Three AI Search Developments, One Number Your Firm Can't Name
Three things landed in the last two weeks that look unrelated and aren't.
On August 13, Profound published a year-long classification of 7.5 million ChatGPT conversations, and found commercial intent climbing from 13.9% to 19.2% of all conversations. On August 18, Google released the August 2026 spam update — global, all languages, the third of the year. Also on August 18, Search Engine Land published nine months of first-party AI Overview tracking across 51,200 events.
Demand for AI recommendations is compounding. The pool of pages eligible to be cited just got pruned. And the visits AI already sends your firm are sitting in your analytics under the wrong label.
Each is a headline on its own. Together they describe one problem: your practice cannot price something it cannot see.
Demand is compounding
Profound's sample covers a full year, June 2025 to June 2026. Commercial share of ChatGPT conversations rose from 13.9% to 19.2%, with most of the movement in the final quarter. Extrapolated against OpenAI's reported user growth, that is roughly 243 million commercial conversations a week rising to 533 million — about 28 billion a year. Per-user commercial conversations climbed 41%, from 0.38 to 0.54 a week. This is not a share shifting around inside a fixed pie. People are simply having more of these conversations.
The vertical breakdown is where it gets interesting for you. Legal Services came in at 19.3% commercial and 49.4% informational — one of the most informational categories measured, second only to Healthcare and Education. Automotive, by contrast, is 59.1% commercial.
Read that correctly. It does not mean people aren't hiring attorneys through AI assistants. It means they arrive with a situation rather than a shortlist. My H-1B extension got an RFE and I have 87 days. My priority date retrogressed. Can I travel with adjustment of status pending? The informational answer is the moment your firm is either named or not. In a category where half the conversations are someone describing a problem, being the practice cited in the explanation is the whole game.
Google pruned the pool
The August 2026 spam update rolled out globally across all languages, per Google's incident dashboard. It is the third spam update announced in 2026, following June.
Spam updates get read as a ranking story. That framing is now incomplete. AI Overviews and AI Mode draw their citations from the same index that spam updates reshape. When a global spam update thins out low-quality pages, it thins the pool of pages eligible to be cited in an AI answer — including thin, templated, syndicated immigration-law content that was previously sitting between a prospective client's question and your firm.
That cuts both ways. If your firm publishes original, attorney-reviewed content on the questions your clients actually ask, less competition for the citation slot is a quiet gift. If your visibility rests on a directory profile or a content mill's boilerplate pages, the ground under you just moved. Either way you would not know without a baseline.
The visits are already there
Search Engine Land's tracking covered 51,200 AI Overview events across 1,661 cited passages. Across the full dataset, 22.4% of that traffic was misattributed in Google Analytics — filed as Direct instead of Organic Search. In the worst month, 29.3%.
The mechanism is mundane. When someone clicks a cited passage in an AI Overview, Google appends a text fragment to the URL — the `#:~:text=` string that scrolls the reader to the highlighted sentence. Your site receives it. GA4 receives it. Nothing surfaces it by default.
The traffic is also brutally concentrated: the average cited passage drove 31 visits, the top one drove 2,276. Somewhere on your site there is very likely one page doing most of this work, and you cannot name it.
Put a dollar figure on it
Measurement without a number attached to it does not survive a budget conversation. So here is the arithmetic, in five inputs a managing partner already knows or can pull:
- Monthly website visits — GA4, last full month.
- Estimated share arriving from AI answers — from an AI-referred segment if you have one. If you don't yet, use 5% as a deliberately conservative placeholder and label it as a placeholder, not a measurement.
- Consultation booking rate — of site visitors, what share books.
- Consultation-to-client rate — of consultations, what share retains.
- Average matter value — your realistic average across your actual case mix.
Multiply: monthly visits × 12 × AI share × booking rate × client rate × matter value.
A worked example. A firm with 2,000 monthly visits, an 8% AI-referred share, a 4% booking rate, a 35% consultation-to-client rate, and a $4,500 average matter is looking at roughly $121,000 a year in AI-referred work. A 25% lift in AI citation share — 8% to 10% — is worth about $30,000. A 50% lift is worth about $60,000.
Those are not projections. They are the value of traffic your firm is very likely already receiving and currently filing as Direct. Run the same math with your own five numbers and you have something you can defend in a budget meeting: one figure, one lift scenario, one line item.
What to do this week
- Check whether the spam update touched you. Google Search Console → Performance, compare the seven days after August 18 to the seven before. Look at impressions, not just clicks.
- Stand up the AI-referred segment. If you haven't done Wednesday's GA4 setup, do it — one custom dimension flagging `#:~:text=` landings, about twenty minutes.
- Run the five-input calculation. Use the placeholder share if you must; recalculate in 30 days with the real number.
- Name your one page. Identify the single page carrying the most AI-referred traffic, then read the passage a prospective client is actually seeing. Confirm the fees, filing addresses, and processing times in it are still current.
- Decide the budget question. Does the lift figure justify a line in next quarter's budget? Write the answer down either way.
The bigger picture
Every one of these developments is a measurement story wearing a marketing story's clothes. Demand is up and you can't size yours. The citation pool changed and you can't see your position in it. The traffic exists and it's mislabeled in your own reports.
Small immigration firms have been told for two years that AI search is coming. It isn't coming. It is running, and the only firms making decisions about it are the ones that took twenty minutes to start counting. The number does not have to be large to be useful. It has to exist.
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 →