Measuring Whether AI Search Traffic Is Actually Turning Into Enquiries

20 September 2026·5 min read

Quick answer: You can't tell whether ChatGPT, Perplexity or Google's AI Overviews are actually helping your business just by watching your traffic numbers climb. Most AI referral traffic lands in your analytics tagged as "direct" or gets buried in a catch-all bucket, which hides the real story. The only way to know if it's working is to track where actual enquiries come from, not just where clicks come from, and to give it a few months before drawing conclusions. Once you set up proper tracking, you can start telling the difference between genuine interest and noise. 📈

Everyone's suddenly obsessed with "showing up in ChatGPT" — and fair enough, it's the new frontier. But we're noticing a gap between businesses chasing AI search visibility and businesses that actually know whether it's putting money in the bank. 💖 Visibility feels good. It's satisfying to see your business mentioned when someone asks an AI assistant a question. But visibility isn't the goal — enquiries are. If you can't connect the two, you're making decisions based on vibes instead of numbers.

What most businesses get wrong

The most common mistake is treating "AI search traffic went up" as a win in itself. A session isn't a sale, and a mention isn't a customer. Here's where the tracking usually falls apart:

Standard analytics setups lump AI referral traffic in with "direct" traffic, so it looks like nobody's coming from anywhere, when really a chunk of it is coming from an AI assistant that didn't pass referral data properly.
Businesses check "are we mentioned" without ever checking "did that mention lead anywhere" — the two get treated as the same question, and they're not.
Any bump in overall traffic gets credited to AI search, when it might just be seasonality, a promotion, or an unrelated backlink doing the work.

The AI Referral Tracking Sheet

Build this as a simple spreadsheet, or a GA4 exploration report if you want more of it automated. Add a row every week, or whenever you review your numbers:

Date — the week you're reviewing
Source — chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, or "AI Overview click" if you can identify it
Landing page — which page the session hit first
Sessions — pulled from GA4 for that source
Enquiry submitted? (Y/N) — cross-checked against your form or CRM
Enquiry type — quote request, phone call, booking, newsletter signup
Days to enquiry — same session, or did they come back later
Notes — anything odd, like a spike tied to one piece of content

Add a "How did you hear about us?" field to every enquiry form with an option for "asked an AI assistant" — this single field does more work than any dashboard, because it catches enquiries that never show clean referral data at all.

A B2B firm with a long sales cycle: Someone might land on the site from an AI assistant in March and not enquire until June. Session-based tracking alone will never connect those dots. Once this business added "how did you hear about us" to every enquiry, they found a handful of leads each quarter that traced back to an AI conversation weeks earlier — small in number, but high-intent, since the person had already done their research before they reached the form.

A local service business with a seasonal dip: AI search traffic spiked during a typically quiet month, and it looked like a breakthrough — until bookings didn't move at all. It turned out the AI Overview was answering the question directly on the results page, so people got what they needed without clicking through. Good for brand exposure, not a lead source. That's a useful thing to know, even if it's not the exciting answer.

A quote-based service business with a high average sale: Adding one dropdown option to the quote form ("I asked ChatGPT or another AI tool for a recommendation") showed roughly one in eight new quotes mentioning it within a few months. Not huge, but growing every quarter — and worth knowing before deciding how much time to spend optimising for it.

How to actually do this

Check your GA4 referral exclusion list. Some AI tools' domains get accidentally excluded or misread, which silently erases the data you're trying to collect.
Create a GA4 exploration or segment that isolates sessions from known AI referrer domains, so you're not digging through "direct" traffic guessing.
Add UTM tags to content you know gets pulled into AI answers or cited elsewhere, so you can trace it back cleanly where possible.
Add the "how did you hear about us" field to every enquiry form, booking form and quote request — this catches what your analytics can't.
Review monthly, not weekly. The volumes are usually still small enough that weekly numbers will just be noise.

💡 Give it three months minimum before you conclude anything. AI search behaviour is still shifting quickly, and small sample sizes will lead you to overreact to a single good, or bad, week. Treat the early data as directional, not definitive.

Mistakes to avoid

Celebrating a traffic bump without checking whether any of it turned into an enquiry.
Relying on GA4's default channel groupings, which weren't built with AI referrers in mind.
Forgetting to ask new enquiries how they found you — it's the cheapest data point you'll ever collect.
Making big strategy or budget decisions off a few weeks of data.
Ignoring zero-click AI answers entirely — they may still be building awareness even without a click.

Frequently asked questions

How do I know if traffic is coming from AI search at all?

Check your GA4 referral traffic report for domains like chatgpt.com, perplexity.ai or copilot.microsoft.com. Some sessions will show as "direct" because the AI tool didn't pass referral data — that's a known limitation, not something you can fully fix on your end.

Is it worth setting all this up for a small business?

If AI search is genuinely part of how your customers search for what you do, yes — but be honest about the scale. For most small and service businesses, AI referral volume is still a small slice of overall traffic. Set it up because it's good practice and the trend is only growing, not because it'll transform your numbers overnight.

Can I just trust what the AI tools themselves report?

Not fully. Reporting from AI platforms is inconsistent and still evolving, and none of them currently give you enquiry-level data. Your own analytics and enquiry form are the more reliable source of truth, even if they take more manual effort to maintain.

What if I don't have the time or tech skills to set this up properly?

That's genuinely common, and worth naming honestly: GA4 exploration reports and referral tracking aren't intuitive, and getting them wrong will give you false confidence either way. If you don't have someone comfortable in analytics, it's worth getting a hand to set it up properly once rather than guessing.

Keep reading 🤍

Written by

Kate, founder of Chronically Online

I help Gold Coast and Brisbane businesses grow with branding, websites and marketing that actually works.

Work with me ✦