A law firm case results page earns AI citations in 2026 only when every number on it is attached to matter facts, a jurisdiction, a year, and a disclaimer placed next to the claim instead of in the footer. ABA Model Rule 7.1 bars communications that create unjustified expectations about results, Florida Bar Rule 4-7.13(b)(8) requires an explicit disclosure that a prospective client may not obtain the same or similar outcome, and the New Jersey Committee on Attorney Advertising requires the line “Results may vary depending on your particular facts and legal circumstances” beside every past performance claim. Meanwhile Google AI Overviews appeared on roughly 48% of queries as of March 2026, and Search Engine Land reported that 68% of US Google searches ended without a click in early 2026, so an undocumented wall of dollar figures is now both an ethics exposure and dead weight in retrieval.
The gap is bigger than most firms think. The 2026 Legal Consumer Choice Report from AttorneySync found that 72% of legal consumers start with Google search and 42% now use ChatGPT when researching an attorney, with Perplexity, Gemini, and Microsoft Copilot splitting the remainder. Those systems build answers from sources they can corroborate: Justia, Avvo, Martindale-Hubbell, Super Lawyers, Lawyers.com, Google Business Profile, court dockets on PACER, and trade coverage in Law360. A page that says “$4.2 million” and nothing else corroborates against none of them.
Why do AI engines ignore most law firm case results pages?
Because a bare dollar amount is an unverifiable numeric claim, and retrieval systems discount those first. Research published in the ACM FAccT 2025 proceedings on source-cited LLM search found that generative engines routinely surface claims with no grounding in the documents they cite, which is exactly the failure mode their retrieval layers are tuned to avoid. When ChatGPT or Google AI Overviews weighs “$4.2 million settlement” from a firm’s own site against a docket entry, a VerdictSearch summary, or a Law360 story, the self-reported number loses.
The second reason is the format. Most case results pages are a grid of 40 tiles, each holding a figure and a two word label like “Truck Accident.” There is no sentence to extract, no entity to anchor, and no query that maps to the page. Engines quote sentences, not table cells.
The third reason is duplication and staleness. The same 12 results usually appear on the home page, every practice area page, and the results page, with no dates. A firm publishing a 2011 verdict beside a 2025 settlement, with neither labeled, gives an engine no way to judge currency, and currency is a live ranking factor when the query implies “who is winning these cases now.”
What do bar rules actually require when you publish verdicts and settlements?
Publishing past results is allowed in every US jurisdiction, but almost all of them require a qualification placed with the claim. ABA Model Rule 7.1 sets the baseline: a communication is misleading if it omits a fact necessary to keep the statement from being materially misleading, and the comment history flags damage awards as the classic example of information that creates unjustified expectations when stated without context.
Five states demand extra care. Florida allows past results under Rule 4-7.13 but treats an incomplete description as inherently misleading and requires the “may not obtain the same or similar results” disclosure. New York requires “Prior results do not guarantee a similar outcome” and an “Attorney Advertising” label on the home page of a firm website under 22 NYCRR 1200. California applies Rule 7.1 with the requirement that disclaimers be clear and conspicuous rather than buried in fine print. Texas Rule 7.02 requires prominent qualifications on any communication describing the amount of a damage award or a lawyer’s record of favorable verdicts. New Jersey requires the specific “Results may vary” sentence on every mention of results, including bio narratives and representative matter lists, not just the results page.
Firms licensed in more than one state should write to the strictest applicable rule and use one disclaimer everywhere. Maintaining separate compliant and non compliant versions of the same content is how a firm ends up with the wrong one indexed.
Not sure whether AI assistants can verify a single number on your results page? Get your free AI visibility audit and see exactly which claims ChatGPT, Perplexity, and Google AI Overviews treat as credible.
What is the anatomy of a case result entry AI engines will quote?
Six elements, in this order, for every entry on the page. Each one exists to make a specific claim checkable, which is the same thing that makes it compliant.
1. A descriptive headline that reads like a query
Not “$4.2M Verdict.” Write “$4.2 million verdict for a warehouse worker in a forklift crush injury case, Cook County, 2024.” That single line carries the amount, the matter type, the venue, and the year, and it maps to how people search and prompt. It is also the sentence an engine will lift.
2. Matter facts a stranger can follow
Three to five sentences on what happened: the injury or dispute, who the defendant was by category (regional trucking carrier, hospital system, municipal employer), what the insurer or opposing counsel argued, and what the pre suit posture was. This is the context bar rules ask for and the substance retrieval systems index.
3. Jurisdiction, court, and year
Name the court and the county or district, plus the year of resolution. This is the highest value addition most firms can make, because it converts an unverifiable number into a claim an engine can align with public records. Redact the case caption where confidentiality requires it and say so explicitly.
4. The obstacle overcome
Every result worth publishing had a problem: a comparative fault argument, a missed reporting deadline, a prior settlement offer of one tenth the eventual number, a defense expert, a jurisdictional challenge. Naming the obstacle is what separates a result from a number, and it is the part that demonstrates real experience rather than volume.
5. A plain language disclaimer beside the claim
Same font size, same block, immediately after the entry. Not gray six point type in the footer. Use the strictest wording that applies to your licensure. Placement is a compliance requirement in Florida, New Jersey, and California, and it reads to a retrieval system as candor rather than promotion.
6. A verification path
Link or cite the public record where one exists: the docket number, the reported opinion, a VerdictSearch or Jury Verdict Research entry, a Law360 or local legal publication story, or a court press release. Where the matter is confidential, state that the amount is confidential and describe the result qualitatively. A firm that publishes 20 results with 12 verifiable citations outranks a firm publishing 200 with zero.
How is a case results page different from a testimonial page?
A case results page publishes the firm’s own factual assertions about outcomes. A testimonial page publishes a client’s opinion about the firm. Different rules, different proof burden, different retrieval behavior, and they should never share a page.
Testimonials are governed by endorsement rules, including ABA Model Rule 7.2(b) on giving anything of value for a recommendation and the FTC Consumer Reviews and Testimonials Rule, and their credibility comes from third party platforms like Google Business Profile, Avvo, and Martindale-Hubbell. Case results are governed by the misleading communications rules and their credibility comes from court records. We cover the client voice side separately in law firm testimonial pages that AI engines can cite.
There is a third format worth separating too. A case result is one entry of roughly 150 words. A case study is a full narrative page of 800 to 1,500 words on a single matter, built to rank for a specific query pattern, and our law firm case study template covers that build. Best practice is to publish 20 to 40 short entries on the results hub and promote the four or five strongest into standalone case studies.
How do AI engines evaluate unverifiable dollar amounts?
They treat them as low confidence and route around them. Generative engines assemble answers from passages they can attribute, and a number with no venue, date, or corroborating source fails attribution. The practical result is that the engine cites Justia, Super Lawyers, or a Law360 article about the same verdict instead of the firm’s own page, and the firm loses the citation for its own win.
Three signals move a number from low confidence to citable. First, corroboration: the same figure appearing in a court record, a legal trade outlet, or a directory profile. Second, specificity: an amount paired with a matter type, a court, and a year is treatable as a discrete fact rather than marketing copy. Third, hedging language that matches reality, meaning the disclaimer is present and the page does not claim the outcome is typical. Engines penalize pages that read as guarantees, which is the same behavior state bars penalize.
This matters most for valuation prompts like “what is my case worth in Cook County.” Gemini and Microsoft Copilot answer those with ranges pulled from sources they can verify, and firms with documented, dated, jurisdiction tagged results are the ones supplying the numbers.
What mistakes kill a case results page?
Five, in order of how often we find them in audits.
The number wall. Forty tiles of dollar amounts with two word labels. No sentences, no dates, no venues, nothing extractable. This is the default output of most law firm website templates and it is unciteable by construction.
Footer disclaimers. A single “past results do not guarantee future outcomes” line at the bottom of a page holding 40 claims fails the placement requirements in Florida, New Jersey, and California, and it fails the plain reading test a regulator applies.
Undated results. A 2009 verdict displayed identically to a 2025 settlement. Add the year to every entry, including the old ones, and let the engine decide what is current.
Inflated or aggregated totals. “Over $500 million recovered” with no breakdown is the single riskiest sentence on most firm sites. It is unverifiable, it invites the unjustified expectations analysis under Rule 7.1, and no engine will repeat it.
Schema that does not match the page. Do not mark case results with Review or AggregateRating markup, which describes opinions, not outcomes. Use Article or BlogPosting on standalone case studies, FAQPage on the questions section, and LegalService plus Attorney on the firm and lawyer entities. Our legal schema markup guide covers the exact property set and the validation workflow.
FAQ
Do I have to publish dollar amounts at all? No. Qualitative results are compliant and often more citable: “case dismissed on summary judgment,” “charges reduced from felony to misdemeanor,” “custody modified in favor of the client.” These work well in practice areas where amounts are confidential or where bar scrutiny is heavy, such as Florida and New Jersey. A page of 25 outcome descriptions with venues and years will outperform a page of 25 naked figures in both compliance review and AI retrieval.
Where exactly should the disclaimer go? Directly beneath each result entry, in the same font size and color as the body text. Florida Bar Rule 4-7.13, the California clear and conspicuous standard, and the New Jersey Committee on Attorney Advertising all target placement, not just presence. A footer line under 40 claims is the version that gets cited in grievances. Repeat the disclaimer once per entry even when it feels redundant.
Can I publish confidential settlements? Only in the form your settlement agreement permits. Many agreements bar disclosure of the amount but allow describing the matter and the fact of resolution. Publish “confidential settlement” plus the matter type, venue, and year, and say why the amount is withheld. That reads as candor to both a bar reviewer and an AI engine, and it is far safer than an amount you cannot defend.
How many results should the page hold? Twenty to forty entries, segmented by practice area, with each segment linked from its matching practice area page. Volume without documentation hurts. Three to five relevant results embedded on each practice area page, with the full set on the hub, beats one 200 item page that no query maps to.
Will this page rank if nobody clicks through anymore? Rankings and citations are now separate outcomes. With 68% of US searches ending without a click in early 2026, the win is being the source the AI Overview or ChatGPT answer names, which drives brand recall and direct search for the firm name. Documented results pages get cited by name in AI Overviews and Perplexity answers far more often than number walls, which get skipped entirely.
What if a result was reported by Law360 or a local outlet? Link to it. External corroboration from Law360, Bloomberg Law, ABA Journal, or a state legal newspaper is the strongest verification signal a case result can carry, stronger than a directory profile and stronger than the firm’s own account. Add the citation directly to the entry with the publication name and date, and keep a copy on file in case the link rots.
The close
Most firms treat the case results page as a trophy case. It is closer to an evidence file. The version that works in 2026 reads like something a clerk could check: amount, matter, court, year, obstacle, disclaimer, source. That format satisfies Rule 7.1 in every jurisdiction, survives a Florida or New Jersey ethics review, and gives ChatGPT, Perplexity, and Google AI Overviews something they can actually repeat with your firm’s name attached.
The firms that keep the number wall will keep watching Justia, Avvo, and Law360 get cited for verdicts those firms won.
Your best results are probably being credited to someone else’s page. Get your free AI visibility audit and find out which of your wins the AI engines can trace back to your firm.
Tagged