# Lead scoring in Excel without a CRM

`=HUNCH.ASK` reads a line of lead notes and returns the probability that this person can approve a purchase on their own. `=HUNCH.SCORE` next to it returns how well they match your ideal customer, so the two columns together give SORTBY something to rank.

Updated 28 September 2026. Source: https://hunchsheet.app/formulas/lead-scoring-excel

## Column B: can this person buy

```
=HUNCH.ASK(A2:A9, "Can this person approve a purchase for their team without asking someone else?")
```
In Google Sheets: `=HUNCH(A2:A9, "Can this person approve a purchase for their team without asking someone else?")`

The answer is a probability from 0 to 1. The question carries the definition of "decision maker", which is what makes the number mean something specific rather than a vague sense of seniority.

## Column C: does this person fit

```
=HUNCH.SCORE(A2:A9, "ICP fit: marketing agency with 5 to 50 employees that manages paid ad campaigns for other companies, based in North America", "no fit|weak|good|perfect")
```
In Google Sheets: `=HUNCHSCORE(A2:A9, "ICP fit: marketing agency with 5 to 50 employees that manages paid ad campaigns for other companies, based in North America", "no fit|weak|good|perfect")`

Write your ideal customer as one sentence: company type, size, who they sell to, where they are. Put that sentence in a cell of its own and reference it in the formula, so changing the ICP re-scores the whole column without touching the formula itself:

```
=HUNCH.SCORE(A2:A9, $F$1, "no fit|weak|good|perfect")
```
In Google Sheets: `=HUNCHSCORE(A2:A9, $F$1, "no fit|weak|good|perfect")`

## Real outputs

Eight lead notes for a marketing-agency tool, with the model's answers as returned on 28 September 2026.

| Lead notes | Can buy? (0–1) | ICP fit (0–3) |
| --- | --- | --- |
| Growth lead at a 15-person paid social agency in Austin. Manages the media budget directly. | 0.71 | 2.99 |
| Freelancer asking about a single seat for a personal project. | 0.17 | 0.09 |
| VP Marketing at a 30-person performance agency in Toronto, signs off on all vendor contracts. | 0.86 | 2.89 |
| Intern researching tools for a school assignment. | 0.08 | 0.18 |
| Owner of a 6-person boutique agency in Miami running Facebook ads for local restaurants. | 0.81 | 2.97 |
| Analyst at a 2,000-person holding company, says budget decisions go through corporate. | 0.15 | 0.06 |
| Co-founder of an 80-person performance marketing shop in London, evaluating for the whole team. | 0.69 | 0.48 |
| Account manager, no budget authority, wants a demo to show her director. | 0.05 | 0.63 |

Row 7 is the one that shows the ICP sentence doing real work: an 80-person performance marketing shop is exactly the kind of company the sentence describes, and the co-founder can clearly buy (0.69), but fit lands at only 0.48 because the sentence says 5 to 50 employees and this company has 80. The size clause is not decorative; loosen it if a larger agency should still count as a fit. Row 8 shows the opposite problem: the note says nothing about the company at all, only that this person cannot approve a purchase (0.05), and fit still comes back at 0.63 rather than 0, because there is nothing in the text to rule it out either. Thin notes do not fail loudly, they drift toward the middle, which is why a one-line summary of company type and size is worth the extra ten seconds per lead.

## Rank and cut

```
=SORTBY(FILTER(A2:C9, (B2:B9>=E1)*(C2:C9>=2)), C2:C9, -1)
```

Put the probability threshold in E1, starting at 0.5, so you can tune it without editing the formula. Rows that pass both filters get a call this week; everyone else gets a nurture email or nothing at all.

> Tip: A probability near 0.5 is the model saying the notes do not say enough to tell. Read that row yourself, or add one more line to the notes, rather than rounding it to yes or no.

## Limits

- The model judges only the text in the cell. It does not look up the company website, LinkedIn, or firmographic data.
- A one-line note gives the model less to work with than three lines, and the score will sit closer to the middle. That is honest, not broken.
- Two formulas per row is two credits per row. A 2,000-lead list is 4,000 credits, about $23 on the Starter and Top-up packs combined.
- Cells calculating together travel in requests of up to 40 rows, and answers are cached for six hours inside Excel.

## Setup in three steps

1. Sideload the add-in from [hunchsheet.app/excel](https://hunchsheet.app/excel).
2. Save your key once: `=HUNCH.SETKEY("hunch_...")`, then delete the cell.
3. Type both formulas in row 2, then fill both down together.

## See also

- [HUNCH.ASK reference](https://hunchsheet.app/docs/hunch) and [HUNCH.SCORE reference](https://hunchsheet.app/docs/hunchscore).
- [The same two-column pattern in Google Sheets](https://hunchsheet.app/formulas/lead-scoring-google-sheets), and [scoring ICP fit alone](https://hunchsheet.app/formulas/icp-fit-score-spreadsheet) if you only need that column.
- [AI formulas in Excel](https://hunchsheet.app/formulas/ai-formulas-excel), an overview of all four formulas.

## Questions

### Does it check the company website or LinkedIn for me?

No, it judges the text you put in the cell. A one-line summary of what you know about the lead is enough; write down title, company size, and anything they said about budget.

### Why two columns instead of one combined lead score?

Authority and fit fail for different reasons, and averaging them hides which one is the problem. An account manager at a perfect-fit company and a founder at a company that is a poor fit can average to the same score for opposite reasons.

### Can I combine the two columns into one number anyway?

Yes, with a normal formula once both columns exist: `=B2*0.6 + (C2/3)*0.4`. Keep the raw columns too; changing the weights then costs nothing.

### What happens when I change the ICP sentence?

Every row referencing that cell re-scores the next time the sheet calculates. Keep the sentence in the words your lead notes use; if your notes say "SMB", say SMB in the ICP sentence rather than an employee-count range.

### How many leads can one column handle?

There is no hard row limit. Requests travel in batches of 40, so a 5,000-row list runs as roughly 125 batches; for very large lists, paste the finished columns as values before adding more rows.