# Forty Questions

> Measures how often AI assistants name a company when buyers ask the questions
> that precede a purchase, then does the work to change it. For B2B software
> companies selling into the UK, Singapore, Australia and New Zealand.

Last updated: 2026-09-22

## What it measures

Forty fixed questions, asked three times each, across four counted assistants —
ChatGPT, Claude, Perplexity and Google AI Mode — for 480 answers a month.
Gemini adds 120 more, reported beside the number rather than inside it, because
its terms do not permit the automated grounded route.

Reported every month: how many answers named the company, in what position,
which competitors were named instead, and every page the assistants cited
before answering.

## What makes it different

- The method is published: frozen question set, frozen configuration, cold
  logged-out sessions, counting rules decided in code before collection.
  No other tool in this category publishes sample size or run counts.
- The subscription includes the fix work, not only the measurement.
- The roster stops at ten companies, because one person does the fix work.

## Measured finding, September 2026

Adding six words to a question — "name specific products, don't hedge" — moved
one Singapore point-of-sale vendor from named in 3 of 5 answers to 5 of 5, and
from second on recommendation share to first. Same assistant, same questions,
fresh sessions throughout. One variable.

Across 1,053 citations in 80 recorded answers, 48% went to ten domains, and
eight of those ten belong to companies selling in that category.

## What it does not claim

It does not promise a higher count. Nobody controls what an assistant says.
It counts naming, not revenue. It measures a logged-out stranger, not a user
with months of chat history.

## Pages

- [Home](https://fortyquestions.io/): method, limits, pricing, sample report structure
- [Method](https://fortyquestions.io/method): the full rulebook — frozen questions, three repeats on three days, logged-out sessions, counting rules fixed before collection
- [What it costs to measure yourself](https://fortyquestions.io/cost): measured per-answer API costs from 151 billed calls — ChatGPT $0.0359, Claude $0.0408, Perplexity $0.0078, Google AI Mode $0.0250, Gemini $0.00
- [Comparison](https://fortyquestions.io/compare): Profound, Peec, Otterly, Athena, Goodie, Evertune, Semrush, Ahrefs — prices and methodology transparency, checked Sept 2026
- [Pricing](https://fortyquestions.io/pricing.md): machine-readable
- [Privacy](https://fortyquestions.io/privacy)
- [Terms](https://fortyquestions.io/terms)

## Contact

Satayu Isariyaphorn, founder — satayu@fortyquestions.io