How to Track AI Conversion Rate in GA4: Step by Step

Every brand I audit asks the same question the moment AI traffic shows up in their reports: what is it actually worth. GA4 will not tell you out of the box. AI referrals get dumped into Direct or mislabeled Referral, your AI Search channel reads as zero sessions, and the conversion rate you most want to see does not exist. From what I have seen, that single gap is why most teams still under-invest in GEO - they cannot prove the channel pays.

This is the step-by-step I ship to fix it: one clean AI Search channel, a conversion report built against it, and the revenue number that ends the argument.

Why GA4 doesn't track AI conversion rate by default

Before the setup, know what you are fixing. GA4 mishandles AI traffic in four specific ways.

• The native AI Assistant channel is partial. Since May 13, 2026, GA4 ships an AI Assistant channel in the Default Channel Group (medium ai-assistant), but it only recognizes ChatGPT, Gemini and Claude - and it is not retroactive.
• Perplexity and Copilot fall through. They land in Referral or Direct, so half your AI Search channel is scattered across two other buckets.
• No-referrer hits become Direct. Answers rendered inside the app, or clicks that strip the referrer, arrive with no source. GA4 calls that Direct, the same bucket as someone typing your URL.
• Default reports blend it all. Even when the source is correct, the standard Acquisition report has no row that isolates AI conversion rate against organic.

The fix is three setup steps, then a report. Do them in order.

Set up an AI Search channel in GA4 (steps 1-4)

Everything in this section happens in Admin, not Reports. You build and reorder channels in Admin (the gear icon, bottom-left); Reports only shows the result once the channel exists. So if you are staring at a report wondering where to click, that is why - switch to Admin first.

1. Step 1 - Turn on GA4's native AI Assistant channel. In Admin, open Data display, then Channel groups, and open the Default Channel Group. Confirm it lists AI Assistant. It needs zero setup and stamps medium ai-assistant for ChatGPT, Gemini and Claude going forward (Search Engine Journal, 2026). Treat it as the floor, not the finish line - it misses Perplexity and Copilot and it does not backfill history. You cannot edit the Default group, which is exactly why the next steps build your own.
2. Step 2 - Create your own AI Search channel group. Still in Admin, Data display, Channel groups, click Create new channel group (top right). Name it "AI Search". GA4 copies in every default channel - Direct, Organic Search, Email and the rest - so you are editing a full list, not a blank page. Hold that thought; it is why step 4 exists. A custom channel group is also retroactive: it re-buckets your historical data, so you get a real before and after. The free GA4 tier gives you two custom channel groups of up to 25 channels each (Julius Fedorovicius, Analytics Mania), and both he and Himanshu Sharma of OptimizeSmart walk this same route for AI traffic.
3. Step 3 - Add the AI Search channel and paste the matching rule. Click Add new channel. Name the channel "AI Search". Under conditions, pick Source, then matches regex, and paste the pattern below - it catches the four major engines plus Claude and the OpenAI variants, so one channel covers the whole AI Search surface. Click Done to save the channel. It lands at the bottom of the channel list.

   chatgpt\.com|chat\.openai\.com|openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|bing\.com/chat

4. Step 4 - Drag AI Search above Direct. This is the step most guides skip and the one that quietly breaks the report. Your new channel sits at the bottom of the list, below Direct. GA4 reads the list top to bottom and files each visit into the first channel it matches, then stops - so if Direct is above AI Search, a no-referrer AI visit lands in Direct and never reaches your rule. To fix it: click Reorder (above the channel list), then grab the drag handle - the stacked-dots icon at the far left of the AI Search row - and drag the row up so it sits above Direct. Click Apply, then Save group, top right. Himanshu Sharma's rule of thumb: specific channels like AI Search high, broad catch-alls like Direct low (OptimizeSmart).

Build the AI conversion rate report in GA4 explorations (steps 5-6)

The channel group sorts the traffic. Now you need a report that actually shows AI conversion rate next to organic, because no default report does. This part is in Reports, under Explore.

5. Step 5 - Create a Free-Form exploration for AI traffic. Open Explore, start a blank Free-Form. Pull in the dimension Session default channel group and the metrics Sessions, Sessions key event rate (this is your conversion rate), Key events, and Total revenue. Set the row to the channel group dimension and you have every channel's conversion rate side by side.
6. Step 6 - Compare AI conversion rate against organic search. Add a filter to keep only AI Search and Organic Search, or just read those two rows. The number that matters is the ratio: AI conversion rate divided by organic conversion rate. If AI is below organic, your post-click experience for AI visitors is broken - that is a separate fix, and it is usually the landing page, not the channel.

Track AI referral revenue, not just conversion rate

Conversion rate wins the analyst. Revenue wins the budget. Add Total revenue and Average purchase revenue to the same exploration and you get the number that ends the GEO argument: dollars from the AI Search channel.

One caveat to bake in, or you will overstate the case. ChatGPT traffic tends to spend less per order even when it converts more often - lower AOV against a higher rate. So report both. A channel that converts at 1.8x but spends at 0.8x AOV is still winning, just not by the headline multiple. The honest number is revenue per session, not conversion rate alone. For the full picture on rates, AOV and the growth curve, see the complete AI conversion rate benchmarks.

The AI attribution gap GA4 still can't close

Even with the channel and the report clean, GA4 undercounts AI. Two leaks you cannot fully patch:

• No-referrer dark traffic. Answers read inside the ChatGPT or Perplexity app often send no referrer. GA4 rewrites missing or excluded referrers to Direct - the same mechanic Simo Ahava documents for referral exclusions - so those conversions silently land in Direct and your AI number is a floor, not a ceiling.
• Last-click blindness. AI frequently does the convincing, then the buyer returns later via Direct or branded search to purchase. GA4's last-click model hands the credit to the final touch and AI gets none.

The practical move: read the AI Search channel as a directional floor, watch branded-search and direct lift in the weeks after AI citations climb, and never argue the channel down to a single GA4 row. AI conversion rate in GA4 is the evidence you can show, not the whole truth.

AI conversion rate benchmarks to compare your GA4 numbers against

Once your report runs, you need something to read it against. These are the public benchmarks I check client numbers into.

Do not treat these as targets - treat them as the shape of the curve. Amsive's 54-site test found no significant gap on some accounts (p=0.794), so your own GA4 ratio beats any benchmark. The table tells you whether your number is plausible; your report tells you what is true. For why the gap shows up at all, see the four mechanisms behind it, and for the metrics that live beyond GA4, see the e-commerce AI search KPIs.

Build the channel once, and AI conversion rate stops being a guess. You get a number you can put in front of a CFO, a ratio you can defend, and a clean read on the highest-converting traffic most teams still cannot see.