Self-reported attribution:
ask buyers if AI found you
Self-reported attribution asks buyers how they found you. For AI search it catches the readers who never clicked, which no analytics tool sees. Add ChatGPT, Gemini, Perplexity, Claude and Google AI as answer options, keep a free text field, and code the answers into your CRM as AI-influenced.
of B2B buyers used generative AI in a recent purchase
Gartner, 645 buyers, Aug to Sep 2025of Google visits with an AI summary ended on a click inside the summary
Pew Research Center, Mar 2025of visits ended the browsing session with vs without an AI summary
Pew Research Center, Mar 2025of buyers said generative AI helped them decide better
Forrester, 2026Customers name AI far more often than analytics credits it
Sample share of signups by source: what customers say, and what analytics records.
Sample data for illustration. Your own numbers will differ.
How to read it. The AI bar is the gap. Customers say AI helped them, but their visit was filed as direct or search. Both views are useful.
In ThriveStack citedby The survey answer is stored next to the tracked source on each contact, so the gap is measured, not guessed. See it for your site →
The share of signups who name AI keeps rising, month by month
Sample share of new signups choosing an AI assistant on the how-did-you-hear question.
Sample data for illustration. Your own numbers will differ.
How to read it. A rising line is your early signal, even when analytics barely moves.
In ThriveStack citedby AI-influenced deals are reported on their own line next to AI-sourced, so the survey signal reaches the revenue report. See it for your site →
A how-did-you-hear question catches the AI buyers analytics cannot see.
Many buyers read an AI answer, skip the link and return later by search or direct visit. They leave no trace in GA4. They will tell you if you ask. The survey is the only direct view of influence.
of visits with an AI summary included a click inside it
Pew Research Center, 900 U.S. adults, Mar 2025ended the browsing session
Pew Research Center, 900 U.S. adults, Mar 2025Use a short list of options plus a free text field.
| Option | Why it is there |
|---|---|
| Google search | Baseline for comparison |
| Largest AI referral source | |
| Gemini or Google AI answer | Covers Google AI surfaces |
| Separate source with its own referrer | |
| Separate source | |
| Another AI assistant | Catches Copilot, Grok and others |
| Other, please tell us | Free text finds new sources |
Code the answers into your CRM and let sales correct them.
Store the answer in its own field. Map any AI option to AI-influenced. After a sales call, let the rep fix it when the buyer says more. Do not overwrite the tracked source, so you can compare the two.
Know the limits of survey data.
People forget, and many skip the question. Read it as a floor and a direction. Compare it with tracked AI-sourced revenue each month.
- People forget
- Many skip the question
- Wording changes results
- As a floor and a direction
- Next to tracked AI-sourced revenue
- Month on month, same wording
Self-reported answers find the buyers who read an AI answer and never click.
Pew found that only 1% of visits with an AI summary ended on a click inside the summary, and 26% ended the session. Gartner found 45% of buyers used generative AI. Put those together and a large share of AI influence leaves no click.
A survey option is the cheapest way to count them. Add "ChatGPT or another AI assistant" to your signup form and your demo request form this week.
of buyers used generative AI
Gartner, 645 B2B buyers, Aug to Sep 2025of AI summary visits ended on a click inside the summary
Pew Research Center, 900 U.S. adults, Mar 2025Ask the question the same way every time, or you cannot compare months.
Keep the wording, the options and the position on the form fixed for at least a quarter. Randomize the order of the options to avoid favoring the first one, and keep an "Other" field.
Review the free text. Buyers often write the product name ("Claude", "Perplexity") instead of the category. Code those to one AI value and keep the raw text.
Keep the options and their position fixed, randomise the order of the options, keep an "Other" field and read the free text. Buyers often write the product name.
“For 2026, the question isn't how to grow AI referral traffic; it's how to grow brand visibility inside AI experiences.”
ThriveStack citedby puts self-reported source next to tracked visits and signups, so you can see where the two agree and where they differ.
Self-reported attribution, answered.
Where do I put the how-did-you-hear question?
On signup, on booking forms and in the first sales call.
Does self-reported attribution replace analytics?
No. It adds the buyers analytics cannot see.
How do I report it?
As AI-influenced revenue, labelled as an estimate.
Sources for this lesson
- Pew Research Center: Google users are less likely to click on links when an AI summary appears (Jul 22, 2025)
- Gartner survey via Business Wire (May 20, 2026) 645 buyers
- Digital Commerce 360: Forrester on B2B buying groups and AI (Jan 22, 2026)
- Conductor: 2026 AEO / GEO Benchmarks Report Updated Jul 6, 2026
Figures were checked against each page in Nov 2026. Vendor studies are marked by their sample. Charts marked sample data are illustrations.
Questions this lesson answers
See all 10 questions in this cluster
- How did you hear about us ChatGPTP0 NEW
- Self reported attribution AI searchP0 NEW
- Add ChatGPT to how did you hear about usP0 NEW
- Survey question for AI discoveryP0 NEW
- Self reported attribution vs software attributionP0 NEW
- Sales call question to find AI influenceP0 NEW
- AI influenced pipeline from self reported sourceP0 NEW
- How to code self reported attribution in CRMP0 NEW
- Free text attribution field AI mentionsP0 NEW
- Post purchase survey AI assistantP0 NEW