AI visibility ROI: a practical model from share of answers to pipeline.

The short answer.

To estimate the return on AI visibility, multiply five things: how many buyer questions are asked in your category, how much more often answers name you, how many of those buyers then visit or look you up, how many visitors enquire, and what an enquiry is worth in pipeline. Compare the result with what the work costs.

The published data gives reference points for some steps and none for others. AI referrals are around 1% of website traffic on average, they usually convert better than organic visits, and a growing share of B2B buyers start research in a chatbot. The rest, especially how many buyers act on an answer without clicking, has to come from your own tracking. That is why the model only becomes reliable once conversion tracking connects AI-referred and self-reported enquiries to your CRM.

The calculator in chapter 9 does the arithmetic. It does not ask for a price: it tells you the most the work could cost each month and still pay for itself, which you can then compare with any quote.

Why AI visibility ROI is hard to show.

Three things break the usual SEO return calculation: the buyer often does not click, the visits that do happen are misfiled, and nobody reports how many people asked the question.

First, the click. Pew Research Center tracked 900 US adults’ Google searches in March 2025 and found that people clicked a link inside an AI summary on just 1% of visits to pages that showed one (published 22 July 2025). A buyer can read your name in an answer, remember it, and arrive weeks later through a branded search or by typing your address.

Second, attribution. Visits from AI apps often arrive without a referrer and land in “direct” in GA4. Google’s own generative AI report in Search Console, which reached all sites worldwide on 31 August 2026, shows impressions in AI Overviews and AI Mode but not clicks separately; Google notes the data is also inside the main performance report.

Third, demand. There is no public “search volume” for ChatGPT questions. And marketers know it: in a release on 26 June 2026, Semrush reported that 45% of marketing leaders cannot accurately measure their brand’s visibility in AI answers, and only 9% have the tools to track all the relevant metrics.

None of this makes ROI unknowable. It means the model has to be honest about which numbers are measured and which are assumed, and has to improve as tracking improves.

The model in five steps.

Each step turns one number into the next. Measure what you can, assume the rest openly, and replace assumptions with data each quarter.

Step one: buyer questions and share of answers.

Start with how many relevant questions buyers put to AI tools each month, then how often answers name you now and how often you aim for them to. The first is an estimate; the second you can measure.

The audience is large and increasingly asks for information. OpenAI’s chief executive told its developer conference on 6 October 2025 that more than 800 million people use ChatGPT every week, as reported by TechCrunch. A National Bureau of Economic Research working paper on how people use ChatGPT (September 2025) estimated that about 49% of messages are “Asking”, seeking information or advice, and that “Seeking Information” grew from 14% to 24% of usage between July 2024 and July 2025.

B2B buyers in particular are moving. G2’s research published on 15 April 2026, surveying 1,076 software buyers in March 2026, found 51% now start research with an AI chatbot more often than with Google, up from 29% in 2025, and 71% use AI chatbots at some point. Forrester wrote on 22 January 2026 that 94% of B2B buyers use AI in their buying process.

How to estimate your monthly questions

There is no ChatGPT keyword tool you can trust for volume, so build a range. Take Google search demand for your category’s commercial questions, apply the share of buyers in your market who also ask AI (G2’s 51% is a reference for software buyers; services buyers may differ), and use a low and a high case. Some AI visibility tools publish modelled prompt volumes; treat them as another estimate, not a count.

Share of answers is the percentage of panel runs in which an engine names you. Measure the baseline before any work starts; our guide on checking AI search visibility sets out the method.

Step two: from answer to visit.

Only a small share of buyers who see your name in an answer click straight through; more look you up later. Model both, and keep them separate.

For the direct click, Pew’s 1% figure for links inside Google’s AI summaries is the only public behavioural number we found. It applies to Google, not to chatbots, where a named supplier with a link may draw more clicks. For the size of what arrives, Conductor’s benchmarks report, covering 1,215 enterprise domains in 10 US industries from May to September 2025, found AI referrals made up 1.08% of all website traffic, with IT the highest at 2.80%, and ChatGPT supplying 87.4% of AI referrals.

The indirect path is the bigger unknown. A buyer who reads your name and later searches for it shows up as branded search or direct traffic. Track branded search impressions in Search Console month by month, and ask every enquirer how they first heard of you. Keep self-reported answers in their own column, never merged with tracked sources; our play on the “how did you hear” question shows how to word it.

On Google’s own results, being named also changes clicks. Seer Interactive found brands cited in an AI Overview earned 35% more organic clicks and 91% more paid clicks than when not cited (4 November 2025), noting that this is a correlation, not proven cause.

Step three: from visit to enquiry.

Visitors from AI answers usually convert better than organic visitors, but the published multiples range from about 1.2 to 23 times. Use your own GA4 data as soon as you have it; until then, use the low end.

• Ruler Analytics (26 May 2026): AI referrals convert at 5.8% on average across industries, against 4.9% for organic search. Ruler counts calls as well as forms.
• Microsoft Clarity (6 November 2025): on 1,277 mostly publisher sites, sign-up rate was 1.66% from AI tools against 0.15% from search.
• Semrush (21 July 2025): the average AI search visitor is worth 4.4 times an organic visitor, based on conversion rate; the rates were not published.
• Ahrefs (16 June 2025): on its own site, AI search was 0.5% of traffic but 12.1% of sign-ups, 23 times the organic rate. One site, and Ahrefs says it may be unusual.

For a B2B baseline, First Page Sage reports visitor-to-conversion rates of 1.1% for B2B SaaS and 1.5% for IT and managed services (18 September 2025), and Ruler reports 2.3% for organic visits to B2B services sites in 2025. A cautious default is your organic enquiry rate with a small uplift. Before relying on your own rate, check whether GA4 conversions match your real leads.

Step four: from enquiry to pipeline.

The last steps come from your CRM, not from any benchmark: how many enquiries qualify, how many become opportunities that close, and what a won deal is worth. We walk through one anonymised client result measured this way in a separate article. If you also run ads, offline conversion tracking sends those same CRM stages back to the ad platforms.

Use the same definitions you use for every other channel. If paid search enquiries are judged on cost per qualified opportunity, judge AI-sourced ones the same way; our article on cost per lead versus cost per qualified opportunity explains why. Use first-year gross profit, not revenue, so the return is comparable with the cost of the work.

Two adjustments are worth making openly:

• Unseen influence. Some buyers who saw your name in an answer will enquire through another channel and never mention it. Add a multiplier only if your self-reported data supports it, and show the result with and without it.
• Time lag. B2B deals close months after first contact. Model a 12-month window and report pipeline created before revenue won.

Connecting AI-referred enquiries to closed deals is covered step by step in our AI-to-pipeline play.

A worked example.

A fictional B2B services firm, with every number invented to show the arithmetic. The point is the shape of the calculation, not the result.

At steady state this firm could spend up to about 3,600 a month on the work and break even on first-year profit, in its own currency. Two lessons follow. The model is most sensitive to deal value and share of answers, so AI visibility pays most for firms with high-value deals and few, well-defined buyer questions. And the result is a range: halve the look-up rate and drop the influence multiplier, and the break-even figure falls by about 70%.

The ROI calculator.

Set each assumption yourself. Defaults are the illustrative values from chapter 8, not benchmarks for your business. The calculator runs in your browser and nothing is sent anywhere.

Read the two figures as a range. If the cautious case already covers a proposal, the decision is easy; if only the optimistic one does, the work needs a measurement plan that proves the assumptions within two quarters.

Where this fits with cost and reporting.

This model answers one question: could AI visibility work pay for itself? Two companion articles cover the questions either side of it.

Use them in that order. The cost guide gives you a range to test; this model gives you a break-even line; the board scorecard tracks the inputs that this model assumed, so each quarter you can replace a guess with a measurement. When the measured share of answers rises but enquiries do not, the problem is usually further down the funnel; our article on why being cited is not being chosen covers that gap.

Assumptions to agree first.

Before any AI visibility work starts, agree these with whoever signs off the budget, so the return is judged on terms set in advance.

• The list of 20–50 buyer questions, and which engines and countries count.
• The baseline share of answers, measured over several runs before work begins.
• How AI referrals are identified in GA4, and that app traffic in “direct” is not counted.
• The exact wording of the self-reported source question on every form.
• The CRM fields for original source and self-reported source, kept separate.
• The qualified-enquiry definition, matching other channels. Our guide to website lead tracking shows how to record it.
• The time window: pipeline at six months, revenue at twelve.
• The cautious and optimistic cases, both written down.

Questions owners ask.

How do you calculate the ROI of AI visibility?

Estimate the extra answers that name you after the work, multiply by the share of buyers who click or look you up, their enquiry rate, the qualified and win rates, and the gross profit per deal. Compare that monthly figure with the cost of the work, using a cautious and an optimistic case.

What is a good AI referral conversion rate?

Published averages vary. Ruler Analytics reported AI referrals converting at 5.8% against 4.9% for organic in May 2026; other studies report far larger gaps on narrower samples. Use your own organic enquiry rate with a modest uplift until your GA4 data gives you a real figure.

How much traffic does AI search send to websites?

Not much yet. Conductor measured AI referrals at 1.08% of website traffic across 10 US industries in 2025, with IT the highest at 2.80%. Much of AI search’s effect happens without a click, which is why the model includes buyers who look you up later.

How long before AI visibility work shows a return?

Share of answers can move within weeks of changes to pages and profiles, but B2B pipeline takes longer. Plan to report share of answers monthly, enquiries quarterly, and revenue over twelve months, and agree that timetable before work starts.

Can I measure AI visibility ROI without paid tools?

Yes. A manual panel of buyer questions, Search Console’s AI impressions report, a GA4 channel group for AI referrers, a source question on your forms and your CRM cover every input in the model. Tools save time at higher volumes. The GA4, form and CRM parts are what our conversion tracking setup covers.

Sources, all read 10 November 2026: Pew Research Center, 22 July 2025; Google Search Central blog, 3 June 2026; Semrush, 26 June 2026 and 21 July 2025; TechCrunch on OpenAI DevDay, 6 October 2025; NBER working paper w34255, September 2025; G2, 15 April 2026; Forrester, 22 January 2026; Conductor AEO/GEO benchmarks report; Seer Interactive, 4 November 2025; Ruler Analytics, 26 May 2026; Microsoft Clarity, 6 November 2025; Ahrefs, 16 June 2025; First Page Sage, 18 September 2025. Search demand for “ai visibility”: Google search data via DataForSEO, US and UK, November 2026. Worked example and calculator defaults are illustrative.