The audit brief
AI search readiness is a useful operational question only when it leads to verifiable work. Can a relevant system retrieve the firm's public pages? Can a reader identify who provides which services, in which jurisdictions, and under what constraints? Can a prospective client contact the firm without encountering an inaccessible form or contradictory office information? These questions are more actionable than an unexplained readiness score.
“Beyond traditional SEO” should not imply abandoning search fundamentals or buying a new class of mandatory markup. Google states that its AI search features do not require special technical optimization beyond established eligibility requirements. This guide extends the audit workflow to entity accuracy, answer evidence, and client-facing pathways; it does not claim a proprietary technical shortcut into recommendations. 1
The framework below is an original checklist methodology based on public documentation, not a certification standard. It does not establish that any particular law firm, or 360AI Legal's automated products, has completed these checks. Use it to define a bounded engineering and editorial engagement, record supporting evidence, and distinguish remediation from business outcomes. Examples are hypothetical and contain no real client information.
01. Inventory the pages and owners that matter
Start with a page inventory tied to business functions: practice pages, attorney biographies, office pages, contact routes, educational resources, and relevant policy pages. Record the intended audience and conversion path for each template. A large crawl export is not an audit conclusion; it becomes useful when someone can explain why a failure matters to the firm's services and prospective clients.
Include representative pages from every important template and deliberately sample exceptions. A location page built by a different agency may behave differently from the homepage. Older PDF resources may carry outdated contact details. A translated page may describe services that the current intake team cannot support. Identify who owns each fact before asking an engineer to make it consistent everywhere.
Create an issue register with the affected URL, observed behavior, evidence date, business impact, proposed change, responsible owner, and verification method. Separate access blockers from content ambiguities and optional improvements. Do not bundle every finding into an urgent package. The first delivery should make the scope reviewable, including items that require attorney approval or third-party access and therefore cannot be promised as immediate fixes.
| Check | Evidence to retain | Acceptance boundary |
|---|---|---|
| Retrieval | Response and applicable access controls | Public page can be fetched |
| Firm identity | Approved factual source and page comparison | Supported facts agree |
| Contact path | Authorized test and delivery confirmation | Inquiry reaches the agreed destination |
Original editorial framework. This exhibit contains no observed client results, performance statistics, or market estimates.
02. Test retrieval, redirects, and crawl controls
Fetch the public URL and inspect its status, redirect chain, headers, and returned document. A visitor-visible page can still be blocked by an edge firewall, authentication challenge, geographic rule, or rate limit. Check approved logs and hosting controls before assuming the application is the only source of a problem. Use authorized diagnostic requests; do not weaken security controls broadly to accommodate a crawler.
Read robots directives by hostname and path. OpenAI documents separate controls for its search crawler and training crawler. Treat those as separate policy choices, and confirm the firm's preference before changing them. A robots file is neither evidence of actual successful retrieval nor a promise of future inclusion. Verify the resulting request behavior after an approved change. 2
For Google, assess index eligibility and snippet controls alongside crawl access. A blocked page and a page excluded from indexing are not the same condition. Preserve evidence of the actual control encountered instead of writing “AI blocked” as a catch-all diagnosis. Where multiple domains or a website migration are involved, inspect the production hostname rather than assuming that staging behavior proves public availability.
03. Compare delivered HTML with the rendered page
A browser screenshot demonstrates presentation; it does not show what every retrieval system receives. Compare the initial HTML with the browser-rendered document for the page title, primary heading, practice description, attorney identity, navigation, and contact details. Note any content dependent on delayed scripts, interaction, personalization, or a third-party widget.
Google documents a JavaScript processing pipeline and practical constraints around rendering and discovery. That documentation is not evidence that every AI service renders a site the same way. Prefer resilient delivery of essential public information rather than assuming a system will execute an elaborate interaction before it can understand the page. Server-rendered or static core content is a design option to evaluate, not a requirement to replace a functioning architecture. 3
Test navigation as actual links, not just visual controls. Confirm that important destinations are discoverable through ordinary anchors and that error pages return appropriate responses. Review content hidden behind accordions for accessibility and retrieval consequences. For a law firm, the goal is not to expose private information; it is to make already-approved public service facts available without unnecessary technical barriers. Record what was tested and what remains uncertain.
04. Establish a reliable firm and attorney identity
Build a small factual register for the firm's public identity: approved name, relevant offices, contact methods, practice descriptions, attorney affiliations, and supported languages. Confirm facts with the appropriate internal owner. Resolve differences between marketing copy, biographies, and intake operations before propagating any preferred version into structured data or external profiles.
Use distinct pages where distinct entities need explanation. An office, a lawyer, and a legal service are related but not interchangeable. A practice page should explain the service and relevant limitations; an attorney biography should establish that person's role and qualifications without implying that every listed lawyer handles every matter. Multi-office firms should avoid presenting a mailing address as evidence that a service is delivered locally.
The audit should produce a correction plan, not an invented credential. If a specialization, award, result, or jurisdictional claim lacks a confirmed source, mark it for editorial review. The ABA's Model Rule 7.1 addresses misleading communications about legal services; applicable jurisdictional rules and the firm's reviewing counsel govern the actual publication decision. This is a marketing-quality control recommendation, not legal advice or a claim that the model rule applies identically everywhere. 4
05. Make content evidence clear and useful
Read priority pages as a prospective client would. Does the page explain what the service covers, who may need it, where the firm operates, and how to request an assessment? Does it distinguish general educational information from advice about a person's matter? Does a claim have a clear source, owner, and review date where one is needed?
For a hypothetical employment practice, a page about workplace retaliation should not collapse every employment dispute into one intake promise. It can explain the kinds of issues the team evaluates and the information an initial conversation may require, subject to attorney review. Avoid adding unsupported deadlines, eligibility rules, or outcome predictions merely to make a page appear comprehensive. Depth is useful when it reduces ambiguity, not when it increases unreviewed legal assertions.
Map meaningful gaps to existing pages before commissioning large volumes of near-duplicate content. When a statement is already clear and current, preserve it. When a public fact changes, update the canonical page and identify other places that repeat it. An AI readiness project is more maintainable when it strengthens an owned factual source than when it creates many inconsistent summaries of the same service.
06. Validate structured data against visible facts
Structured data should describe the page's supported facts, not create a second, more ambitious version of the firm's story. Google requires structured data to align with visible content and makes clear that eligibility does not guarantee a rich result. A successful validation test is evidence about syntax and supported requirements, not proof of better AI recommendations. 5
Inventory existing markup before adding more. Look for conflicting organization names, stale URLs, duplicate entities generated by multiple plugins, and properties whose values cannot be verified. Identify the source of each markup block so fixes survive the next publishing cycle. Keep identity references coherent across the templates under the firm's control.
Do not treat an unsupported “AI schema” package as an automatic audit requirement. Review proposed changes against the actual documentation for the intended consumer and the site's content. If an optional experimental file is proposed, state what it is intended to help, what evidence supports it, and what is not known. Conventional markup, readable content, and clear internal links are easier to validate than a claim that a new file guarantees ingestion by multiple proprietary systems.
07. Audit the contact path and measurement boundary
Follow the journey from a relevant article or practice page to the intended inquiry route. Test phone links on an appropriate device, form validation, required fields, error feedback, success states, and the handoff to the firm's intake system. A visual confirmation should correspond to an actual accepted submission, not a front-end animation disconnected from delivery. Obtain permission before sending test inquiries into a live workflow and label authorized tests clearly.
Inspect analytics separately from form delivery. A button click does not prove that an inquiry arrived, and a successful inquiry does not establish that the firm accepted the matter. Maintain event definitions and test records that distinguish those states. Avoid putting case narratives, names, email addresses, or other sensitive details into analytics events, URLs, or page titles. Google's Analytics guidance specifically warns against sending personally identifiable information. 6
Assess accessibility through keyboard navigation, visible focus, labels, error associations, and usable layouts at narrow widths. Automated checks can find some problems, but they cannot establish that every interaction works for every user. Assign manual testing to a named reviewer and preserve the limitation when device or assistive-technology testing is unavailable. This is particularly important when embedded scheduling or intake tools are outside the website team's direct control.
08. Prioritize, implement, and verify bounded changes
A useful prioritization matrix weighs business impact, evidence strength, effort, dependencies, and reversibility. A confirmed broken contact route deserves different treatment from a speculative content idea. Agree the intended correction and its acceptance test before editing. Keep a recoverable version of the current site, use a preview where feasible, and avoid modifying unrelated service terms during technical cleanup.
After implementation, rerun the original failing check under comparable conditions and retain the result. For a redirect issue, record the new response chain; for missing server-delivered content, save the corrected response; for a form issue, verify the approved delivery endpoint with the responsible team. A checklist marked complete without evidence is not a verification record.
Keep indexing, platform selection, and recommendation outcomes as later observations rather than immediate acceptance criteria for code changes. A technical fix can remove a barrier without producing a measurable commercial change. The report should say exactly that. This distinction protects both the firm and its agency from interpreting engineering completion as a guaranteed marketing result.
09. Handoff, limitations, and the next decision
Deliver a concise executive explanation supported by a technical appendix. The appendix should include the tested URL set, issue register, evidence captures, change list, verification results, remaining dependencies, and owners. State whether the audit was sampled or comprehensive. Describe inaccessible systems explicitly; do not imply that a website review established the quality of a separate CRM or intake operation.
This framework does not certify compliance, predict AI ranking, or substitute for attorney review. It uses current public documentation to organize practical checks and adds editorial recommendations for law-firm workflows. Vendor interfaces and crawler behavior may change. Recheck documentation before implementation, especially when changing access controls or making claims about a platform's requirements.
The next investment should follow the evidence. Fix confirmed access and contact failures first, reconcile public facts with authorized owners, and then establish a separate visibility benchmark. That sequence creates a defensible foundation for growth work without selling a technical audit as a completed client-acquisition program. A well-run handoff leaves the firm with a maintainable system of checks, not a mysterious score it must keep paying to interpret.
Sources & references
Official documentation reviewed November 13, 2026. Platform features and policies can change. The linked sources support the cited facts; proposed workflows are editorial analysis.
- Google Search Central: AI features and your websitehttps://developers.google.com/search/docs/appearance/ai-features
- OpenAI: Overview of OpenAI Crawlershttps://developers.openai.com/api/docs/bots
- Google Search Central: JavaScript SEO basicshttps://developers.google.com/search/docs/crawling-indexing/javascript/javascript-seo-basics
- ABA Model Rule 7.1: Communications Concerning a Lawyer?s Serviceshttps://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_7_1_communication_concerning_a_lawyer_s_services/
- Google Search Central: General structured data guidelineshttps://developers.google.com/search/docs/appearance/structured-data/sd-policies
- Google Analytics: Avoid sending personally identifiable informationhttps://support.google.com/analytics/answer/6366371


