Software that measures ROI from AI search traffic: our client setup

The question we get after the first quarter of any AI search program is always the same: is this making us money? Mention counts do not answer that. ROI from AI search traffic needs software that sees the visits, ties them to conversions, and links them back to what we did.

Why AI traffic is hard to measure

Several things get in the way. Chat engines do not always pass a clean referrer. Some visits arrive as direct traffic. Plenty of answers are read without any click. And standard analytics reports lump AI sources into "other" or "referral" unless you set up segments.

So we begin with attribution, before any optimization, because history cannot be reconstructed afterward.

The four numbers we report

1. AI-referred visits, split by engine.
2. Landing pages, so we can link traffic to content work.
3. Conversion rate compared with other channels, using the client's own definition of a conversion.
4. Cost, which includes tool fees, content time and engineering time.

Upstream, we also report which prompts and citations feed those visits. That is how a client decides whether to double down on a topic.

What we use

We run client programs on Promptwatch for this reason. Its visitor analytics report AI-referred traffic and conversions across ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Kimi, Mistral and others, set up with a lightweight script, a Google Tag Manager template or the WordPress plugin.

Where it earns its place is the join. 360 Insights combines crawler logs, citation clicks and AI-referred visits per page. So we can show a client that a page was crawled by GPTBot, cited for a prompt, then visited and converted. That is a story a finance lead will accept, because each step is observable.

Two customer cases are worth knowing about, with the caveat that they are those customers' results and not a promise. Crisp reported that AI-referred traffic converted at twice the rate of traditional channels. TreeHouse, behind Pararius and Huurwoningen.nl, reported AI search reaching about 25% of organic conversions, 99% of them from ChatGPT, after fixing crawler access and automating weekly reporting.

Volume matters on the plan. Essential at $95 a month includes 200K visitor events, Professional at $245 includes 1M, and Business at $579 includes 10M. Agency plans from $199 a month include visitor analytics across unlimited projects.

Other approaches

You can build AI segments in GA4 by hand. That works for a rough count but cannot tell you which prompt or citation produced a visit. AthenaHQ lists GA4, Shopify and Webflow revenue attribution next to a GEO score. Amplitude includes an AI visibility feature in every plan. Scrunch AI and Profound report bot and agent traffic. All are useful in context. We need the full chain from prompt to conversion in one place.

Setting expectations on ROI

AI-referred volumes are often small at first, and early conversion rates swing on little data. We show confidence honestly and trend over months, not weeks. We also avoid claiming that AI search caused a sale when it assisted one.

A practical start

1. Install attribution on day one.
2. Define conversions with the client.
3. Track a focused set of buyer prompts tied to landing pages.
4. Fix crawler access early so pages can be cited at all.
5. Report AI-referred conversion rate against other channels each month, with costs.

No software can promise AI traffic. What good software does is tell you, with evidence, whether it is paying back.

One more thing on attribution

If a client has a long sales cycle, AI-referred visits may convert weeks later and show up under another channel. We ask sales to add "how did you hear about us" to forms and calls, and compare the answers with what the analytics show. The two rarely match exactly, but together they give a fuller picture than either one alone.