A score your reps will argue with, and then agree to.
Most account scores fail for one reason: nobody can explain them. Kairo shows the six touches behind an 87, lets you change what counts, and keeps the history so a score movement is always traceable.
| Signal | What it measures | Weight |
|---|---|---|
| Committee breadth | Distinct job functions engaged in 30d | 30% |
| Depth | Touches per engaged contact | 20% |
| Seniority | Director-and-above involvement | 20% |
| Recency | Days since last qualifying touch | 15% |
| Creative stage | Bottom-funnel asset engagement | 10% |
| Site behaviour | Pricing / docs visits, if tag installed | 5% |
You set the weights. Kairo shows the arithmetic.
Start from a default model that works for most B2B programmes, then move the weights as you learn what actually precedes a meeting in your market. Every change is versioned, and you can compare two models against the same 90 days before you commit to one.
- Per-account expansion showing the touches that produced the number
- Score history, so you can see what a score was on the day a rep called
- Model versioning and A/B comparison on historical data
- Separate models per target list — enterprise and mid-market rarely score the same way
Four stages, and the honesty to include the first one
Not served
On the list, never delivered an impression. Usually an audience-size or budget-allocation problem, and always the cheapest thing to fix.
Reached, thin
Impressions delivered, one function, low frequency. Not a signal. Do not hand this to a rep.
Multiple functions
Two or more job functions engaging inside the window, with repeat exposure. Worth a nurture sequence, not yet a call.
Committee moving
Breadth, depth and seniority all present, with recent bottom-funnel engagement. This is the list your AE should get on Monday.
A score is only useful where the work happens
Into the CRM
Score, stage, last movement and the top three contributing touches written onto the account record in HubSpot, Salesforce or Attio.
Into sequencing
Threshold crossings pushed to Clay, Outreach or Salesloft so an account entering In market can trigger a sequence without a human noticing first.
Into anything else
Signed webhooks on every transition, plus the REST API and MCP server if you would rather pull than be pushed.
Build a score model your sales team will actually use.
Start on the default, change one weight, and compare both against the last 90 days before you roll it out.