/leads in the main menu.
The leads table with the product picker marked 1, the quality filters marked 2 and a row's review button marked 3
1

Pick the product

Scores belong to a product, so the ranking changes entirely when you switch. Without a product there is nothing to rank.
2

Filter by quality

Each pill carries its own count, so you can see how the list splits before opening anything.
3

Review a lead

Opens the Human review box for that person, described below.
A lead is a customer scored against a product. Pick a product at the top of the page and the table ranks your customers by how well they fit it. Without a product there is nothing to rank, so create one first — see Product Catalog.

​
Reading the table

Columns are the contact, their fit, when the score was last updated, and the action button. Fit shows as a signed percentage from −100% to +100% and a coloured badge:
BadgeFitMeaning
Hot70% and upStrong match on the signals you defined
Warm30% to 70%Partial match
Neutral−30% to 30%Nothing much either way
Low fitbelow −30%The signals point away from this customer
Not scored—Too few signals assessed to publish a number
The pills above the table filter to one badge, and the search box narrows by name or email.

​
Why a customer scored what they did

Expand a row and the right-hand side breaks the score into three groups:
  • Who they are — what their profile and company say.
  • What they say — what the conversation revealed.
  • How they engage — opens, clicks, replies.
Each group lists its signals with a ✓, ✗ or – and the reason the scorer gave. A group nobody has assessed yet is dropped from the overall rather than counted as zero, which is why a thin profile reads “Not scored” instead of 0%.

​
Human review

The clipboard button on a row opens Human review: one box for what you know that the scorer does not.
1

Write what changes the picture

For example: “Customer is junior, not a decision maker yet — discount technical-fit weight.” Or: “Company under 50 employees, not a strong fit.”
2

Submit

The note is added to the customer, and a re-score runs across every signal on the product.
3

Wait for the worker

Scoring runs in the background; the table refreshes with the new ranking.
What you write becomes context for the next scoring run, so specifics beat verdicts. “Their stack is all Java, our SDK is Node-only” teaches the scorer more than “bad fit”.

​
Re-score one customer

Customer drawer → ”…” → Re-score now.