Platform · Scorecards
LiveThe weakest behavior in every store, with the clip that proves it.
Weekly scorecards by store, district and associate. Behavior hit rates, trend over the quarter, and a review loop that makes the model better on your playbook.
Every scorecard states the capture coverage behind it.
The weekly scorecard
Weekly scorecard
Northgate Fine Jewelers, Store 14
Week of March 10 · 214 conversations scored
| Associate | Greeting | Budget framed | Occasion asked | Second item | Warranty | Follow-up set | Trend |
|---|---|---|---|---|---|---|---|
| Marcus D. | 96% | 88% | 82% | 74% | 61% | 70% | |
| Priya N. | 92% | 71% | 68% | 44% | 52% | 58% | |
| Dana R. | 89% | 64% | 55% | 31% | 40% | 47% | |
| Chris L. | 84% | 52% | 47% | 22% | 35% | 29% | |
| Elena V. | 91% | 60% | 51% | 38% | 44% | 41% | |
| Tomas B. | 78% | 45% | 39% | 18% | 27% | 24% |
Second item is the weakest behavior in this store. Six coachable clips are attached.
Sample data on a sample playbook.
Capabilities
What a manager gets every Monday.
Weekly, by store, district and associate
The same grid rolls up three ways, so a district leader and a store leader argue from one number instead of two reports that disagree.
Behavior-level hit rates
Not a composite score. The rate at which each behavior actually happened, per associate, so the weakest behavior in a store is visible at a glance.
Every score carries its clip
Each cell opens the conversations behind it. A coaching conversation that starts from a recording does not turn into a debate about whether the number is fair.
Trend views
Rolling windows over the quarter separate a good week from a changed habit, which is the only distinction that matters when you are deciding whether coaching worked.
Peak-season readiness
Ahead of the season, a readiness view shows which behaviors are strong enough to carry volume and which stores will lose money on attach when the queue builds.
A manager review loop that compounds
Managers agree or disagree with individual scores. Those decisions train the taxonomy, so the model gets more accurate on your playbook every week it runs. This is why year two is better than year one.
In practice
How this works in a store
On Monday morning a district leader opens one page for four stores. Revenue already told her Store 14 is behind. The scorecard tells her that Store 14 offers a second item in twenty-two percent of eligible conversations while Store 9 does it in sixty-one, and that the gap sits with three specific associates.
She clicks the weakest cell and hears two conversations where the moment passed. She does not need to visit the store to know what the coaching conversation is about, and she arrives with evidence instead of an impression.
The store leader, meanwhile, disagrees with one score: the customer had explicitly declined an add-on earlier in the conversation. He marks it, the score is reviewed, and the taxonomy learns that this pattern is an exclusion rather than a miss. Next quarter, that mistake happens less.
Technical detail
The technical detail, in plain language
- Cadence and windows
- Scorecards publish weekly, with four-week and thirteen-week rolling views for trend. Season views compare against the same weeks last year where data exists.
- Rollups
- Associate, store, district and region. Rollups are behavior rates, not averaged composite scores, so a large store does not hide a weak counter.
- Denominators
- A behavior is scored only on conversations where it was eligible. Hit rates are therefore rates over eligible conversations, and the eligibility rule for each behavior is documented.
- Evidence links
- Every cell resolves to the underlying conversations, each with transcript and audio, subject to the retention window you configure.
- Review loop
- Manager agree or disagree decisions are stored against the conversation and used as labelled feedback when the taxonomy is retrained.
- Capture coverage
- Every scorecard states the capture coverage behind it, so a week with a device offline is not read as a behavior collapse.
FAQ
Questions we get asked
Build this grid from your behaviors.
We will encode three of your behaviors on a demo call and show you the scorecard they produce.