Ask Questions That Make the SEO Plan Executable

Begin discovery by defining the client's publication boundaries. Ask which laws, professional rules, contracts, brand policies, platform requirements, privacy obligations, and internal standards affect website content.

Do not assume that the marketing contact has final authority. Identify the person or function that can approve factual claims, professional guidance, comparisons, testimonials, pricing statements, outcomes, service availability, and disclaimers.

Request existing materials that show the rules in practice: a compliance guide, advertising policy, approved disclaimer set, claims matrix, brand standard, privacy requirement, or prior review comments.

If the client has no central record, document that absence as a project risk rather than filling the gap with agency assumptions. Ask what evidence must accompany a claim and where that evidence is stored.

A statement about a credential, award, service, result, or regulatory status should not be published merely because someone remembers it.

The workflow should state who drafts, who checks sources, who reviews subject accuracy, who performs legal or compliance review when required, who approves publication, and who can stop a release. Ask for the typical review time and the escalation route when a deadline is missed.

A single word such as 'guaranteed' or 'best' may be unacceptable in a specific context, so the agency needs the actual boundary rather than a general instruction to be careful.

Discuss rejected work as well as approved work. A rejected page or campaign can reveal the difference between a factual correction, a prohibited claim, a missing disclaimer, an unsupported comparison, and a disagreement about tone. That evidence helps the team create briefs that are publishable before drafting begins.

The output should be a content governance record containing approved and prohibited claim types, required evidence, mandatory language, responsible reviewers, turnaround expectations, escalation contacts, and unresolved issues.

The owner is the client representative with authority to confirm the rules. Measurement at this stage is operational: approval time, rejection reasons, revision count, and the proportion of planned work that can proceed without an unresolved compliance dependency.

The next discovery task is to identify the people and organizational evidence that can support accurate content. Ask who holds the relevant experience, who can explain the service, who can validate technical or professional details, and who is willing to be publicly associated with the material.

The top three experts named by leadership may not be the only useful sources. Experienced practitioners, product specialists, operations staff, researchers, customer support teams, and sales personnel may hold different parts of the knowledge needed for a complete page.

Build an evidence inventory rather than a list of impressive labels. For each person or organization, record the role, subject coverage, credentials, public biography, approved profiles, publications, presentations, memberships, research, internal documents, and availability.

Every credential or recognition proposed for public use should have a source the client can verify. The existence of structured data or an author page does not prove expertise by itself. Those elements should accurately describe information that the client is prepared to publish and maintain.

Ask whether real names and biographies may be used, whether contributors can be quoted, and whether the client prefers review without public attribution. Anonymous or brand-authored content is not automatically unsuitable, but the decision affects what evidence can be shown and how responsibility is communicated.

The agency should not claim that naming an author guarantees rankings. The useful objective is to make authorship, review, evidence, and accountability clear to readers when appropriate.

Design a knowledge extraction process that respects limited schedules. A 15-minute interview can be useful when it has a focused brief, prepared questions, recording permission, a defined output, and a review owner.

It should not become a justification for creating claims the expert did not make. Ask which internal reports, training materials, memos, case records, policies, or research can be used and what must remain confidential.

The output should be an expert and evidence register that links topics to approved sources, interview availability, publication permissions, reviewer roles, and maintenance responsibilities. The content owner uses it to decide whether a proposed page has enough support to proceed.

Measurement includes source coverage, interview completion, approval latency, unsupported claim removals, and whether published biographies and credentials remain current.

Technical discovery must test implementation capacity, not merely identify defects. Ask which platform runs the site, who maintains it, where it is hosted, how releases are approved, whether a staging environment exists, and which teams control templates, redirects, structured data, analytics, scripts, forms, and server settings. Record the actual access available to the agency and the process for requesting anything it cannot change directly.

A site can look polished while carrying architecture, crawl, indexation, performance, template, or integration constraints that alter the strategy. Ask about legacy code, duplicated templates, active subdomains, microsites, migrations, language versions, JavaScript rendering, customer portals, booking systems, CRM embeds, security controls, and vendor-managed components.

Some areas may be no-go zones because of privacy, security, contract, or operational requirements. Those limits need to be known before recommendations depend on them.

Map the request-to-production path. If every meta tag change requires a Jira ticket and a two-week sprint, the roadmap must reflect that. Identify the technical point of contact, the approver, the developer, the tester, the release window, and the rollback owner.

Ask what documentation is required, how urgent fixes are handled, and what happens when marketing and IT disagree on priority.

Structured data should be discussed as an implementation and accuracy question, not a guaranteed ranking tactic. Can the platform support custom Schema.org markup? Can the client maintain it when content changes?

Who validates that it matches visible information? The same approach applies to canonicals, redirects, internal links, page templates, mobile behavior, and performance work. A recommendation is not complete until the responsible team can implement and test it.

The scope may need a technical foundation phase before large content investment. That tradeoff should be explicit: delaying publication can be justified when templates, indexation, tracking, or release controls would otherwise neutralize the work.

The output is a technical dependency register with issue, business effect, access owner, implementation owner, approval path, effort, risk, release stage, rollback plan, and test result. Progress is measured by implemented and verified changes, not recommendations delivered.

Discovery should connect search demand to the business the client wants and is able to serve. Ask which services are strategically important, which are profitable, which create repeat or referral value, which have available capacity, and which should not be promoted.

A high-volume topic may be a poor priority when it attracts unsuitable locations, budgets, jurisdictions, use cases, or customer types.

One 'qualified lead' can matter more than 10,000 informational visits in a high-value service, but the client must define qualified. Ask for the attributes that make an inquiry useful and the conditions that make it unsuitable.

These may include service need, location, legal eligibility, business size, urgency, budget, decision authority, technical fit, or required documentation. Also ask who validates lead quality and where that result is recorded.

Review the sales process. What questions appear on the first call? Which misconceptions must be corrected? What information causes a prospect to proceed or withdraw? Where do customers hesitate? Which pages, documents, calculators, comparisons, or explanations support the decision?

Sales and customer service language can reveal useful content needs, but recordings and customer data must be handled with appropriate permission and privacy controls.

Historical traffic growth should be interpreted cautiously. A previous report may show 200 percent growth while revenue remained flat, but that observation does not establish that traffic caused the revenue outcome.

Ask what changed in service mix, tracking, market demand, advertising, sales capacity, or qualification during the same period. Separate visibility, inquiry volume, qualified demand, accepted opportunities, and closed business.

The client may know the current lead-to-close ratio for organic traffic, or the data may be incomplete. Record the source, definition, period, and limitations rather than treating an unsupported ratio as a fact.

Avoid turning irrelevant organic queries into a so-called negative keyword list as though organic search can be excluded like paid advertising. Instead, use exclusion criteria to decide which topics not to target, which pages need clearer qualification, and which reports should segment unsuitable demand.

The output is a commercial intent record linking services and audience needs to query themes, decision stages, qualification rules, exclusions, page types, conversion actions, and owners. The strategist uses it to prioritize work.

Measurement should distinguish visibility from qualified inquiries and should document data gaps rather than filling them with assumptions.

AI-related discovery should begin with the client's information governance, not a promise to optimize for an undocumented mechanism. SGE was a historical experimental name. For current Google products, discuss Google AI Overviews and other Google AI features that appear for relevant searches. Other services, including ChatGPT or Claude, may also be observed when the client considers them commercially useful.

Ask which branded, service, problem, comparison, local, and reputation queries should be tested. Record the date, product, account or location context where relevant, response type, source citations, business mentions, factual errors, omissions, and uncertainty.

The classification must distinguish a source citation, a brand mention, a general category description, and no appearance. Do not translate an observed mention into a claim that the system recommended or selected the business unless the recorded response actually used that classification.

Create a verified brand fact record. Ask which legal name, services, locations, credentials, people, policies, pricing details, process descriptions, statistics, and public profiles are approved and supported.

A list of consistently maintained facts can reduce internal contradiction, but it does not guarantee an AI citation. Structured data can describe visible information in machine-readable form, yet it should not be presented as a special requirement for AI inclusion.

Ask which third-party publications, professional databases, directories, or registries the client considers reliable. The agency should verify the relevance, accuracy, and ownership of those profiles rather than treating every directory as authoritative.

If the client is incorrectly represented, document the correction process and evidence. If the brand name conflicts with another entity, record the disambiguating facts that can be published honestly.

Content format is also a governance choice. An answer-first structure may help readers reach a clear response, but it is not a documented guarantee that a model will quote the page. Transparent pricing or process information may be useful when accurate and commercially appropriate, but the client should decide what can be disclosed.

The number one goal is not to feed a model. It is to publish useful, accurate information that the business can support and maintain.

The output is an AI visibility observation plan and brand fact register. The reporting owner tracks agreed queries, exact recorded classifications, citations, errors, corrections, and related qualified actions without claiming causation.

The client should also define how the brand ought to be described in a 2-sentence AI summary, then verify that every element of that preferred description is supportable.

Most clients coming to you have worked with an SEO agency before, and many of them have 'agency trauma.' I ask: 'What specifically did your last SEO partner do that frustrated you the most?' and 'Can you share the reports they gave you?' This isn't about bad-mouthing competitors: it's about diagnosing the failure.

What I've found is that many failures aren't due to bad SEO, but bad communication or a lack of alignment. However, sometimes it is bad SEO. You need to ask: 'Have you ever received a manual action notice in Google Search Console?' and 'Have you ever bought links or used automated content generators in the past?' You need to know if you are walking into a 'clean' site or one that has a hidden history of spam.

In practice, I also ask: 'What is the one thing you were promised by a previous agency that never happened?' This helps me manage expectations. If they were promised #1 rankings for a massive term in 30 days, I need to reset their reality immediately.

We build a documented, measurable system, and that takes time. Understanding their past disappointments allows me to build a partnership based on evidence over promises.

Generic discovery guides treat goals and competitors as complete answers. In practice, a stated goal can conflict with capacity, permissions, commercial priorities, or technical reality. A client may request broad visibility while approving only a narrow set of claims.

The marketing team may want frequent publishing while the authorized expert has no review time. The business may ask for leads from markets where it cannot provide the service. The website may require a development process that cannot support the proposed release schedule.

The missing questions concern evidence and execution. Who is authorized to approve a professional statement? Which source supports each credential, service description, result, or comparison? Who owns the CMS, analytics, profiles, call tracking, and customer data?

How does a recommendation move from request to production? Which leads are commercially useful, and who validates their quality? What happened under previous agencies or internal teams?

Modern discovery must also distinguish the people and organizations behind the content from the words on the page. The useful question is not only what topic the client wants to cover. It is whose knowledge supports the page, what evidence can be published, what review is required, and whether the business wants that person publicly associated with the material.

That information allows the agency to design a realistic content, technical, and measurement system rather than a generic keyword plan.

Earlier in my work, I treated discovery as a chance to demonstrate knowledge. The more useful purpose is to uncover the operating conditions that decide whether a recommendation can move from a document into production.

A sound plan can stall because the authorized expert is unavailable, the compliance owner was never identified, the IT queue is longer than expected, the analytics events are unreliable, or the agency cannot access the system it is expected to change.

The questions about internal governance and workflow are therefore not softer than technical questions. They identify who can say yes, who can say no, what evidence is required, how work is released, and where it is likely to stop.

I once spent months planning around a healthcare site before learning that the internal IT team would not approve the requested Schema implementation. The lesson was not that a plugin should have been forced into the stack. It was that platform policy and implementation ownership belonged in discovery.

A dependable engagement begins with a documented route from input to decision, production, approval, evidence, and measurement. Hundreds of hours can be wasted when that route is left implicit. The strategy matters, but its value depends on the client's capacity to support and implement it.

Day 1-5

Send the governance, compliance, expertise, and evidence questions to the primary stakeholders.

Outcome: Identification of publication boundaries, approval owners, required evidence, and available subject matter experts.

Day 6-12

Conduct 15-minute interviews with selected internal experts and the technical lead using focused briefs.

Outcome: A technical dependency inventory and an approved record of useful brand, service, and subject facts.

Day 13-20

Review Search Console, prior reports, change records, access, links, content history, and measurement definitions.

Outcome: A reconciled baseline of known assets, historical risks, data limitations, and unresolved questions.

Day 21-30

Present the proposed visibility roadmap with owners, dependencies, approval stages, deliverables, and measurement rules.

Outcome: A signed-off strategy with realistic stage timing, documented workflows, and explicit tradeoffs.