Your Firm Is Probably Already Getting AI-Referred Visits — Your Analytics Is Filing Them as Something Else

On August 18, Search Engine Land published the longest first-party look yet at where AI Overview traffic actually goes. Nine months of tracking, 51,200 recorded events across 1,661 cited passages, all captured by one team that got tired of waiting for Google to report this natively. The finding that should stop every managing partner mid-scroll is not about content. It is about counting.

Across that entire dataset, 22.4% of AI Overview traffic was misattributed in Google Analytics — filed under Direct instead of Organic Search. That is 11,468 visits from people who asked Google a question, read an AI-generated answer, clicked a citation, and landed on the site — all recorded as if they had typed the URL from memory. In the worst month, May 2026, the misattribution rate hit 29.3%. In the best, April, it was still 16.8%.

Earlier this month we wrote about writing the sentence AI lifts. This is the other half of the same problem: seeing the visits that sentence is already sending you.

The Signal Is Already in Your Data

Here is the mechanism, and it is simpler than it sounds. When someone clicks a cited passage inside a Google AI Overview, Google often appends a text fragment to the destination URL — the `#:~:text=` string that scrolls the reader to the exact highlighted sentence. Your website receives that fragment. Google Analytics receives it too. It just does nothing with it by default.

Standard GA4 setups have no dimension that separates "arrived via an AI Overview citation" from "arrived via a blue link." So those visits get bucketed into ordinary search traffic when the referrer survives the hop, and into Direct when it does not. Either way, the number you would most want to know — how many people are reaching your firm because an assistant recommended it — is sitting in your property, unlabeled.

That matters more for an immigration practice than for most businesses. The questions your clients ask are exactly the questions AI Overviews are built to answer: what happens if I miss my priority date, can I travel while my adjustment of status is pending, how long do I have to respond to an RFE, does a DACA renewal reset my work authorization. If assistants are answering those questions and citing someone, you want to know whether that someone is you.

Why the Misattribution Is Worse Than It Looks

A 22% gap in channel reporting is not a rounding error. It is a decision-making problem in three directions.

First, it understates organic. AI Overviews only exist inside Google Search. Every one of those Direct-labeled visits is organic traffic. If you have ever looked at a flat organic line and concluded your content work was not paying off, some portion of the payoff may have been sitting in the Direct column the whole time.

Second, it hides which pages are earning the citations. The same study found the traffic is extremely concentrated: across 1,661 cited passages, the average drove 31 visits while the single top performer drove 2,276. That is a seventy-fold spread. For a small firm, this is the difference between refreshing the H-1B page that is actually being quoted and spending a Saturday rewriting a page no assistant has ever touched.

Third, it makes the trend unreadable. Over the full nine months, AI Overviews accounted for 7.53% of organic sessions — but the distribution was nowhere near flat. It peaked at 16–17% in February and March 2026, meaning roughly one in six organic visitors arrived through an AI answer, then fell back to 2–4% by the time the study published. If you cannot see the channel, you cannot see swings like that, and you will keep attributing them to the wrong cause.

Set Up the Measurement This Week

This is a configuration change, not a content project. A non-technical operator or your web person can do it in about twenty minutes.

1. Create a custom dimension in GA4 that captures the page location, including the fragment, on each session's landing page. Name it something you will recognize six months from now — "AI-referred landing" beats "custom_dim_3."
2. Fire it on a condition: the landing URL contains `#:~:text=`. That is your AI-referred flag. Everything else stays untouched.
3. Build one exploration report with that dimension as the row and sessions as the metric. This is your AI-referred traffic view.
4. Add a second view breaking those same sessions down by default channel grouping. The share showing up as Direct is your firm's own misattribution rate — the study's 22.4% is one industry's number, not yours.
5. Record a baseline in week one and log it weekly for a month. Total AI-referred visits, top landing page, top cited passage. Four rows. That is the entire first month.
6. Note the caveat in your log. The `#:~:text=` fragment is also used by featured snippets and People Also Ask, so this is directional, not surgical. It is still far better than the zero you have now.

Then act on it. Whatever page shows up at the top of that list is your highest-leverage asset in AI search — refresh it first, keep its dates and fee figures current, and make sure the passage being quoted still says what you want it to say. Immigration gives you a natural cadence for this: priority dates move monthly, USCIS updates fees and filing addresses, H-1B and TPS windows land on the calendar. Each of those is a legitimate reason to refresh a page that is already earning citations.

The Bigger Picture

For twenty years, the measurement question in search was "where do we rank?" It was a proxy — imperfect, but at least visible. AI search has quietly broken the proxy without replacing it. Google has not exposed AI Overview traffic in Search Console, so the default state for every firm is a channel that is real, growing, volatile, and invisible.

The firms that will make good decisions about AI visibility over the next year are not the ones with the biggest content budgets. They are the ones who can see the number. Twenty minutes of GA4 configuration puts your practice in that group, and gives you a baseline you have never had: how many people are actually walking through your door because an assistant sent them.

The worksheet below walks through the setup step by step, with the exact fields to enter and a four-week baseline log to fill in.

Clientory helps immigration law firms become visible to the families and individuals who need them most — turning AI search into consultations, and consultations into clients.

See Your AI Visibility Score → CLIENTORY.ORG