Limitations
- Sample size — Preview returns up to 100 records. For audiences with millions of members, this is a small window into the full membership.
- No ordering guarantee — The preview results are not ordered in any specific way. The 100 records returned are determined by the warehouse’s query execution plan.
- Stale trait values — If a trait hasn’t been evaluated recently, the preview shows the last materialized values, which may not reflect the latest source data.
- Preview vs. full evaluation — The preview result is not cached or stored. It’s a one-time query for validation purposes. The actual membership list is only created when you save and evaluate the audience.
Preview from the Audience Detail Page
For saved audiences, you can also access preview from the audience detail page:
- Navigate to Segment > Audiences and select an audience
- Click the Preview tab
- The preview runs against the current filter conditions and displays sample members
This is useful for monitoring an existing audience without entering the edit mode.
Next Steps
- Size Estimation — Get approximate member counts
- Creating an Audience — Full audience creation guide
- Audience Syncs — Activate your audience to destinations