One week, five steps
Step one — schema markup on every important page
Day one and two of the audit. Inventory the most-visited or most-revenue-driving pages. Add Schema.org JSON-LD blocks to each, matched to the page type: Product on product pages, Article on blog posts, FAQPage on FAQ sections, HowTo on procedures, LocalBusiness or Organization on the home page, BreadcrumbList everywhere. Validate every block in Google's Rich Results Test, the Schema.org validator, or one of the tools in the SEO analysis directory before deploying. This step produces one of the larger visibility improvements of the audit for traditional search and rich results. It remains a cornerstone even though Google's May 2026 guidance notes that structured data is not strictly required for generative AI features; the rich-results pathway alone is usually worth the work. It is easy to verify objectively. The only rule is honesty: do not mark up content that is not actually what the schema claims.
Step two — rewrite key pages around the four questions
Day three. Take the same pages and edit them so the answers to the four questions — what, who, cost, how — appear in plain language inside the first screenful. Use real headings that match likely user queries. Replace one or two long paragraphs with a table or a list where the underlying content is naturally tabular or sequential. The brand voice elements remain; they just move down the page or live alongside the literal copy rather than instead of it. Where possible, add a short FAQ block at the bottom with three to seven questions a real customer would actually ask. Validate FAQ schema separately — incorrect FAQ markup is one of the most common reasons rich results disappear.
Step three — open the site to agents
Day four. Audit robots.txt and allow GPTBot, ClaudeBot, CCBot, Google-Extended, OAI-SearchBot, and ChatGPT-User explicitly, unless there is a deliberate policy reason to block one. Verify that sitemap.xml is current and that its lastmod dates are accurate. Decide whether to add an llms.txt. Some assistants honour it; Google has stated its own systems do not require it. The right call depends on which assistants matter most to the site's audience. An AGENTS.md is a low-cost win for developer-facing or SaaS sites, even though it is not a Google-search signal. Where the site has a product feed, make sure it is publicly reachable and uses standard formats. For larger or developer-facing sites, plan a more substantial step: an OpenAPI specification, an MCP server, or an NLWeb or WebMCP deployment, depending on the platform.
Steps four and five — build off-site signal and lock in freshness
Day five and the start of an ongoing routine. Audit independent review platforms and ensure the brand has up-to-date listings on the major ones for its sector. Audit press and citation history; identify two or three credible publications worth pitching this quarter. Audit NAP consistency across directories. Then commit to the freshness routine from lesson 11: a quarterly review cycle for high-value pages, an annual review for everything else, with visible last-updated dates and matching dateModified. The audit ends with a measurement baseline using the three layers from lesson 13 — agent-readability score, server-log filter, prompt testing — so the next month's progress can be compared cleanly.
Things to consider — cost, timeline, and ownership
Three practical questions decide whether the audit actually gets done. Who owns it? In small businesses the founder usually owns it; in larger organisations it sits awkwardly between marketing, IT, and product, and that is where it dies. Naming an owner with both authority and bandwidth matters more than picking the 'right' team. What does it cost? A DIY audit costs a team-week of focused work and produces meaningful improvement; a small-agency audit costs roughly two to ten thousand pounds and adds professional schema validation and ongoing measurement; a larger consultancy engagement runs higher and is rarely justified for sites under significant traffic. How long until impact? Schema and on-page changes show up in agent-readability scores within a week; in citations within four to six weeks; in attributable revenue within a quarter, if the broader content and off-site work has also been done. AI SEO is steady compounding, not a single launch.
Block out one week on the calendar. Pretend the goal is to take your site from where it is today to the level described in this course. What is the first task you do on Monday morning?