A 30-day implementation roadmap for AI search discoverability
A sequenced four-week plan for an SMB that has decided to address AI search discoverability seriously. Written for the operator delivering the work, with sufficient detail for an owner-manager to commission and supervise it competently.
The intent of this roadmap is to set out, in week-by-week detail, what a competent operator delivering the first foundational engagement on an SMB website should do, in what order, and to what standard. The sequencing matters. A number of items depend on others being in place first, and getting the order wrong adds cost without proportionate benefit.
The roadmap is calibrated to a typical professional services SMB with an existing website that hasn't previously addressed AI search. It is approximately the scope of the Retrofit package: a two-week implementation engagement preceded by a one-week audit and followed by a one-week review and refinement window.
The roadmap is also a useful template for an owner-manager commissioning the work from a third party. Each phase is self-contained enough that progress can be reviewed, and the deliverables for each phase are explicit.
Week 1: audit and baseline
The first week is exclusively diagnostic. No structural changes are made to the site until the audit is complete.
Day 1 to 2: structured audit. Run the audit framework in full. The output is a written assessment across the six categories, with each item scored.
Day 3: external corroboration audit. Pull the firm's LinkedIn, Companies House, professional body listings, review profiles, and any other third-party presence. Assess consistency with the website. Note any contradictions for remediation.
Day 4: test-query work. Run a curated set of 20 to 30 queries against ChatGPT, Claude, Perplexity, Microsoft Copilot, Google's AI Overviews and Gemini. The queries should be the ones the firm wants to be cited on. Note which assistants cite the firm, which cite competitors, and which decline to recommend anyone. The baseline is the reference against which the post-engagement work will be measured.
Day 5: written report and remediation plan. Synthesise the audit into a single document: the firm's starting position, the priority remediation list, the effort estimates for each, and the recommended sequencing.
Deliverable at end of week 1. A written audit report (8 to 12 pages), a prioritised remediation list, and a baseline test-query result set.
Week 2: foundational structured data
The first week of implementation work addresses the structural items. These are the foundational layer on which everything else depends.
Day 6 to 7: schema.org Organization and WebSite. Add or correct the site-wide markup. This includes Organization (or ProfessionalService, or LocalBusiness), the principal Person entities, and WebSite at the site root. The markup should be implemented in the layout template so it renders on every page; for template platforms, the appropriate per-platform mechanism is used. (See the practitioner's guide to Schema.org for the specifics.)
Day 8: per-page Service and BreadcrumbList. Each substantive service page gets its own Service markup. Every non-root page gets BreadcrumbList markup.
Day 9: validation. Run every page with structured data through the Google Rich Results Test and the Schema.org Validator. Fix any validation errors before proceeding.
Day 10: llms.txt. Draft and publish the llms.txt at the domain root. (See Understanding llms.txt for the structure and contents.) The draft should be reviewed by the firm's principal before publication.
Deliverable at end of week 2. All foundational structured data in place, validating cleanly. llms.txt published.
Week 3: content density and FAQ markup
The third week addresses the content-layer interventions. These aren't a full content programme (that's the ongoing investment) but the targeted interventions that make the foundational work pay off.
Day 11 to 12: FAQ schema on service pages. For each substantive service page, draft a FAQ section of 6 to 10 questions, with substantive specific answers. Mark up with FAQPage schema. (See FAQ schema.) The questions should be the ones the firm's principal genuinely hears on discovery calls.
Day 13: principal biography page. Draft or refine the principal's biography to be substantive and consistent with LinkedIn and professional body listings. Mark up with Person schema.
Day 14: content review on existing pages. Read every page on the site. Flag any pages whose content is thin (under 400 words), generic, or contradicts other claims on the site. Where possible, expand or rewrite. Where a full rewrite is out of scope for the engagement, flag for the follow-on content programme.
Day 15: canonical-answer pieces. Where the firm has positioning questions that aren't yet substantively answered anywhere on the site, draft canonical-answer pieces. Typically 800 to 1,500 words each. These become permanent reference pages, not blog posts.
Deliverable at end of week 3. FAQ schema on all substantive pages. Principal biographies in place. Content density addressed on existing pages, with any out-of-scope items flagged for the follow-on programme.
Week 4: classical hygiene and review
The final week addresses the classical SEO hygiene that the generative search layer also depends on, and reviews the work against the baseline.
Day 16: robots.txt and sitemap. Confirm robots.txt explicitly allows the AI crawlers (GPTBot, ClaudeBot, PerplexityBot, CCBot, Google-Extended, Applebot, Amazonbot, Bytespider, cohere-ai). Confirm sitemap.xml is current and properly structured. Submit to Google Search Console and Bing Webmaster Tools.
Day 17: canonical URLs and meta. Check canonical tags on every page. Verify meta titles and descriptions are unique, substantive, and accurate. Fix any duplicate or missing canonicals.
Day 18: page speed and accessibility. Run Lighthouse on the firm's primary pages. Address any accessibility violations and any Core Web Vitals failures.
Day 19: test-query re-baseline. Run the same 20 to 30 queries from week 1 against the same six assistants. Note changes. The full effect of the engagement is unlikely to appear within the engagement window (the search interfaces typically take 4 to 8 weeks to fully re-index) but early signals are usually visible.
Day 20: handover. Written summary of the work, with before/after structured data, before/after test-query results, the canonical-answer list, the content programme recommendations, and the maintenance commitments going forward.
Deliverable at end of week 4. Engagement complete. Site is in the AI-search-ready state. Handover document is in the firm's hands.
Beyond the engagement
The foundational work is, by design, finite. Once it's done, the firm's ongoing commitment is the content investment (the sustained publishing cadence that the Authority package productises) and a periodic maintenance review.
The recommended maintenance cadence:
- Monthly. Test-query check against the tracked query set. Note any changes.
- Quarterly. Review the
llms.txt. Update if the firm's positioning or service lines have materially changed. - Annually. Full structured-data audit. Confirm the schema.org work is still validating, that AI crawler allowances are still appropriate, that the principal biographies are still current.
The Authority Maintenance Retainer productises this ongoing cadence for firms that prefer to delegate it.
A note on expectations
The work described in this roadmap reliably moves a firm from "structurally invisible" to "competitively visible" in AI search. It doesn't, and can't, guarantee that the firm will be the top recommendation for every query in its category. The selection of which firm is recommended depends on factors (principal-level reputation, third-party citations, sustained content investment) that no four-week engagement can fully address.
The work also takes time to fully propagate. The major generative search interfaces typically take four to eight weeks to fully re-index the firm's pages after substantial structural changes. The firm should expect to see partial results within the engagement window and the full effect within two to three months.
The state of AI-driven search in 2026 note covers the broader market context for the work.