GrowthAgents AI

Tag: AI Search

  • ChatGPT Is Losing Ground to Claude and Gemini

    ChatGPT Is Losing Ground to Claude and Gemini

    No, ChatGPT is not the only assistant worth tracking anymore, and treating it that way now means missing nearly half the picture. New industry data shows ChatGPT’s share of AI prompt volume has been cut from dominant to roughly even with the rest of the field in just six months. Claude and Gemini picked up almost all of the difference.

    Is ChatGPT Still the Only AI Assistant That Matters?

    It’s still the biggest single assistant. It’s no longer the only one that counts. Comscore’s Q2 2026 AI Intelligence Report found ChatGPT’s share of total AI prompt volume fell from 70% to 50% between January and June 2026. That’s still real traffic — ChatGPT reached 168 million desktop conversations in June 2026 — but it’s half the share it held at the start of the year.

    Some of this is people using more than one assistant rather than switching entirely. Multiplatform visitation rose to 99 million users, up from 59 million a year earlier. Prospective clients aren’t settling on one AI tool. They’re asking the same question in more than one place.

    Where Is the Traffic Actually Going?

    Mostly to Claude and Gemini. Claude’s share of AI prompt volume rose from 2% to 11% over the same six months. Its desktop conversations climbed from 1.1 million in October 2025 to nearly 40 million in April 2026, before leveling off at 22.3 million in June 2026 — about 20 times its October 2025 level.

    Gemini’s growth is bigger in raw numbers. Its share rose from 17% to 30%, and its desktop conversations grew from 8 million to 89 million over the past year, with a similar slowdown showing up in early summer.

    And more people are showing up to an AI assistant at all. By June 2026, 35% of desktop users and 29% of mobile users had visited an AI assistant, up from 25% and 12% in January 2025. The pool of people asking AI tools questions is growing, and it’s splitting across more tools as it grows.

    Why Does This Matter for a Law Firm’s Client Research?

    Because the person typing “best [practice area] attorney near me” into an AI assistant doesn’t check which one leads the market first. They ask whichever one is already open on their phone or laptop. If your firm’s AI-visibility work only checks whether ChatGPT names you, you have no idea what the Claude answer says, or the Gemini answer — and those two now account for roughly four in ten prompts, by share of volume.

    That gap doesn’t show up gradually, either. A firm can look fine on one engine and be invisible on another in the same week, simply because each assistant draws on different sources and weighs them differently. Checking only one tells you nothing about the other two.

    We call this metric Citation Share: the percentage of AI-assistant answers in your market that name your firm. A Citation Share measured on ChatGPT alone was already a partial view. This data makes the gap bigger, not smaller.

    What Should a Small Firm Actually Track?

    Citation Share across ChatGPT, Claude, and Gemini — not one engine standing in for all three. The report itself attributes ChatGPT’s falling share to people spreading their usage across more tools, not to ChatGPT losing users outright. So the point isn’t to drop ChatGPT tracking. It’s to stop treating it as the whole picture.

    Does This Mean ChatGPT Is Losing Users?

    No. The data points to people adding assistants, not abandoning ChatGPT. Its raw conversation volume is still large. Its share of total volume is smaller because the total pool of AI usage, and the number of tools splitting it, both grew.

    Should We Stop Tracking ChatGPT?

    No. It’s still the single largest assistant by volume. The point is to stop tracking it alone. A firm that only measures ChatGPT visibility is measuring a shrinking fraction of where its prospective clients are actually asking questions.

    Is This Legal Advice?

    No. This post is informational commentary on third-party industry data about AI assistant usage. It is not legal advice, and nothing here should be read as guidance on your firm’s specific marketing, advertising, or compliance obligations. Talk to your own counsel about those.

    How Do We Find Out Where Our Firm Stands Right Now?

    Start by checking where your firm’s own site and visibility currently stand. Run a free SiteScan to see what it finds.

  • Google Is Paying Publishers for AI Answers

    Google Is Paying Publishers for AI Answers

    Google is now paying a small group of publishers when their content visibly shapes an AI-generated answer. The program is called AI Contribution, it is a pilot, and it covers roughly 100 publishers worldwide. No small law firm is in that pilot, and none will be invited. But the move itself is the story: Google just put a dollar figure on being the named source behind an AI answer.

    What is Google’s AI Contribution pilot?

    Reported by eMarketer on September 14, 2026, AI Contribution pays select publishers when their content “significantly contributes” to answers across AI Overviews, AI Mode, Gemini, and other Google AI products. It runs on a pay-per-value model, publishers can opt out, and earnings show up as monthly figures inside Search Console. What Google has not disclosed is just as telling: no public dollar amounts, no full list of participants, and no explanation of how a given answer’s value gets calculated.

    How much are publishers actually getting paid?

    Numbers are thin and inconsistent. One analyst told eMarketer the program lets “publishers price their work while letting Google retain control over how it is valued.” A separate report pegs the pilot at around 100 publishers, with one publisher reportedly pulling in over $1 million while most participants earned far less. That spread matters. It means Google isn’t paying for simple inclusion. It’s paying for outsized influence on the answer, and most publishers, even inside the pilot, don’t have it.

    Why should a small firm care about a program it can’t join?

    Because the news isn’t the payout. It’s the admission. For years, Google has pulled answers from the open web, surfaced them in AI Overviews and Gemini, and kept every click, reader, and dollar of attention on its own page. AI Contribution is Google saying, for the first time, that the source behind the answer has measurable value worth paying for.

    That logic doesn’t stop at 100 publishers. It applies to every answer AI gives, including the one a prospective client gets when they ask ChatGPT, Gemini, or Perplexity which family lawyer to call in their city. If Google is now pricing “being the source,” a firm should be asking whether it even is the source, in its own market, for the questions its clients are already asking an AI assistant.

    What’s the real takeaway for a law firm or professional services business?

    Not “sign up for Google’s publisher program.” You can’t, and even the publishers who can are mostly earning little for the effort. The takeaway is narrower and more useful: visibility inside an AI answer is now a recognized asset, not a hypothetical one. Google just validated it with real money.

    At GrowthAgents, we call this visibility Citation Share: the percentage of AI-assistant answers in your market that name your firm by name. A firm with low Citation Share is invisible in exactly the moment a prospect is deciding who to call. A firm with high Citation Share shows up in that answer, with no check from Google required.

    Be clear-eyed about the difference. Google’s pilot pays a royalty. Citation Share earns attention and credibility, the kind that gets your firm into the conversation before a prospect ever searches your name. That’s the win available to a small firm right now, and it’s one worth measuring on its own terms.

    FAQ

    • Will my firm ever get a check from Google for AI Overviews? Almost certainly not. The pilot is limited to roughly 100 publishers, with no stated path for small firms or professional services sites to join.
    • What should a firm track instead of waiting on a payout? Whether it shows up by name in AI answers for its market, alongside standard, provable metrics like search impressions, rankings, and clicks.
    • Is this legal advice? No. This post is informational commentary on a reported industry program, not legal advice for your firm or jurisdiction.

    Want to see where your site stands before worrying about what Google pays anyone else? Run it through the free SiteScan tool, it grades five pages and rewrites your weakest one with a before-and-after score.

  • Is Your Agency Selling Rebranded SEO as GEO? 7 Questions

    Is Your Agency Selling Rebranded SEO as GEO? 7 Questions

    Is your agency’s “AI search” pitch just SEO with a new coat of paint?

    Some agencies now sell “GEO” or “AEO” as a new service line, but the work underneath hasn’t changed. Law firm marketers at recent conferences report hearing AI search mentioned in every meeting while nothing in the actual deliverables shifts. Here are seven questions that separate real work from a relabeled invoice.

    This matters because GEO work is harder to audit than traditional SEO. A keyword ranking is a number anyone can check in five minutes. An AI assistant’s answer changes by the hour, varies by region, and differs between ChatGPT, Perplexity, and Google’s AI Overviews. That makes it easy for an agency to describe activity instead of outcomes — and easy for a managing partner to not know the difference until the invoice arrives.

    Why this question matters right now

    Consumer research behavior is moving fast. A July 2026 survey of 1,110 U.S. consumers found Google usage for researching attorneys fell from 86.7% to 71.9% in a single year, while ChatGPT usage climbed from 28.1% to 41.9% (iLawyer Marketing, 2026). At the same time, a separate analysis of law firm website traffic found AI referrals averaging just 0.47% of sessions (LawFuel, Sept 2026). Translation: consumers are asking AI assistants about attorneys more than ever, but most firms can’t yet see it in their own analytics. That gap is exactly where a relabeled SEO pitch hides. One industry breakdown put it plainly: some agencies talk about AI search “in every meeting” while “nothing they’re actually doing has changed” (MeasureMarketing, 2026).

    7 questions to ask on your next agency call

    1. Which exact prompts do you track, and across which AI assistants?

    Real answer: a named list of prompts tied to your practice areas, tracked across more than one assistant (ChatGPT, Perplexity, Gemini, etc.), reviewed with you.
    Dodge: “We monitor your AI visibility” with no prompt list and no named engines.

    2. How many runs per prompt, and how often?

    Real answer: a stated run frequency and count — AI answers vary run to run, so one-off checks are noise.
    Dodge: “We checked and you showed up” — a single snapshot presented as a result.

    3. Do you report brand mentions and linked citations separately?

    Real answer: two distinct numbers — how often your firm is mentioned by name, and how often a response links to your site. This is the idea behind what we call Citation Share: mentions and citations never get blended into one inflated figure.
    Dodge: a single combined “AI visibility score” that hides which number is actually moving.

    4. What changed in our actual content because of this work?

    Real answer: specific pages rewritten, specific structural changes made — not just a new report format.
    Dodge: a shinier dashboard covering the same unchanged pages.

    5. How do you structure pages differently for AI extraction versus classic SEO?

    Real answer: a concrete explanation — direct-answer structure near the top, clear headings, extractable facts — distinct from keyword-density tactics.
    Dodge: “We optimize for both” with no description of what’s actually different on the page.

    6. What’s your plan when an AI engine’s behavior changes?

    Real answer: an acknowledgment that engines update constantly and a stated process for re-checking and adjusting.
    Dodge: silence, or an assumption that today’s approach works forever.

    7. Can you show one provable before/after example?

    Real answer: one specific page, one specific prompt, a documented change over time.
    Dodge: vague claims about “overall improvement” with nothing you can verify yourself.

    FAQ

    Is GEO the same thing as SEO?

    No. Classic SEO targets search engine rankings; GEO (generative engine optimization) targets how AI assistants summarize and cite your firm in conversational answers. They overlap but aren’t identical.

    Why don’t AI referrals show up much in my analytics yet?

    Research suggests AI-driven research is rising fast, but much of it doesn’t generate a trackable website click — someone asks an assistant, gets an answer, and calls the firm directly. That’s part of why relying on referral traffic alone to judge an agency’s GEO work is misleading.

    What should I ask for in writing?

    The specific prompts tracked, the engines covered, and separate mention/citation numbers — not a single blended score.

    Not sure if your current agency’s GEO work is real? Run your site through our free SiteScan — it grades your site across five pages and rewrites the weakest one, so you can check their claims against an independent read of your own site.

  • No Law Firm Cracks the Top 20 AI Visibility List

    No Law Firm Cracks the Top 20 AI Visibility List

    Ask ChatGPT, Gemini, Perplexity, or Grok how to find a personal injury lawyer, and the answer won’t have your firm’s name in it. A new industry benchmark released October 1, 2026 checked 4,000 AI responses to legal-consumer questions and found zero individual law firms among the top 20 most-visible names. Directories, bar associations, and rating sites own that answer right now, which means the fight for AI visibility is happening at a different layer than most firms are watching.

    What did the benchmark actually find?

    Zen Media ran 1,000 legal-consumer prompts through four AI assistants using its ZAVI AI visibility engine, producing 4,000 total responses from 165 consumer personas, split roughly 71.7% comparison-style prompts and 28.3% informational ones. The headline finding: no single law firm cracked the top 20 names AI assistants surface when consumers ask how to find a lawyer. The names that did show up were directories, rating platforms, and professional associations, not firms.

    Which names are AI assistants actually naming?

    By company, Avvo appeared in 45% of responses, Martindale-Hubbell in 32%, Super Lawyers in 24%, FindLaw in 15%, and LegalZoom in 13%. Zoomed out by category, legal directories and rating platforms accounted for 45% of all company mentions, professional associations 20%, and online legal services 15%. That’s the entire visible map. There’s no slice left over labeled individual firm.

    What do AI assistants cite as sources?

    Citations tell a similar story. State bar associations were cited in 92% of sourced answers, ahead of Avvo at 85%, Martindale-Hubbell at 78%, and Super Lawyers at 65%. Zen Media CEO Sarah Evans put it plainly: “When someone asks an AI assistant how to find a personal injury lawyer, the answer almost always points them to the state bar, to attorney directories and to questions about contingency fees,” a quote reported in the original release.

    What is the AI actually trying to answer?

    The benchmark also tracked recurring themes inside the answers themselves. Credential verification appeared in 92% of responses, specialization in 87%, and fee transparency in 85%. Put those together and a pattern emerges: AI assistants aren’t trying to recommend a lawyer. They’re trying to help someone vet one. That’s a different job than ranking on Google, and it rewards different signals.

    Is AI visibility a lost cause for law firms, then?

    No, but the benchmark is a useful correction to how most firms think about it. If the goal is to show up when someone asks an AI how to find a lawyer in general, directories and bar associations have already won that query, structurally, because they’re the credential-and-directory layer AI assistants are built to lean on. The more useful question for a firm isn’t “do I rank,” it’s whether the AI assistant names you by name when a prospective client asks something specific to your practice area, your city, or your case type. We call that second thing Citation Share: the percentage of high-intent questions prospective clients ask AI assistants in your market that actually name your firm. It’s a different measurement than rankings, and it’s the one that matters once you accept that generic queries go to Avvo.

    How do you find out if your firm gets named at all?

    Most firms don’t know the answer because nothing in their existing SEO reporting tracks it. Rankings measure where you sit on a results page, not whether an AI assistant says your name out loud. You can sit at the top of page one and be invisible in every AI answer, because these engines answer questions rather than match keywords. The first step is simply seeing what your site gives an AI assistant to work with in the first place.

    Frequently asked questions

    Does this mean AI assistants never mention a specific law firm?

    No. The benchmark measured the top 20 most-visible names across broad, comparison-style legal queries, and none were individual firms at that aggregate level. Narrower, more specific prompts can still surface individual firms, the benchmark just shows they don’t dominate the general “how do I find a lawyer” conversation.

    Should a law firm stop investing in SEO because of this?

    No. Traditional search and AI-assistant visibility are different surfaces with different rules, and both matter. This benchmark is a reason to add AI visibility as a thing you measure, not a reason to drop anything you’re already doing.

    What is Citation Share, in plain terms?

    It’s the percentage of the high-intent questions your prospective clients ask AI assistants that actually name your firm, as opposed to naming a directory or a competitor instead.

    Why do state bar associations and directories dominate instead of firms?

    Per the benchmark, AI assistants lean heavily on credential verification, specialization, and fee-transparency signals when answering legal questions, and bar associations and directories are built around exactly that kind of structured, verifiable information.

    Want a quick read on how your own site presents that kind of information to an AI assistant right now? Run it through SiteScan. It’s free, takes about two minutes, grades your site across 5 pages, rewrites your weakest one, and shows you the before-and-after score. No card required.

    This article summarizes a third-party industry benchmark for informational purposes and is not legal advice.