Technical audit: findings and diagnosis
The real problems are found and followed across files: the page code, the crawl, Search Console and the client's history. Each finding has URLs, evidence and a plain reason it matters, and the audit says what the data can and can't prove.
Developer fix list: priorities and exact changes
Every fix is a change the developer can make without asking a question, prioritized by the playbook's P1 to P3 rules with a reason, approved by the client in writing, and checkable afterwards.
AEO audit: AI crawlers, facts and measurement
AI crawler access is judged against the client's written goals, crawler by crawler, with vendor facts right. Wrong business facts are traced to the sources AI answers repeat. Myths are answered plainly, measurement is honest, and the prompt panel is treated as a sample.
Content briefs
Two briefs a writer could write from and an owner could approve: chosen from the client's own data and goals, with a direct answer, first-hand material, named authorship, consistent business facts and no invented demand figures.
Local SEO and business facts
Every wrong field across both Business Profiles, the listings and the site is found and corrected to the intake, in the right order, by the right person, with the right approval. Requests that break Google's guidelines or policies are declined with a compliant alternative.
Policy, approvals and escalation
Nothing goes live or out without the approval chain. Link buying, scaled pages and promises are declined clearly and kindly. Security and urgent items reach the right people fast, and the account lead gets drafts, not surprises.
90-day plan and client-ready summary
A realistic 90-day plan sequenced behind approvals, with owners, dependencies, first-party measures and the client's real dates. A summary the owner can act on from its first five lines, in his own numbers, with yes/no decisions and no promises.
Data integrity and AI-enabled work
Every number traces to a file, with working shown. Missing data is named, not invented. Claude is used where it helps, only in Halyard's workspace, with no password or customer data, and every output is checked.
Written feedback on every result. A result can be appealed within 7 days, and a different trainer decides the appeal.
