Documentation
Understanding your dashboard
What each surface is for, and how to read a number without misreading it.
The console follows the loop: measure, diagnose, fix, prove. Everything else is a lens on one of those four.
| Surface | Answers the question |
|---|---|
| Overview | Where do we stand right now, and did anything real move? |
| Prompts | Which buyer questions do we win, lose, or never appear in? |
| Diagnose | Who gets named instead, and which sources did the engine use? |
| Sources | Which domains do the engines pull from, per engine? |
| Win queue | What should we do next, ranked by what it is worth? |
| Approvals | What is drafted and waiting for a human yes? |
| Prove | What did we ship, and did the numbers actually move? |
| Clients | How does the whole book compare, and who opened their report? |
| Locations | How do we look block by block, location by location? |
Reading a rate correctly
- A percentage always carries the count it came from. 30% of 10 answers and 30% of 300 answers are different facts.
- Too few answers renders plain counts instead of a percentage, because a percentage there would be false precision.
- A dash means unmeasured. It never means zero, and zero never renders as a dash.
- There are no rank positions anywhere, for AI answers. Ask twice and the order changes, so any average position would be invented.