Discovery: reading your map
How to read a finished run: the segment landscape as a map of your customer base, KPI lenses as overlays, and each segment's field guide.
When a segmentation run finishes and Neuralift publishes it, the Discovery tab shows every segment in the run: the segment landscape sized by population, KPI lenses to recolor it by any metric you care about, and a detail panel per segment with its insights, defining features, and persona. These segments were discovered, not defined: nobody wrote rules for them, the model found the structure in your data.
The whole tab reads like a map. Data preparation was the survey: the engineered features are its instruments, measuring each customer. The landscape is the map those measurements produce: each segment a region, its area the population living there. KPI lenses are the elevation overlays you drape across it, and each segment’s insights and persona are the field guide for one region. Later, the Actions section is where you plot routes across that terrain.
The reading order
Read the map wide-to-narrow:
| Read | What it answers | Where |
|---|---|---|
| 1. The segment landscape | Who are my customers, and how big is each group? | Discovery tab |
| 2. KPI lenses & lift | Where does each KPI concentrate, and which segments over- or under-index? | Discovery tab, lens bar |
| 3. Inside a segment | What defines this group, and who are they as people? | Discovery tab, right-hand panel |
Start with the landscape, the whole terrain in proportion, so you get a feel for the shape of your customer base before reading any single story. Then activate the lenses: recoloring the landscape by one KPI at a time shows where revenue, churn risk, or engagement actually concentrates, which is rarely where the biggest regions are. Only then go inside a segment, where its field guide is waiting: a summary, confidence-ranked insights, defining features, and a persona that makes the group memorable to your wider team.
Both result tabs stay disabled in the sidebar until the run has results to show. If you’re a member rather than an admin, you’ll see results once Neuralift publishes the run; see Roles & permissions.
Where segments come from
Segments aren’t rule-based buckets you defined up front. Neuralift’s engine learns a compact neural representation of every customer from your prepared dataset, then finds naturally dense groups of genuinely similar customers within it. The goal and KPIs you set in the use case definition are used to interpret and present those groups (how they’re named, explained, and prioritized), not to guide the discovery itself. Customers that don’t fit any group with confidence are left unclassified rather than forced into a poor match. For the full method, see How the segmenter works.
Where to go next
Once you can read the map, the next section is about moving across it:
- Start with the bridge page: From insight to action.
- Review the run-wide plays: Actions & action plans.
- See the evidence behind them: Segment tactics.