One in Five Top-Ranking Law Firm Pages Is Even After the September Update
CLM Sequoia scored every law firm page ranking in the top 8 Google results for 2,418 competitive legal searches on October 16, 2026, eight days after the September spam update completed.
AI Update Changed Nothing
AI CONTENT IN GOOGLE
What's on the law firm pages ranking today
On October 1, Google quietly rewrote its page on AI-generated content. The new version says it is critical to manually fact-check everything a language model produces before publishing it, down to the title tags and the alt text. A week later the September spam update finished rolling out. Law firms have been asking the same question ever since: does this mean AI content is finally getting penalized?
Rather than guess, eight days after the spam update completed, Custom Legal Marketing pointed CLM Sequoia at the top 8 Google results for 26 competitive legal searches in 93 U.S. metros, 2,418 queries in all, and scored every law firm page that showed up for AI-generated content. That came to 5,182 unique pages and 6,044,850 words. It is the largest study of its kind CLM has run, and it follows the smaller study published in February, before any of this year's updates landed.
The short version: a little over a quarter of the words on ranking law firm pages are machine-written. Where a page ranks has nothing to do with it. What the page is about, how long it is, and who published it have a lot to do with it. And nothing Google did this fall changed any of that.
What's on the law firm pages ranking today
CLM Sequoia scored every law firm page ranking in the top 8 Google results for 2,418 competitive legal searches on October 16, 2026, eight days after the September spam update completed.
Source: CLM Sequoia, October 16, 2026.
Law firm SEO agencies love talking about content, links, and page speed while treating URL structure like an afterthought.
That is a mistake, and not just because we empirically debunked Google PageSpeed scores having any effect on law firm rankings.
We cranked up our super AI law firm marketing platform, CLM Sequoia's research tool and analyzed 31,977 ranking URLs pulled from 9,216 Google searches across 288 U.S. markets and 8 major practice areas to see what the top-performing law firm pages actually look like.
Real world data is what makes CLM Sequoia smarter than any SEO platform so we wanted to know what is really showing up in the rankings, at scale, across some of the most competitive legal searches in the country.
Our report shows how URL paths, keyword placement, geographic targeting, site depth, and page structure align with stronger organic performance. While some of our studies have focused on debunking old common wisdom, this study on URLs actually pointed us to something very important that is not at all a vanity metric.
Let's dive into the ultra-powerful yet under-respected humble URL.
How much AI content is on the law firm pages ranking today?
Powered by CLM Sequoia
A quick word on the measurement. For every law firm page that ranked, CLM Sequoia pulled the page text and ran it through AI content detection. The detection model reads the writing and returns one number: the percentage of words on the page it believes a machine wrote. A page at 3% reads as human. A page at 97% reads as generated. It is a judgment about writing patterns, not a record of how the page was made, and the end of this report comes back to what that means.
The distribution is sharply two-sided. 60% of pages score 10% or less. 21.6% score 76% or more, and 13.4% score exactly 100%, meaning the detection model flagged every single word. The middle is nearly empty. Fewer than one page in five lands anywhere between 11% and 75%. Lightly edited AI drafts, the kind that would score in the 40s and 50s, turn out to be rare. Law firms are either writing their pages or generating them, and very few are doing the thing Google actually asked for, which is generating and then editing hard.
The middle is empty. 60% of ranking pages score 10% AI or less and 21.6% score 76% or more. Fewer than one page in five lands in between. Firms are either writing their pages or generating them.
Where a page ranks tells you nothing about its AI score
This is the finding law firms keep asking about, and the answer is blunt. The share of pages scoring 50% or higher is 28.3% at position 1 and 29.4% at position 8. Across all 11,738 law firm ranking rows, the correlation between position and AI content is zero for practical purposes. A page scored at 95% AI ranks first for competitive terms in this data. A page it scores at 2% sits at position eight. Both happen all the time. Whatever Google's quality systems are rewarding, it is not something a detector can see.
Share of pages with 50%+ AI content, by ranking position
Positions 1 through 8, 11,738 law firm ranking rows, 93 metros, October 16, 2026.
Ranking position is not correlated with AI content share (Spearman rho = 0.017, p = 0.07). Position #1 vs. #8 share of 50%+ pages: p = 0.55 (chi-square). Source: CLM Sequoia.
The keyword matters. The city does not.
The biggest markets do not look different from the smaller ones. The 24 largest metros in the study had a 50%+ share of 27.7%; the 69 smaller metros came in at 28.2%. No individual metro stands out once sample size is accounted for. A firm in Boise is competing against the same amount of AI content as a firm in Chicago.
The search term is a completely different story. The share of mostly-AI pages runs from 14.8% for "dwi lawyer" to 40.3% for "assault lawyer", and that spread is not noise. Estate planning and trust searches, corporate law, car accidents, and assault defense sit at the top. The DUI and DWI gap is the strangest thing in the dataset. The two terms describe the same offense, yet "dui lawyer" pages are more than twice as likely to be mostly AI as "dwi lawyer" pages. The likeliest explanation is that DWI is the statutory term in a smaller set of states, so those pages are more often written by a local criminal defense lawyer for one jurisdiction, while DUI pages get produced at national scale by firms and lead-gen operations that cover everywhere at once.
The six most and six least AI-heavy searches
Share of law firm ranking rows scoring 50%+ AI, by keyword, across 93 metros. Full October dataset, 26 keywords.
Keyword is a significant predictor of AI content share (p < 0.001). Note that "dui lawyer" and "dwi lawyer" return very different pages: 35.6% vs. 14.8%. Source: CLM Sequoia.
AI content by practice area, October 2026
Full October dataset: 93 metros, positions 1 to 8, 5,182 unique law firm URLs. Averages are per unique URL; the 50%+ share is per ranking appearance.
| Practice area | Law firm URLs | Avg AI content | Pages 50%+ AI |
|---|---|---|---|
| Estate planning | 658 | 30.3% | 30.9% |
| Criminal defense | 1,036 | 27.4% | 30.4% |
| Medical malpractice | 520 | 28.4% | 30.3% |
| Family law | 540 | 29.4% | 28.1% |
| Business law | 520 | 23.6% | 26.4% |
| Personal injury | 1,304 | 29.7% | 26.1% |
| Employment law | 604 | 23.8% | 24.4% |
Source: CLM Sequoia, rankings captured October 16, 2026.
The multi-market firms are the heaviest users
154 law firm domains rank in 11 or more of the 93 metros. Between them, they occupy 37.2% of every law firm result in the top 8. Those firms score 50%+ AI on 33.8% of their ranking pages, against 26.2% for firms that rank in a single metro and about 24% for everyone in between. That is the clearest signature of scaled content production in the whole study. A firm building practice-area pages for dozens of cities is far more likely to be doing it with a language model, and that will surprise no one who has looked at those pages.
It is also exactly where the risk Google described on October 1 concentrates. A hallucinated fact in a template gets copied into every city the template serves. One wrong statute of limitations becomes sixty wrong statutes of limitations. The group is not uniform, though. Nearly half of those multi-market domains average 10% or less AI content across their pages, so operating at scale does not require it. Some of the biggest firms in the country are still writing their own pages.
Firms ranking in 11 or more metros carry the most AI content
Share of law firm ranking rows scoring 50%+ AI, grouped by how many of the 93 metros the firm's domain ranks in. Full October dataset.
154 domains rank in 11 or more metros and together hold 37.2% of all law firm ranking rows in the study. Their 50%+ share is higher than every other group (p < 0.001). The group is split, though: 48.1% of those domains average 10% or less AI content and 29.9% average 50% or more. Source: CLM Sequoia.
AI pages are longer, buried deeper, and harder to read
Pages scoring 50% or more AI have a median length of 1,149 words. Pages scoring 10% or less have a median of 704. Pages over 1,500 words are mostly AI 37.0% of the time, compared with 21.5% for pages under 1,000. CLM's earlier research on whether word count matters for rankings, found that length by itself does not move the needle. What this study adds is that the long pages are disproportionately the generated ones.
Page type matters too. Homepages are the least AI-heavy page type on law firm sites at 18.3%, against 27.2% for interior practice-area and location pages. That fits a pattern CLM has documented before: Google rewards internal pages while AI engines cite homepages, and firms seem to put their own words on the homepage and let the machine fill out the rest. The generated pages also read harder. Pages scoring 90% or more have a median readability score of 30.4, versus 36.1 for pages at 10% or less, on a scale where lower means more difficult. The AI output is wordier and denser than the pages lawyers write themselves, which is the opposite of what a potential client wants from a practice-area page.
Longer pages score higher for AI content
Share of unique law firm pages scoring 50%+ AI, by total word count. Full October dataset.
Median length of a page scoring 10% or less: 704 words. Median length of a page scoring 50% or more: 1,149 words. Spearman rho between word count and AI score = 0.2. Homepages (503 pages) score 50%+ at 18.3%, compared with 27.2% for internal pages (p < 0.001). Source: CLM Sequoia.
What Google actually changed this fall
Google's search status dashboard lists two confirmed core updates in 2026, in March and May, and four spam updates. The most recent, the September 2026 spam update, started on September 24 and took almost 14 days to finish, wrapping up on October 8. Search Engine Land documented three separate waves of ranking volatility during that window. No October core update had been announced as of October 19, despite plenty of forum chatter, according to Search Engine Watch. Firms that were hit can start with CLM's guide to recovering from a core update still applies.
The documentation change landed on October 1. Google's Search Central page on using generative AI content picked up language from the Search Quality Rater guidelines: generative models predict likely word sequences rather than retrieve facts, their output can contain hallucinations, and it is critical to manually fact-check and review all AI-generated content for accuracy before publishing. Search Engine Roundtable pointed out that Google almost never uses the word critical in its help docs.
Read what it does not say. There is no penalty for AI-generated content. The spam policies did not change. Google's position is the same one it has held for years: the question is whether the content is helpful and original, not how it was produced. The scaled content abuse policy that already covered mass-produced low-value pages is untouched. Google added a warning label. It did not add a rule.
What Google shipped between the two studies
Confirmed ranking updates and documentation changes, February through October 2026, in sequence (spacing is not to scale). No October core update had been announced as of October 20.
How this compares to February 2026
CLM Sequoia ran a smaller version of this study on February 7, before any of the year's updates: 24 metros, top 5 results, 1,579 law firm pages. 528 keyword and city queries appear in both runs, which allows a clean comparison. On that matched set, the share of law firm pages scoring 76% or higher was 19.7% in February and 19.8% in October. The share at 50% or higher went from 28.3% to 25.2%, which is inside sampling noise. Position was flat in both runs.
The study also tracked the firms themselves, because a snapshot can miss a penalty if the penalized sites simply disappear and get replaced by similar ones. Of the 834 law firm domains that ranked in the matched February queries, 36.9% still ranked in the same queries in October. Domains that averaged 0 to 10% AI content in February survived at 36.1%. Domains that averaged 51% or more survived at 33.0%. If the spam update or the guidance change had gone after AI-written law firm pages, that second number would be much lower than the first. It is not.
Did AI-heavy firms get pushed out since February?
Of the 834 law firm domains ranking in the 528 queries CLM Sequoia ran in both February and October, the share still ranking in the same queries in October, grouped by their February AI content average.
Overall survival rate: 36.9% of domains. The mostly AI group survived at a rate statistically indistinguishable from the low AI group (Mann-Whitney p = 0.85). Source: CLM Sequoia.
One caution for anyone putting the two reports side by side. AI detection models are tuned regularly, and between February and October the scoring changed. The evidence is direct: 243 law firm pages ranked in both studies with an identical word count, meaning nobody touched them. In February, 105 of those pages scored exactly 1%. In October, 98 of the same pages scored 0%. The model went from "a few machine-sounding words" to "nothing here" with no change to the text, and a smaller group of pages moved by larger amounts, almost all downward. That is why the February report said 99% of pages contained some AI content and this one says 56%. The floor moved. The pages did not. The numbers that hold up across both runs are the share of pages at 76% or higher and the survival rate, and both stayed put. If anything, a model that now scores the same text lower and still finds one in five pages mostly AI suggests the real number has not gone down.
What law firms should take from this
Eight months, two core updates, four spam updates, and an explicit new instruction from Google to fact-check AI output did not change the share of mostly-AI pages ranking at the top of competitive legal searches. That holds in every metro, at every ranking position, and in markets of every size. It is consistent with what Google has said all along. Its systems evaluate whether content is useful, accurate, and original. Detecting AI authorship is not the mechanism, and nothing in this data suggests it is becoming one.
So the October 1 guidance is still the most important document here, because it describes the actual risk. Google did not say AI content ranks lower. It said AI output contains mistakes, and publishing it without review is the problem. For a law firm that is not abstract. A hallucinated statute of limitations, an invented case result, a wrong filing deadline on a YMYL page a potential client relies on. This study shows where that exposure sits: in the long interior pages, in estate planning, corporate, DUI, and car accident content, and above all in the multi-market firms producing those pages for dozens of cities from one template. The scaled content abuse policy already covers firms that publish hundreds of thin city pages, and the spam updates are where that enforcement happens.
What this study cannot show is whether the pages scoring high for AI are any good. Detection tools measure patterns in prose. A carefully edited draft that started in a language model will often score high, and a thin page a paralegal wrote in twenty minutes will score low. The questions worth asking about a firm's own site are different. Has a lawyer checked every factual claim on every practice-area page? Do the pages say something a competitor's pages do not? And is the firm described the same way on its site and in the directories AI engines use to verify it, because those engines rarely agree on which firms to recommend and consistency is what tips it. CLM Sequoia will run this study again after the next confirmed core update, and the results will be in the research library when we do.
FAQs About AI Content in Google
Does Google penalize AI-generated content on law firm websites?
Not based on this data. CLM Sequoia found no relationship between a page's AI content score and its ranking position across 11,738 law firm results, and law firms whose pages averaged 51% or more AI content in February kept their rankings through the September 2026 spam update at the same rate as firms under 10%. Google's October 1, 2026 guidance update added a warning to fact-check AI output; it did not add a penalty.
How much of top-ranking law firm website content is AI-generated?
About 27.9% of the words on law firm pages ranking in the top 8 Google results for competitive legal searches were flagged as AI-generated, based on 5,182 unique pages across 93 U.S. metros. 21.6% of those pages were 76% or more AI, and 13.4% scored exactly 100%. At the other end, 60% of pages scored 10% or less.
Does AI content affect where a law firm page ranks in Google?
No measurable effect. 28.3% of first-position pages were mostly AI, compared with 29.4% of eighth-position pages, and the correlation between position and AI content across all results was effectively zero. Pages scoring 95% AI rank first for competitive terms, and pages scoring 2% rank eighth, and both are common.
Which practice areas and keywords have the most AI-generated content?
By keyword, the share of mostly-AI pages ranged from 14.8% for "dwi lawyer" to 40.3% for "assault lawyer". Estate planning and trust searches, corporate law, car accident, and DUI searches sit at the top. By practice area, the spread is narrower: medical malpractice, criminal defense, and estate planning lead, while employment law and business law trail. Market size and individual metro made no difference.
What changed in Google's October 2026 AI content guidance?
On October 1, 2026, Google updated its Search Central page on using generative AI content with language from the Search Quality Rater guidelines. The new text says generative models predict likely word sequences rather than retrieve facts, that their output can contain hallucinations, and that it is critical to manually fact-check and review all AI-generated content, including titles, meta descriptions, structured data, and alt text, before publishing. The spam policies and the scaled content abuse policy were not changed.
The Fine Print (Disclaimers)
This analysis is part of a comprehensive study conducted by Custom Legal Marketing using its proprietary CLM Sequoia platform. The full dataset includes 19,138 SERP-level data points across 26 legal keywords, 7 practice areas, and 93 U.S. metropolitan areas, captured on October 16, 2026, with 5,182 unique law firm pages and 6,044,850 words scored for AI-generated content. The February 2026 comparison draws on the 24 metros and 22 keywords common to both studies at positions 1 through 5.
Rankings and page content represent a point-in-time snapshot and may fluctuate. Google's search results vary by device, location, personalization, and time of day. AI content scores are probabilistic estimates of writing patterns; no AI detection method is fully accurate, and a score does not establish how a page was actually produced. Sequoia referenced Winston AI for AI detection analysis. Detection scoring changed between the February and October studies, so average AI percentages are not directly comparable across the two; the share of pages at 76% or higher and domain survival are the measures that hold across both. Only the single ranking page content was scored. All correlation findings describe observed patterns in the data and do not establish causation, and the absence of a relationship between AI content and ranking position does not mean content quality is irrelevant to rankings. AI content is one of many attributes of a ranking page, alongside content quality, accuracy, backlink profiles, domain authority, technical performance, and user engagement signals. The study analyzed top-level, high-intent practice area keywords; findings may not generalize to long-tail queries, informational content, or AI answer engines.
© 2026 Custom Legal Marketing, an Adviatech Company. All rights reserved. This research was conducted using proprietary CLM Sequoia technology and methodology.
Jason and his team are the best. He has solid ideas that produce results for the long term. We had tried other companies, but none of them produced the quality results we received from CLM.
- RICHARD WIRTZ
Wirtz Law APC