We discover and update the factors answer engines use, so you can manage what matters about your visibility.

Built for B2B SaaS marketing teams with 50–500 published pages. Currently in beta; founding-customer programme open through customer #10. Vertical expansion considered from 2026-08-11 onward, after the first five founding customers ship.

Answer engines choose what to cite. The factors change constantly. AnswerGraph tracks them across ChatGPT, Perplexity, Google AI Mode, and Claude — and tells you what to do about it.

Grow your citation share

See why answer engines cite your competitors instead of you. Get a weekly action plan: what to change, predicted lift, effort estimate. Monitor whether each action moved the needle.

Protect your visibility

See what answer engines are saying about your brand right now. Get alerted when something changes — a new competitor cited, a factual error surfacing, your share dropping. Know what to do about it and whether the fix worked.

Every recommendation ships with a confidence interval, a sample size, and a registered hypothesis. If we can't show those three, we don't ship it.

The methodology page versions every claim we publish. Read it.

Get notified when we open to external customers

We're currently measuring on internal properties. We'll be in touch when measurement is mature enough to bring on external customers — expected within 8–12 weeks pending data accumulation.

Register interest
486
Panel observations per day across 4 engines and 3 verticals.
Source: AnswerGraph panel (live since 2026-05-03)
9,115
Training rows in the inference engine after first crawler expansion.
Source: AnswerGraph changelog (2026-05-04)
0.34
Median confidence interval width after coverage expansion (69% reduction from day one).
Source: AnswerGraph changelog (2026-05-04)

Internal test properties

amvia.co.uk

B2B telecoms

Measurement in progress — baseline established May 2026

comparefibre.co.uk

B2C/SMB fibre comparison

Measurement in progress — baseline established May 2026

surfaceloop.com

EASM cybersecurity

Measurement in progress — baseline established May 2026

How the engine works

Continuous panel measurement. Statistical inference via mixed-effects logistic regression. Drift detection. Counterfactual attribution. Pre-registered hypotheses tested and published regardless of outcome.

The methodology is public so you can verify the engine works — not because transparency is the product.

Read the methodology