Now optimizing law firms for AI search: Google, ChatGPT, and AI Overviews.

← Back to all articles

ChatGPT vs. Claude: How AI Assistants Recommend Law Firms Differently

More potential clients are skipping Google entirely and asking an AI assistant a simpler question: "Who's the best personal injury lawyer near me?" At Lexigate, we've been auditing how the two biggest assistants, ChatGPT and Claude, answer that question for the markets our clients compete in.

Here's the headline: they don't think alike. The two assistants pull from different sources, trust different signals, and explain their picks in very different language. A firm that looks great to one can be nearly invisible to the other.

Here's a side-by-side of the patterns we've observed, followed by what's behind each row.

The Short Version

SignalChatGPTClaude
Reviews Gets Google ratings secondhand, mostly via aggregators (Birdeye, Trustindex, Expertise.com) Reads Google listings directly. Sorts by star rating first, review count second, so quality beats quantity
Firm website Primary source by far. Takes claims at face value Uses it but labels claims as "self-reported" and discounts them
Third-party credentials Mentions awards when found, but not a major ranking driver Major ranking driver. Best Lawyers, Super Lawyers, Martindale, and documented verdicts push firms up
How it explains rankings Positioning and perception: local vs. regional, boutique vs. high-volume, TV-ad heavy Data-driven: Google sort order, verifiable credentials, or simply search-result order
Consistency checks Rarely flags discrepancies Catches mismatches across sources (founding dates, review counts, attorney rosters in cached pages)
Directories and sources Avvo, Justia, bar associations, state agencies (e.g., Law Society of Alberta, NY Workers' Comp Board) BBB, Yelp, LawInfo, FindLaw, USLegal, legal industry press

Reviews: Quality Beats Quantity for Claude

Claude reads Google Business Profiles directly, and when it ranks firms on review data, it sorts by star rating first and review count second. In one market we audited, a firm with 350 reviews at 4.8 stars landed below every 5.0-star firm, including ones with a third as many reviews.

ChatGPT tends to get review data secondhand, through aggregators like Birdeye, Trustindex, and Expertise.com, plus its Yelp partnership as a review source.

For Claude, a 0.1-star difference can matter more than hundreds of extra reviews.

If your firm's rating has drifted below your market's top performers, that gap may now be costing you AI recommendations, not just Google clicks.

Your Website: One Assistant Believes You, One Doesn't

ChatGPT leans heavily on the firm's own website. In one audit, the firm's site accounted for more citations than every other source combined. It also takes claims largely at face value.

Claude reads your site too, but it's skeptical. It routinely labels claims like "$500M recovered" or "30 years of experience" as self-reported, and discounts them accordingly.

The takeaway cuts both ways: specific, detailed content on your site clearly helps with ChatGPT. For Claude, the same claims need to be backed up somewhere you don't control.

Third-Party Proof: What Claude Actually Rewards

When Claude explains why it ranked firms the way it did, the reason is almost always independent recognition: Best Lawyers, Super Lawyers, Martindale ratings, documented verdicts. The firm's own landing pages didn't carry the same weight.

ChatGPT explains its picks differently. Its reasoning sounds more like positioning language: "local vs. regional," "boutique vs. high-volume," or "known for TV advertising."

Consistency: Mismatched Info Doesn't Go Unnoticed Anymore

Claude cross-checks sources against each other. In our audits it caught a founding date on a firm's website that didn't match the BBB record, two different Google review counts on two pages of the same site, and a Yelp profile with a phone number that appeared nowhere else.

ChatGPT rarely flags discrepancies like these. But if Claude is any indication of where AI search is heading, the era of letting old directory listings drift out of date is over.

A Caveat Before You Reprioritize Everything

Both tools shift their sources depending on how a question is phrased, so we treat these observations as patterns, not rules. We re-run these audits regularly to see how both assistants evolve, and we expect the details to keep changing.

What This Means for Your Firm

A few practical moves follow directly from what we've seen:

If you've noticed other nuances in how your firm shows up in ChatGPT or Claude, we're happy to compare notes.

Share this article: LinkedIn X Copy link

Want to Win in Google and AI Search?

Let's talk about turning organic visibility, from Google rankings to AI answers, into more signed cases for your firm.

More From the Blog