GUMSEK

Getting a law firm named on sites it does not own

Only about 19% of the website citations in AI answers end up in the brand's own content. Directory, review and video work is the housekeeping of the law firm world, but for that reason it's become the main event. Referral traffic from directories happens to be the best converting traffic in legal.

Tell me your need

How the citations split between your site and everybody else's

Wix pulled 1,056,727 citations out of 75,000 AI answers by doing some sampling across ChatGPT, Google AI Mode and Perplexity1.

80.9% Of website citations went to third parties rather than to the brand being discussed. Only 19.1% went to the brand itself1.

Meanwhile, more than half of everything cited is one of three formats, and disproportionately listicles at that1.

  • Listicles, 21.9%
  • Articles, 16.7%
  • Product pages, 13.7%

Profound's category questions are an even more extreme example. These are the questions a shortlist of providers is made to answer. Owned content is 4.3% of the citations on those questions3.

I checked Whitespark's figures at the local level and they said the same thing. Only 40% of the AI Overview sources in their 540 searches are going to the businesses themselves and 60% going to the Reddits, Quoras, Yelps and Thumbtacks2.

What that means in your market

If you ask an AI system for a personal injury lawyer in your city, you'll likely find a directory, a best-of list, a review site or a news story behind most of the answer. Here in AI search, the firm's own website seems to be a supporting act, rather than the main source.

You can't win that query with your own website no matter how hard you try.

The directories that matter, and what they are worth

3× More likely to be chosen as a source by ChatGPT, for domains with profiles on established review platforms. SE Ranking's examples were Trustpilot, G2, Capterra, Sitejabber and Yelp4.

The legal equivalents are Avvo, Justia, Martindale, Super Lawyers, Best Lawyers, your state bar's own directory, and your Google Business Profile.

Sure, there are all kinds of factors that determine whether a profile pays.

  • Completeness beats count. Anytime profiles are left half finished, they are a weak source.
  • Details have to match exactly. Mix up the firm name, address, phone or practice descriptions across profiles, and it makes the entity tying them together very, very weak.

Once you do this, you see a second payoff in conversions which has nothing to do with AI.

8.8% Conversion rate for referral traffic from directories, review sites and partner pages in legal, the highest of any channel Ruler Analytics measures, against 7.3% for organic search11.

If someone has found you on a directory, and they call you, then they are already half sold.

One distinction to keep straight. Paying for a directory profile is a separate thing from paying for a link. The ways firms used directories in the old link days are different now, and low-quality directory links are considered link spam17. Odds are, even with all those spammy directories, there are a few legit directories that clients and referring attorneys actually use. The value of those is in citations and referrals, not ranking credit. Where do you draw the line on what counts as a directory and what's a link scheme? I suppose that is what the links page is for.

Video is the open goal, and almost nobody local is taking it

Of all the findings in this guide the YouTube one is by far the most useful6.

  • YouTube mentions (what Ahrefs was really counting when it checked titles, transcripts and descriptions across 75,000 brands) are the most strongly correlated with visibility in Google's AI Overviews
  • YouTube is the most cited domain in AI Overviews
18.2% Of AI citations to pages that did not rank in Google's top 100 were YouTube URLs6.

This number is the one that, in my opinion, matters most in practice. The YouTube route doesn't require you to rank in Google's top 100 at all. Even a modest firm website can get cited on questions it has no hope of ranking for. I can't prove YouTube mentions cause citations, but I am certainly going to act on that conclusion with the firms I work with.

People are looking for legal videos, and they're easy to make. One in six of the people FindLaw surveyed who looked for legal information online was searching for video. Of those that did, 79% used YouTube, so there is one obvious platform to upload to12.

What this actually takes

All it takes is one attorney with a phone and ninety seconds a question. You can shoot the videos yourself today without a production company.

What each video needs is a title, transcript and description that repeat the question and its answer in text.

Where else these systems look

Reddit, ChatGPT gives it a big share of its social citations (anywhere between 51-76%, and 57% in the US) depending on the country8.

Reddit is the most volatile citation source in the whole data set. Within a few weeks, Semrush saw the share of ChatGPT responses citing Reddit drop from roughly 60% to about 10%. The Reddit numbers on other platforms were not any lower afterwards9. Worth watching for the way ChatGPT changed its mind. So just don't build a strategy on that.

LinkedIn is the most cited domain for professional questions in AI search. Over Profound's tracking period, LinkedIn roughly doubled its citation frequency, as it climbed from about eleventh to fifth overall. It was cited in around 15% of the AI Mode responses7.

For a law firm, that might mean making sure every attorney profile is current and actually written, rather than a logo and a job title.

What happens when somebody asks about your firm by name

As part of that kind of search two things happen3.

  • Social citations went from 5.4% to 15%
  • Reddit alone accounts for about 8% of ChatGPT's citations on those questions

What strangers have said about your firm in public is a pretty significant part of the answer a prospective client gets.

Some systems are more likely to trust a claim with two sources than one from a single new article. Claude and Perplexity were less willing to take on claims from one new article without corroboration10. You keep one consistent story across your site, your profiles and the press, this goes further than polishing one strong page.

In terms of reviews on the other hand you absolutely should read Google's policy and make sure you get it right. The offenders Google names are reviews paid for directly or in kind, reviews written under a conflict of interest such as current or former employment, or selectively soliciting positive reviews16. The best way to ask for reviews is to follow the compliant method on the local SEO page.

The fake-list tactic, and the rule it runs into

Because listicles are so heavily cited at the moment, it only makes sense for agencies to publish "top ten firms in [city]" pages with their own clients at the top, and that's why we see them. Cyrus Shepard at Zyppy mentioned the use of fake lists, with the marketer at the top, for steering AI recommendations13.

The ninth result for "law firm seo" is an agency's "top 10 law firm SEO agencies" list14.

If you are tempted. Publishing your own "best lawyers in [city]" list with yourself first is potentially dangerous, but you may want to do it anyway. This is in possible conflict with Model Rule 7.1, which prohibits false or misleading communications about a lawyer's services. The question of whether a comparative claim can be verified is an important one, and the answer to that apparently varies from state to state15. But Florida goes further and requires comparisons of skill, experience or record to be objectively verifiable. You need to check it first, which is what the advertising rules page is all about.

Model Rule 7.3 on solicitation is important because it governs approaching people about legal services15. Answering a general legal question in public is one thing, but approaching somebody with a live matter is another. Where that line sits is going to be set by your state, not your marketing agency.

What I would do, one afternoon a week

  1. Week one: audit what already exists. My usual first step is to search for the firm name and each attorney's name and see what comes up. Google Business Profile, state bar, Avvo, Justia, Martindale, Super Lawyers, LinkedIn. Flag the gaps and note the disagreements.
  2. Week two: fix the disagreements. One canonical version of the firm name, address, phone and practice descriptions, used on every profile. This is very very boring but make yourself do it, because without a consistent foundation you are diminishing the impact of everything else.
  3. Week three: complete the profiles. Practice areas, bar admissions, photographs, the full description.
  4. Week four: start asking for reviews, with no incentives. Send the request to every client who reaches the moment you've picked for asking, including the ones you suspect are even unhappy. The reviews help you feed the map pack, AI citations and conversion rate.
  5. Week five onwards: record two videos a week answering the questions clients actually ask, with real titles and transcripts. You do 20 of them and you've covered a practice area. I would say YouTube mentions are one of the strongest off-site signals measured so far6.
  6. Ongoing: earn genuine local mentions. Bar association roles, a case the local paper would cover, a community thing you actually do. One real local mention is far better than buying fifty directory submissions.

For context, citation consistency across the primary search engines scores 28th of all 187 factors in the Whitespark 2026 survey. Its ranking is high and most link factors are lower. In Whitespark's own reading of the survey, three of the top five AI visibility factors are citation factors rather than link factors5.

Why consistency comes before video

Video is the stronger signal, provided your profiles agree with each other. Get the identity right and the video work follows.

References

  1. Wix AI Search Lab, The content types most cited by LLMs Third-party versus self-promotional citations; content type shares
  2. Whitespark (Miriam Ellis), AI Overviews for local business searches 540 manual queries across three cities and six industries, personal injury lawyers among them
  3. Profound, Who shapes AI answers: enhanced citation categories Owned versus third-party citations; brand-name prompts and social sources
  4. SE Ranking, How to optimise for ChatGPT: 129,000 domains and 216,524 pages Referring domains, domain trust, content length, section length, review profiles, speed
  5. Whitespark (Darren Shaw), Local search ranking factors 2026 (2026) 47 local search practitioners scoring 187 factors for map pack, local organic, conversion and AI visibility
  6. Ahrefs, 38% of AI Overview citations pull from the top 10 Citation and ranking overlap, YouTube share, brand mention correlation
  7. Profound, LinkedIn is the most-cited domain for professional queries in AI search LinkedIn citation growth and AI Mode share
  8. Profound, How query language reshapes AI citations Reddit share of ChatGPT social citations by country
  9. Semrush, The most-cited domains in AI: a three-month study Reddit and Wikipedia citation collapse in ChatGPT; LinkedIn growth
  10. Lily Ray, Which AI search tools are the most gullible? How different systems treat a single new source
  11. Ruler Analytics, Legal marketing statistics and conversion benchmarks Conversion rates by channel for legal, and the share of legal conversions that arrive by phone
  12. FindLaw, Statistics on how people look for a lawyer How fast legal consumers act, their use of reviews, and their use of video
  13. Cyrus Shepard, Zyppy, 17 content types to survive Google’s zero-click era (2026-08-22) Zero-click share from Datos and SparkToro; content types rated by strength
  14. Gumsek, Keyword and SERP snapshot for the legal SEO terms (September 2026) Ahrefs volumes, difficulty, CPC and the top 12 results for the head terms
  15. American Bar Association, Model Rules of Professional Conduct Rules 7.1 to 7.3 on communications, advertising and solicitation
  16. Google, Prohibited and restricted content, Maps user generated content policy Fake engagement, conflict of interest, incentives, review gating
  17. Google Search Central, Spam policies for Google web search (updated 2026-08-28) Scaled content abuse, doorway abuse, link spam, site reputation abuse