By Steve Merrill · October 31, 2026
Why do AI assistants name some firms and ignore yours?
AI assistants name firms they can read and trust. When someone asks ChatGPT or Google AI for "a tax accountant near me," the tool picks firms with clean structured data, real bios, and consistent contact details across the web. If your firm is a pretty website with no machine-readable facts, you don't get named.
Here's the thing. Most law and accounting firms are invisible to AI right now.
Not because the work is bad. Because the website was built to look good to a human visitor, and nothing underneath tells a machine who you are, where you practice, or what you do. I've looked at a lot of these sites. Same story every time. Beautiful homepage. Zero facts a model can actually use.
How does AI decide which attorney or CPA to recommend?
AI search reads your structured data first, then checks whether the rest of the web agrees with it. It wants a clear entity, real credentials, and matching contact details in enough places to feel safe recommending you to a stranger. Confidence wins the referral.
Think about what a referral actually is. Someone trusts a source enough to send a client your way. AI works the same way. It won't send a person to a firm it isn't confident about, because a bad tax or legal referral is a real problem for the person asking.
So the firms getting named are the ones that make confidence easy. Everything checks out. The data lines up.
What structured data does a professional services firm need?
You need LocalBusiness or ProfessionalService schema with real contact fields filled in. That means your legal name, address, phone, hours, service area, and the exact practice areas you handle, all marked up so a machine can read them without guessing.
Most firms either skip schema or paste in a generic block with half the fields empty. That's worse than useless. According to schema.org, the ProfessionalService type supports fields for areaServed, address, telephone, and the services you offer. Fill them in. All of them.
Google's own Local Business structured data documentation spells out which properties they read for local results. Match it exactly. Don't improvise.
One rule I give every firm: the phone number in your schema has to be the phone number that actually rings your office. Sounds obvious. You'd be surprised how often it's a dead tracking line or a number from three offices ago.
Why do attorney and CPA bios matter so much to AI?
Bios are your E-E-A-T, and AI leans on them hard for professional services. A real bio with license numbers, bar admissions, years in practice, and the specific matters a person handles tells a model this is a credentialed human doing real work. That earns the recommendation.
A stock photo and two sentences won't do it.
Google's helpful content guidance puts heavy weight on experience and expertise, and for law and accounting that signal lives in the bio. Name the bar admissions. List the CPA license. Say how many years, what kinds of cases, which industries. The more specific, the more a model trusts it.
I tell firms to write bios like a referral partner would describe you to a client. Plain, specific, proud of the actual work.
What makes a practice-area page work for AI search?
A practice-area page works when it answers the real questions clients ask, right near the top, in plain words. Each practice area gets its own page. "Estate planning," "small business tax," "personal injury." One page, one focus, the answer up front where a model can lift it.
Here's what I mean. Someone asks AI "do I need a will or a trust in Texas?" The firm that gets cited is the one with a Texas estate-planning page that answers that exact question clearly, not the firm with a vague "we handle estate matters" paragraph buried on a services page.
Write the way a client asks. Answer the way a trusted advisor would. Short answer first, detail after.
How do reviews and NAP data affect AI referrals?
Reviews and consistent NAP data are the trust check AI runs before it names you. Reviews show real clients had a real experience. NAP, your name, address, and phone, has to be identical everywhere so a model can confirm you're one real firm, not a confusing mess of conflicting listings.
Let me take these one at a time.
Reviews. Earn them on Google and on the directories that matter for your field, like Avvo or Martindale for law, or your state CPA society listing for accounting. Keep them current. Respond to them. A firm with 80 recent reviews beats a firm with 6 reviews from 2021, every time.
NAP consistency. This is the boring one nobody fixes. Your firm name on the website, your Google Business Profile, your bar directory, and every legal or accounting directory has to match character for character. "Smith & Jones LLP" and "Smith and Jones, LLP" look the same to you. To a machine they might be two different firms, and that confusion costs you the referral.
The American Bar Association has covered how client search behavior keeps moving toward AI-assisted tools in its Law Technology Today coverage. The firms that clean up their data now get the head start.
What should a firm do first?
Start with the data a machine reads before anything else.
Add ProfessionalService schema with every contact field filled in. Rewrite your attorney and CPA bios with real credentials and specific experience. Give each practice area its own page that answers client questions up top. Earn current reviews on Google and your industry directories. Then make your name, address, and phone identical across every listing you can find.
Do those five things and you stop being invisible. Skip them and you stay a nice-looking website that no AI will ever recommend.
I've watched firms treat this like a nice-to-have for two years running. It's not. The client who used to ask a friend for a lawyer is now asking ChatGPT. That referral is going to someone. Might as well be you.
Frequently Asked Questions
Will AI search replace word-of-mouth referrals for law and accounting firms?
No, but it's becoming a major source of them. People still ask friends and colleagues, and more of them now ask ChatGPT, Google AI, and Perplexity first. Firms that show up in both win the most clients.
Do I need a developer to add ProfessionalService schema?
Not always. Many firm websites let you paste a JSON-LD block into the page head, and tools can generate it for you. A developer helps if your site is custom-built, but the fields themselves are straightforward once you have your real contact details.
How many reviews do I need before AI starts recommending my firm?
There's no fixed number, but recent and consistent beats old and sparse. A steady flow of current reviews on Google and your field's directories signals an active, trusted firm. Aim to keep them coming in every month rather than hitting a one-time total.
What's the single biggest mistake firms make here?
Inconsistent NAP data. Different versions of the firm name, old phone numbers, and outdated addresses scattered across directories make a model unsure you're one real business. That confusion alone can keep you out of AI recommendations.
Ready to get your firm named by AI?
If your firm is invisible to AI search, we fix the plumbing underneath, the schema, the entity data, the pages, so the tools can actually recommend you. See how it works at WRKNG Digital's agentic commerce page.
