Two scores, and most tools quietly blend them into one
A lead can be a good fit and completely unresponsive, or a poor fit and extremely engaged. Those are different facts and they deserve different numbers. Below is what each of the two scores is made of — including the full point table, which is not usually something a product prints.
Two different scores
| Fit scoreShould we be talking to this business at all? | AI, ₹2.00 |
|---|---|
| Behaviour scoreWhat has this lead actually done? | Points, free |
A single blended number hides which of the two moved, which is exactly the thing you need to know before deciding what to do next.
The fit score
You describe your company and what you sell, once. Each lead is judged against that description and segmented, with a suggested opening line written for that specific business.
What it is made of
A business listing: what the business is, where it is, its category and public signals like rating. That is a genuinely useful basis for “is this the kind of company that buys what I sell” and a useless one for “do they have budget this quarter”.
Read it as a sort order
Its real job is putting the fifty most plausible businesses at the top of five hundred, so your first two hundred messages go somewhere sensible. It is not a verdict on any individual company.
The behaviour score
Points, added as things happen, across every channel. No model, no training data, nothing to be surprised by — you can audit a lead’s score by reading its history and doing the arithmetic.
The shape of the table is its whole argument: a reply is worth fifteen times a send, because a reply is the only thing on the list the lead had to choose to do.
Every weight
| Meeting booked | +50 |
|---|---|
| Meeting completed | +30 |
| Voice call — said they were interested | +25 |
| Marked a hot lead | +20 |
| Replied to email | +15 |
| Replied to SMS | +15 |
| Replied on WhatsApp | +15 |
| Clicked an email link | +10 |
| Voice call completed | +10 |
| Opened an email | +5 |
| Read a WhatsApp message | +5 |
| SMS delivered | +2 |
| WhatsApp delivered | +2 |
| Voice call initiated | +2 |
| Email sent | +1 |
| SMS sent | +1 |
| WhatsApp sent | +1 |
| Meeting no-show | −5 |
| Email bounced | −10 |
| Unsubscribed | −50 |
The negative ones
Three events subtract, and they are the ones most scoring models leave out because they make the numbers look worse.
- Unsubscribed, −50. The largest single movement in the table, in either direction. It should be: they have answered.
- Bounced, −10. Not a judgement on the lead, a judgement on the address. Keeping bounced contacts in rotation is how a sending reputation dies.
- No-show, −5. Small on purpose. People miss meetings for ordinary reasons; it is a nudge, not a verdict.
What a score cannot tell you
- Whether they have budget. Nothing in either score sees a balance sheet.
- Whether they already buy this from someone else. The single most common reason a well-fitting lead never converts, and it is invisible here.
- Whether the person who replied can decide anything.
- Whether now is the right time. A score has no sense of when.
Which is the honest summary of lead scoring generally: it is a good way to decide what order to work a list in, and a bad way to decide whether to work a lead at all.
Questions
- Why is a booked meeting worth as much as an unsubscribe costs?
- Because both are near-certain signals in opposite directions, and the symmetry is deliberate. A lead who unsubscribes has told you the answer; keeping them anywhere near the top of a list wastes the next message and risks a complaint.
- Does sending more messages raise a lead’s score?
- Barely, and that is on purpose. A send is worth 1 point; a reply is worth 15. If activity counted for much, the score would measure your effort rather than their interest, which is the classic way a scoring model becomes a mirror.
- What does the AI fit score cost?
- ₹2.00 per lead, per run, charged when you score. The behaviour score costs nothing — it is arithmetic on events that already happened.