Reporting AI-Referred Conversions to Finance Without Overclaiming

Sooner or later someone from finance asks what AI search is worth. The question is fair, and the temptation is to answer with the biggest number you can defend. Resist it. Finance teams remember the first inflated figure for years, and your credibility on every later number depends on how that first one holds up.

The goal is a number that is small enough to survive scrutiny and clearly labeled for what it includes and leaves out.

What you can count

Start with what is directly observable.

AI-referred visits. Sessions where the referrer identifies an AI platform, such as ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Kimi or Mistral. Promptwatch's visitor analytics reports these using a lightweight script, a Google Tag Manager template or the WordPress plugin.

AI-referred conversions. The subset of those visits that completed a defined goal: signup, demo request, purchase. Promptwatch reports conversions alongside AI-referred traffic.

Pages receiving the traffic. Which landing pages draw AI visitors, so you can see what content earns clicks and not only citations.

That is your floor: directly measured, with a visible referrer.

What you cannot see cleanly

Be upfront about the missing parts.

Visits without a referrer. Some AI-driven traffic arrives with no identifiable source and lands in direct. You may suspect part of your direct traffic is AI-influenced, but you cannot prove it per session.

Influence without a click. A user sees your brand in an answer, remembers it, and later searches for it by name. That shows up as branded search or direct traffic, not as an AI referral.

Answers that never produce a visit. Many AI answers satisfy the user. You were mentioned, which may matter, and there is no session to count.

Present these as limits, not as hidden upside. If you add an estimate for them, label it separately and show the method.

A report structure finance can accept

Use three tiers, kept visually apart.

Tier 1: measured. AI-referred visits, conversions and the conversion rate, with the date range and the tools used.

Tier 2: correlated. Movements in branded search or direct traffic alongside visibility changes. Show both lines. Make no causal claim.

Tier 3: unmeasured. A plain statement that some influence is not captured, and why.

Only Tier 1 goes into any revenue number. If someone wants to combine tiers, make them do it in front of you, with their assumptions written down.

Use cases, not hype, for comparisons

If you report a conversion rate by channel, check sample sizes. A rate computed from a small number of conversions swings widely. Say the count along with the rate, and avoid declaring one channel better than another on thin data.

Promptwatch's own published case studies give examples of how others have reported. Crisp reported that AI-referred traffic converted at twice the rate of traditional channels, and TreeHouse, which runs Pararius and Huurwoningen.nl, reported AI search reaching about 25% of organic conversions, almost all of it from ChatGPT. These are those customers' results, in their contexts, and they are not benchmarks for your site. Use them as proof that the measurement is possible, not as targets.

Tie traffic back to visibility

Finance will also ask what you did to earn it. Connect the measured visits to the work behind them.

• Which pages receive AI visitors, and when were they last updated?
• What was their citation trend in Citation Trends?
• Did crawler logs show the bots fetching them without errors?

The 360 Insights view joins crawler logs, citation clicks and AI-referred visits per page, which puts those links in one table. A page that is fetched, cited, clicked and converted is a story a finance person can follow.

Setting up the measurement

Do the unglamorous setup first.

1. Install the visitor analytics script, GTM template or WordPress plugin.
2. Define conversion events with the people who own revenue reporting. Agree on what counts before measuring.
3. Exclude internal traffic.
4. Check that referrers from AI platforms are classified, and review unclassified sources monthly.
5. Record the date tracking began. Nothing before it counts.

If you also use Google Analytics, check that its channel definitions and the platform's classification agree, and note differences instead of forcing a match. For help with GA4 itself, see our separate walkthrough on measuring chatbot traffic there.

Plan notes

Visitor analytics is available on paid plans, with visitor event allowances that scale: Essential includes 200K events, Professional 1M and Business 10M. Agency plans start at 500K on Kick-off. Check your traffic against the allowance, since high-traffic sites can use it up.

What to promise

Do not forecast revenue from AI search on the basis of a few months of data. Say what was measured, what changed, and what you plan to test next. If finance wants a target, set one for something you control, such as the number of pages with a healthy citation trend or the share of visits tracked correctly, rather than a revenue figure you cannot promise.

The recommendation

Report the measured floor, label everything else, and tie the numbers to specific pages. Promptwatch provides visitor analytics with conversions, plus the page-level citation and crawler data that explains where the visits came from, so the report and the story come from the same system. Set up the script this week so you have a clean start date, and look at the first full month before you present anything. Start at promptwatch.com.