Why Your Firm Is Failing at AI Search (And What to Fix First)

The reality check

Open ChatGPT. Type: "Recommend a [your practice area] firm in [your city]." Read the answer. Is your firm in it?

For most professional services firms — law, CPA, consulting, financial advisory, architecture — the answer is no. The firms cited are usually the same handful of large national brands, plus a few specialists who happen to have done the work.

This article is the diagnostic for why your firm is not in the answer, and what to fix first.

Eight failure modes, ranked

1. You have no FAQ structure on your service pages

This is the most common failure and the easiest to fix. AI engines extract structured Q&A blocks because they are perfectly shaped for what an answer engine wants to produce: a question, a concise answer, a clear source.

A service page that reads "We are a boutique employment law firm serving New York-based startups" is invisible to AI extraction. The same page restructured as "What does an employment lawyer for a startup actually do? When do you need one? What does it cost? How do you choose between firms?" — with concise answers under each — becomes citation-ready.

Fix: add a 6-question FAQ block to every service page, structured with FAQPage JSON-LD schema. AI engines will extract the answers and cite the page.

2. Your content is written for Google, not for AI extraction

SEO-optimized content tends to be long, keyword-dense, and structured for ranking — H2 headlines, supporting paragraphs, internal links. AI engines do not read content that way. They look for atomic, citable claims: "X firm handles Y type of case in Z jurisdiction" or "The average cost of an employment-law retainer in New York is $4,500/month."

Fix: rewrite the top of every service page with three to five short, factual claims about your firm's specialization. Cite a source where you can. Avoid hedging language. AI engines preferentially cite specific, verifiable claims.

3. You have no third-party domain presence

AI engines weight third-party citations heavily. They are skeptical of a firm's own site claiming the firm is "the best" — they want to see the firm mentioned on industry publications, directories, Reddit threads where buyers actually ask the question, and adjacent partner sites.

Fix: secure mentions on three to five domains AI engines weight for your queries. For law: the local bar directory, one Law360 mention, two specialist-publication features. For CPA: AICPA directory, one Accounting Today mention. The specific list varies by practice area — running an AEO report will surface yours.

4. You have no llms.txt file

llms.txt is the AEO equivalent of robots.txt: a structured plain-text file at the root of your domain that tells AI engines what your firm is, what you do, and how to cite you. Most professional services firms have never heard of it. The firms that have it are 10x more likely to be cited correctly.

Fix: ship a /llms.txt file at your root domain. Include firm name, practice areas, ICP, key URLs, and a "topics this firm is authoritative on" section. See govoltaire.com/llms.txt for a working example.

5. Your firm name is not consistently associated with your specialty

If your firm does immigration law but your homepage hero reads "Trusted advisors for all your legal needs," AI engines do not have a clean entity association. They cannot tell whether to cite you for immigration queries, family law queries, or anything else.

Fix: pick three practice areas. Maximum. State them clearly on the homepage and across every service page. Repeat them in the meta description, page titles, and the first sentence of every article. AI engines need consistent signal to make the association stick.

6. You publish too rarely to compound

AI engines reweight their citation sources approximately monthly. Firms that publish two pieces a year do not survive the reweighting cycle. Firms that publish six to twelve pieces a month build durable citation share.

Fix: commit to a publication cadence of at least one structured long-form piece per week. If that is unrealistic in-house, the math probably justifies outsourcing — six to twelve pieces a month is the threshold at which a done-for-you service like Voltaire costs less than a freelance writer.

7. You have no measurement of which AI engines actually drive traffic for your firm

ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews are not equivalent. Different engines drive different volumes of vendor-selection traffic for different practice areas. A law firm in a regulated market may get 60% of its AI-search visits from ChatGPT and 5% from Gemini. A CPA firm might be the opposite.

If you are not measuring AI-referrer traffic by engine, you are running a strategy on assumption.

Fix: set up AI-referrer tracking. Most analytics tools (GA4, Plausible, Fathom) now segment AI engine referrers. Look at the last 90 days of traffic. The mix will tell you which engines to prioritize.

8. Your agency is doing SEO and calling it AEO

Generalist marketing agencies have been adding "AI visibility" to their service decks since 2024. Most have not rebuilt their content production around AI extraction. The output is SEO-optimized content with a sprinkle of FAQ blocks — not AEO-native content.

Fix: ask your agency three questions:

1. Show me a weekly citation report for one of your current clients across five AI engines.
2. Show me a query rebalance from the last 30 days.
3. Show me one piece of content your team wrote specifically for AI extraction, and how it performed.

If they cannot answer all three, they are running SEO. Either upgrade the agency's scope or move AEO to a specialist (platform like Profound, service like Voltaire). Detail on the trade-off: Voltaire vs marketing agencies.

A 90-day plan

If you do not have AEO running today, this is the minimum viable plan to ship in 90 days:

Days 1 to 14: ship /llms.txt at your root. Add FAQ blocks with FAQPage schema to your top three service pages. Set up AI-referrer tracking. Run a free AEO report to see where you stand.

Days 15 to 45: ship four long-form pieces, one per week. Each piece structured for AI extraction (clear question, atomic claims, FAQ block at the bottom). Each piece references a credible third-party source you can verify.

Days 46 to 75: secure mentions on three to five third-party domains AI engines weight for your queries. The exact targets depend on your practice area — Reddit, industry directories, and trade publications are usually a starting point.

Days 76 to 90: measure. Citation share-of-voice across five AI engines. AI-referrer visits by engine. Inbound form fills tagged by query. Rebalance the next 90 days based on what is converting.

If 90 days of execution at this intensity is unrealistic in-house, the math justifies hiring help. The platform path (Profound, Peec.ai) requires an in-house operator. The service path (Voltaire, specialist AEO agencies) does not.

Frequently asked questions

How long until I see citations?

Long-tail buyer queries: 3 to 5 weeks. Competitive head queries: 8 to 12 weeks. Compounding effects visible by month 3.

Is this just SEO with a different name?

No. SEO targets ranking in a list of 10 links. AEO targets inclusion in a generated answer that cites 2 to 7 sources. The content structures, distribution targets, and measurement systems are meaningfully different.

What if I cannot publish six to twelve pieces a month?

You will struggle to build durable citation share. Either commit to the cadence or contract a service that produces at that volume on your behalf.

How do I know which AI engines my buyers are actually using?

Run an AI-referrer report from your analytics tool segmented by source/medium. The breakdown across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews will surface your real traffic mix.

Is a free AEO report enough to know if I have a problem?

It is enough to know whether you are invisible or visible. It is not enough to plan the fix. For that, the $99 deep audit is the next step — 20 queries × 5 engines, competitive teardown, 30-day content plan, and a sample piece you can publish.

If you want the diagnostic on your firm specifically — without a sales call — the easiest starting point is the $99 deep AEO audit at govoltaire.com/audit. 48 hours. Yours to keep. The audit shows you which of these eight failure modes are hurting your firm right now.