How to Track AI Search Leads When GA4 Can’t

I was staring at a client’s GA4 dashboard and the numbers weren’t adding up. The intake team said they signed three cases from people who found the firm through ChatGPT. Analytics showed zero visits from ChatGPT. Zero.

So I’m looking at this client’s GA4 dashboard maybe six weeks ago and the numbers aren’t adding up and I keep thinking there’s something wrong with the tracking because the intake team is telling me they signed three cases from people who said they found the firm through ChatGPT but the analytics show zero visits from ChatGPT and I’m sitting there going through the referral sources one by one and it’s all just “Direct” and “Referral” with no detail and I realize the problem isn’t the tracking exactly; the problem is that GA4 was built before AI search existed and nobody at Google has fixed the default channel groupings to account for it, which you’d think would be a priority but apparently not.

And this isn’t a small gap we’re talking about. A 2026 study from Conductor looking at 3.3 billion sessions found that AI platforms drive roughly 1.08% of total web traffic which sounds tiny until you look at what that traffic actually does once it lands on a page and I’ve been writing about why AI search converts 6x better than Google separately but the short version is that a Seer Interactive study from October 2024 through April 2025 found ChatGPT traffic converting at 15.9% compared to Google organic at 1.76% and Perplexity at 10.5% and I’ve seen similar patterns with the firms I work with where the AI traffic is a trickle but the people clicking through are basically pre sold on hiring someone, which changes the math on everything if you know how to actually see it in your reports.

But honestly the analytics gap isn’t usually the problem people come to me about; they come because their managing partner is asking why they’re spending money on AI visibility when the reports show no results and the marketing person can’t prove anything because GA4 is hiding the evidence in a bucket labeled “Other” or “Direct” and the whole conversation turns into a budget fight instead of a strategy conversation, which I’ve watched happen maybe four or five times now with different firms.

And the data from Similarweb’s 2026 Global Report puts ChatGPT conversion rates at 11.4% compared to 5.3% for standard organic and I don’t have an exact number for legal specifically but the pattern holds across every vertical I’ve looked at; AI traffic converts at multiples of what traditional search sends you. A Microsoft Clarity study analyzing over 1,200 publisher sites found that Copilot referrals converted subscriptions at 17x the rate of direct traffic and Perplexity converted at 7x the baseline rate and those aren’t typos, which maybe should tell you something about where these platforms are headed.

So if your highest converting traffic source is invisible in your reports, what decisions are you making based on data that’s lying to you?

Step 1: Go to Google Analytics

Step 2: Click Channel Groups

Step 3: Create new channel group

Step 4: Name channel group

Step 5: Add Regex and save

The rough rule on this is you’ve got maybe a 20 minute setup and then you’re done forever and here’s the actual steps because I know people want the specifics. In GA4 you go to Admin then Data Settings then Channel Groups and you click “Create new channel group” which starts with a copy of the defaults and you add a new custom channel and name it something like “AI Search” or “Gen AI Traffic” and set the condition to Source matches Regex and paste in the string I’ll show you below.

Source: GA4 Custom Channel Grouping documentation — expand this regex as new AI platforms gain market share

And this is the part people miss; you have to use the reorder function to drag this new channel above the standard Referral channel because GA4 processes rules top to bottom and if your AI channel sits below Referral it never gets triggered and all that ChatGPT traffic just keeps disappearing into the generic bucket and you’ll never know it was there, which is basically what’s been happening to every law firm in the country that hasn’t done this yet.

“I keep hearing about AI search but I have no idea if it’s actually sending us anything.” That’s what an intake manager at a PI firm told me maybe three months ago and she was right; she had no way to know because their form had four options which were Google, Friend Referral, TV Ad, and Other and nobody was selecting Other and writing in ChatGPT because people don’t think of AI search as a distinct thing yet, they just think of it as “I searched online” and pick Google even if they spent 20 minutes in a conversation with Perplexity before clicking through to the firm’s website.

A SparkToro study analyzing 332 million queries found that 58.5% of Google searches in the US end with zero clicks and when AI Overviews show up that number jumps to 83% which means the majority of your potential clients are getting answers without ever visiting your site and the only attribution method that works is asking them directly, which feels low tech but it’s honest and it’s accurate and I’ll take that over a fancy dashboard that’s lying to me.

At Percy we treated intake like an emergency room and part of that was tracking every single lead back to its source. The same principle applies here; if you’re not asking where people found you with enough specificity to catch AI referrals then you’re flying blind on what might be your highest quality lead source

The data on this is pretty clear and I still don’t fully understand all the psychology behind it but here’s what I do know. When someone types “best medical malpractice lawyer in Miami” into Google they get a list of 10 options and their brain is in comparison mode scanning for something to click and most of those clicks are exploratory and they’re window shopping. When someone asks ChatGPT the same question they get a conversation that narrows down to one or two recommendations and by the time they click through to your site they’ve already been told you’re the answer and I wrote about why AI leads convert like referrals in more detail separately but the short version is there’s this thing in behavioral economics called effort justification where the 20 minutes they spent refining prompts makes them psychologically committed to whatever the AI suggests.

Source: 2026 AI Overview citation study — cross referenced organic + paid click data

So it’s not just about the traffic coming directly from AI; being cited in AI results creates what I’d describe as a halo effect on everything else you’re doing and it changes how you should think about your PPC and SEO budgets working together instead of treating them as separate line items, but that’s maybe a different conversation.

Once you can actually see AI traffic as its own channel the first thing that usually happens is the conversion rate shocks people because it’s so much higher than everything else and then they start asking whether they should shift budget and honestly the answer depends on your situation. If you’re spending $15,000 a month on Google Ads and your cost per signed case from PPC is $8,000 and then you look at the AI channel and see three signed cases from zero ad spend; that’s not a signal to stop running ads but it is a signal that putting resources into making your firm more visible to AI models is probably the highest ROI work you can do right now.

And the way I think about this for firms I work with is you’re not choosing between Google and AI; you’re making sure the work you do for one helps the other. The effort compounds instead of splitting your focus which is kind of the best case scenario for any marketing investment, or at least that’s how I’ve been approaching it with the firms I work with and so far the tracking from click to signed case seems to support it but I’ve only been measuring this for maybe six months so I don’t want to overstate what I know.

The firms that start tracking AI attribution now are going to have six months of data proving ROI while everyone else is still arguing about whether AI search matters.

P.S. If you’re curious about what your current conversion tracking and attribution setup is actually missing, or if you want to understand the difference between what your agency shows you versus what the data actually says, those are probably worth reading first. And if your firm handles multiple practice areas, the attribution model page walks through how to separate which channels are driving which case types, which gets relevant fast once you start seeing AI traffic as its own source.