An SEO audit should reduce uncertainty about a defined website problem and produce changes your team can implement. TANTU AI agrees the page groups, market context, access and depth before reviewing technical delivery, content and the inquiry path. Findings distinguish confirmed defects from hypotheses and state how much of the site was inspected. The handover identifies affected URLs or templates, evidence, priority and a verification method. A complete audit is separately scoped from the $990 AI search diagnostic, and implementation responsibilities are explicit.
Define what will be inspected
Confirm the domains, templates, target markets and key page groups before collection. Identify whether the task is a routine performance review, an unexplained decline, an ecommerce issue or preparation for a site change. Each requires a different depth of investigation.
Access to Search Console, analytics, the CMS and server information can improve the evidence, but only authorized access is used. If an account is unavailable, the report states the resulting limitation instead of implying a complete historical diagnosis.
Review the layers in sequence
| Layer | Inspect | Example finding |
|---|---|---|
| Access and delivery | Response codes, redirects, access controls and readable HTML | A priority page is unavailable to the intended crawler. |
| Index and canonical intent | Preferred URLs, directives and duplicate patterns | Conflicting versions are being promoted. |
| Architecture and content | Navigation, page purpose, evidence and overlap | Several pages compete to answer the same buying question. |
| Search performance | Available query, page and time-series data | A decline is concentrated in one template or demand segment. |
| Conversion | Offer clarity, actions and inquiry handoff | Relevant visitors reach a page without a useful next step. |
The examples illustrate a method. Your findings must come from your site and the actual evidence collected.
Make every recommendation assignable
- Affected URL or shared template.
- Observed behavior and supporting record.
- Commercial relevance and estimated scope of impact.
- Proposed change, dependencies and responsible owner.
- Acceptance check and release-review method.
A large issue list is not a priority plan. Group common template problems and explain which repairs affect the most important journeys. A specific broken quotation route can deserve attention before a minor metadata refinement.
Distinguish the SEO audit from the $990 AI diagnostic
The AI search diagnostic samples 15 questions in one market and language across three agreed platforms. A full SEO audit can require a crawl inventory, template analysis and historical performance evidence. They are different scopes and are not automatically included in one another.
Implementation, large migrations, development work and third-party tools are separately confirmed. An audit may produce a well-defined brief for your developers rather than a completed code change.
Use a review meeting to make decisions
Walk through the highest-priority evidence with the commercial and implementation owners. Confirm which findings can be acted on immediately, which require more information and which should be deferred. Preserve the original records so the retest can examine the same issue.
A repaired technical obstacle is not proof of a future ranking. Track completion, index behavior and business performance as separate stages. This keeps the audit useful after the initial presentation.
Prepare a focused request
Share the website, priority products, target countries, timing and known changes around the problem. Identify any existing technical audit or migration plan so the work can build on it. We then confirm the scope, required access, deliverables and fee before starting.
Select a sample that represents how the site works
Group URLs by the systems and templates that generate them, then consider their commercial importance. A useful sample might cover category pages, product pages, articles, language variants and parameter-driven views when those features exist. Include known problem URLs and comparable pages that appear to work, so the audit can examine the difference rather than collecting only failures.
State the inventory source and sampling limits. Testing several pages from one template supports a narrower conclusion than inspecting all relevant templates. Expand the review when findings vary within a group, when a defect appears tied to a particular condition or when the commercial risk justifies more coverage. For an illustrative catalog, failures limited to filtered URLs call for investigation of that URL pattern; they do not automatically establish that every product page is affected. Any larger crawl or additional depth must fit the agreed scope.
Ask what evidence would disprove the finding
A strong issue record contains the observation, proposed explanation and a test that could challenge it. If a traffic decline is attributed to a release, inspect timing, affected pages and comparable unchanged pages where data permits. If demand or offer changes could explain part of the pattern, record them. Do not convert correlation into a confirmed technical cause merely because a release happened nearby.
Assign a confidence description based on the available evidence and keep unverified explanations separate from reproducible defects. The review meeting can then prioritize confirmed high-impact repairs while deciding whether uncertain findings warrant more investigation. Acceptance of the audit means that the scope, limitations, records and actions are understandable and assignable. Acceptance of a repair is a later check against the original observed behavior. Search recovery and commercial results are evaluated afterward and remain outside any guarantee that the audit will produce a particular ranking.
Common questions
Is an audit included in every managed plan?
Only the specifically agreed review work is included. A comprehensive standalone audit or migration review needs its own scope.
Can you investigate traffic loss without assuming AI caused it?
Yes. We examine timing, affected pages, technical changes, demand and available search data before attributing the decline to a particular cause.