The Reliability Audit: Can AI Find You, Cite You, and Be Trusted When It Does?

Three developments landed in AI search in early July 2026, and they share one theme: reliability. Search Engine Land's research into ChatGPT's hidden retrieval pipelines found the model's primary source changed 11.6% of the time across repeated runs of the same prompt. A Fractl/Search Engine Land study of 1,008 consumers found the share who call AI search more helpful than traditional search fell from 82% to 54% in a single year. And Meta AI — sitting inside apps with 3.56 billion daily users — is quietly becoming a search surface most firms have never checked.

For a managing partner, the through-line is uncomfortable but useful: it's no longer enough to appear in an AI answer once. Your citation has to be consistent, it has to survive a skeptical reader's click, and it has to hold up on surfaces beyond the one or two you're watching.

This audit walks you through all three checks. Set aside 60 minutes, open a notepad (or print this page), and answer each prompt about your own practice. Score each section honestly — the gaps you find are your marketing plan for the next quarter.

Section 1: Citation Reliability — Does Your Firm Show Up Consistently?

Because ChatGPT's retrieval shifts between hidden backends, being cited once is not the same as being cited reliably. Run the same question a few times and watch what changes.

1. Pick your highest-value question (for example, "best H-1B attorney in [your city]"). Ask ChatGPT 3–5 times in fresh chats. Does your firm appear every time, sometimes, or never?
2. When you are cited, is it your own firm page — or a directory or aggregator standing in for you?
3. Repeat for a second practice area (family petitions, asylum, removal defense). Same consistency, or worse?
4. When a competitor appears instead, which specific page of theirs gets pulled?
5. Is the information ChatGPT attributes to you accurate and current, or is it citing an outdated page?

Write down a score from 1–5 and note the pattern you saw. If your firm appeared in fewer than half your runs, Section 4 should start here.

Section 2: Reader Trust — Does the Answer Survive Scrutiny?

Trust in AI answers is falling even as usage rises; 39% of consumers now say heavy AI use by a brand would reduce their trust in it, up from 20% a year ago. A skeptical reader clicks through to verify. Audit what they find when they do.

1. On your top practice page, is there a named, real attorney attached to the guidance — or is it anonymous "firm" copy?
2. Are your claims dated and specific (case types handled, years in practice, jurisdictions) rather than vague?
3. Does the page show proof a wary reader can check — bar credentials, real outcomes, verifiable contact details?
4. If a prospective client cross-checked your AI-cited answer against your website, would the two agree?
5. Where does a human, verifiable signal appear on the page — above the fold, or buried at the bottom?

Score 1–5 again. The pattern to look for: pages that read well but verify poorly. Those are the pages that lose the skeptical clicks AI is now sending.

Section 3: Unaudited Surfaces — Where Haven't You Looked?

Meta AI now answers questions inside Facebook and Instagram, where discovery already happens. Most firms have never checked whether they exist there. Take inventory of the surfaces beyond Google and ChatGPT.

1. Ask Meta AI (in Facebook or Instagram) for an immigration attorney in your city. Does your firm come up?
2. Have you checked Perplexity and Google's AI answers for your top three practice areas this month?
3. Is your Facebook and Instagram business presence complete and consistent with your website — name, services, contact?
4. Which surface sends you the most inquiries today — and which have you never measured at all?
5. If a client discovered you on a social platform's AI instead of a search engine, would your profile there convert them?

Score 1–5. Most firms score lowest here, and that's not a failure — it's the cheapest opportunity on this page, because so few competitors have checked either.

Section 4: Closing the Gaps

Now turn your three scores into a plan.

1. Across Sections 1–3, which reliability gap is costing you the most right now — inconsistent citation, thin trust signals, or an unchecked surface?
2. What is the single fix that would most improve how consistently you get cited?
3. What one trust signal — a named attorney, dated proof, a verifiable outcome — could you add to your top page this week?
4. Who owns each fix, and what is a realistic go-live date?

Pick the two or three highest-priority actions and put a date on each. An audit without a deadline is a reading exercise.

The Bigger Picture

Reliability is becoming the axis AI visibility turns on. The models are getting less predictable about whom they cite, readers are getting more skeptical about what they're told, and the number of surfaces where a client can ask about you is multiplying. Firms that treat visibility as a one-time win will watch it flicker. Firms that audit it — consistently, across surfaces, with proof a human can verify — will be the ones AI keeps coming back to.

Clientory helps immigration law firms become visible to the families and individuals who need them most — turning AI search into consultations, and consultations into clients.

See how your firm shows up: APP.CLIENTORY.ORG