Hunch

Lead scoring in Google Sheets without a CRM

Updated · Edoardo Panichi, maker of Hunch

You have a sheet of lead notes and no time to read them. Two columns rank the list: who can buy, and who fits. Both come from one formula each, written in plain English.

Column B: can this person buy?

=HUNCH(A2:A2000, "Can this person approve a software purchase for their team without asking someone else?")

In Excel: =HUNCH.ASK(A2:A2000, "Can this person approve a software purchase for their team without asking someone else?")

The answer is a probability from 0 to 1. The question carries the definition, which is what makes the number mean something: "decision maker" alone would leave the model guessing what you count.

Column C: does this person fit?

=HUNCHSCORE(A2:A2000, "ICP fit: B2B software company with 20 to 500 employees that sells to sales or support teams, based in the US or Europe", "no fit|weak|good|perfect")

In Excel: =HUNCH.SCORE(A2:A2000, "ICP fit: B2B software company with 20 to 500 employees that sells to sales or support teams, based in the US or Europe", "no fit|weak|good|perfect")

Write your ideal customer as one sentence and paste it into the question. Change the sentence and the whole column re-scores.

Real outputs

RowABC
1Lead notesCan buy? (0–1)ICP fit (0–3)
2Maria K., VP Sales at Northwind Logistics (300 employees, Chicago). Owns the sales tooling budget.0.672.34
3Tom R., SDR at a 12-person agency in Leeds. Asked for a free trial for himself.0.170.84
4Priya S., Head of Customer Support, Lumen Health (180 staff, Austin). Signs off on tools under $50k.0.792.36
5Student building a side project, wants API access for a hackathon.0.110.07
6Jonas W., CTO and co-founder of a 40-person B2B SaaS in Berlin.0.802.74
7Procurement analyst at a 5,000-person bank. Cannot say who decides.0.090.24
8Director of Revenue Operations, 250-person fintech, New York. Asked for pricing for 20 seats.0.532.62
9Elena P., Support Team Lead at a 90-person SaaS in Amsterdam. Her manager approves purchases.0.072.46

Elena fits the profile (2.46) but cannot sign (0.07); Jonas does both. That is exactly the split a single "lead score" hides, and why two columns beat one.

Rank and cut

=SORT(FILTER(A2:C2000, B2:B2000 >= E1, C2:C2000 >= 2), 2, FALSE)

Put the threshold in E1 (start at 0.5) so you can tune it without editing the formula. Rows above the line get a call this week; the rest get a nurture email.

Probabilities near 0.5 are the model saying it cannot tell from these notes. Read those rows yourself or add a line to the notes; do not round them to yes or no.

Questions

Does it read the company website or LinkedIn?
No. It judges the text in the cell. Put what you know in the notes column; a one-line summary per lead is enough.
How many rows per formula?
A range formula handles about 1,000 rows within Google's 30-second limit. For bigger lists use Hunch menu › Ask about selection, which runs up to six minutes and writes values.
Will the score change when I reopen the sheet?
Results are cached for six hours. When a column is final, select it and use Hunch › Freeze selection to values.

Run it on your own column

Copy the template (or use the Excel add-in), paste your key, point the formula at your data. Every formula here has its Excel spelling underneath. 100 rows free to start, then $29 for 5,000 rows. Credits never expire and there is no subscription.

Copy the templateGet 100 free rows

Reference: HUNCH · HUNCHSCORE

More guides