SEO Specialist interview questions

I would model reachable demand by query cluster rather than quote total keyword volume as the opportunity.

• I group queries by intent and landing-page type, then remove overlap so variants are not counted as separate demand.
• For each cluster I apply a realistic rank distribution, SERP-specific click-through rate, and the share our authority could plausibly reach.
• I convert the resulting visits with current or benchmark conversion and value ranges, with explicit low, base, and high assumptions.

I treat search share of voice as a directional index and triangulate it with first-party demand data.

• I weight tracked queries by demand and expected clicks, then compare our visibility with the competitors that actually appear in each cluster.
• I segment brand and non-brand because brand demand can make a weak acquisition position look healthy.
• I validate movement against Search Console impressions and clicks, while stating that neither source represents the whole market.

SEO mainly captures expressed demand, while demand creation makes more people search for the problem, category, or brand later.

• Comparison and solution pages meet existing intent and can be evaluated against known query demand.
• Research, tools, and original stories may create awareness before relevant search volume exists, so keyword volume understates their value.
• I measure the two differently: non-brand acquisition for capture, and branded search growth plus assisted journeys for creation.

I would publish scenarios with visible assumptions, not one promised traffic number.

• The model separates demand, indexable coverage, rank distribution, click-through rate, conversion, and implementation timing so each assumption can change independently.
• Low, base, and high cases use ranges grounded in current performance and comparable page groups rather than arbitrary multipliers.
• I update the forecast as pages index and early impressions arrive, and I show forecast error instead of rewriting the original baseline.

I rank initiatives by expected incremental value adjusted for confidence, cost, time to impact, and downside.

• A template fix that affects valuable indexed pages usually outranks publishing more pages into a broken architecture.
• I reduce confidence where the causal link or reachable demand is weak instead of hiding uncertainty inside an inflated impact score.
• I also price delay and reversibility, because a defensive fix before a platform release can beat a larger but slower growth bet.

I separate them because they represent different demand and respond to different levers.

• Branded clicks often reflect awareness created elsewhere, so counting all of them as SEO acquisition overstates channel impact.
• Non-brand segments reveal category reach, but I split informational and commercial intent so traffic growth does not hide weak buying demand.
• Brand still matters defensively, especially for reputation, resellers, and SERP features, so I report it rather than discard it.

I reserve capacity for protecting proven value before funding uncertain expansion.

• Defensive work covers regressions in indexability, migrations, expiring content, and SERP ownership around the brand or core products.
• Growth work targets new clusters, markets, or templates only after the underlying crawl and conversion paths can support them.
• I compare expected loss avoided with expected gain created, because prevention has economic value even when traffic stays flat.

I move investment when the next unit of effort produces less incremental value than another credible opportunity.

• Improving a stable top result may yield fewer clicks than bringing an entire high-intent cluster from page two into contention.
• More content can reduce returns if it fragments links, duplicates intent, or adds maintenance cost to the indexable inventory.
• I compare marginal upside and opportunity cost, not historical channel performance, before extending a mature program.

I use revenue evidence as a range of contribution, not a single ownership claim.

• I join landing pages and query groups to qualified actions in analytics and the customer system, while keeping identity and consent gaps visible.
• I compare first-touch, last-touch, and assisted paths because informational search often influences a later branded or direct conversion.
• Where stakes justify it, I use holdouts or time-series evidence to estimate incrementality beyond observed attribution.

I pair early delivery and search signals with business outcomes that mature later.

• Leading metrics include valid indexable coverage, crawl discovery, impressions, and visibility for the intended non-brand clusters.
• Lagging metrics include qualified leads, revenue, retention where relevant, and incremental acquisition rather than raw sessions alone.
• I define expected timing by initiative so a slow revenue result is not mistaken for failure before pages can compete.

I would make indexability an explicit product state owned at the template and URL-family level.

• Each family has a declared purpose, demand case, canonical behavior, sitemap eligibility, and minimum content requirements.
• Inventory reporting reconciles generated, crawlable, canonical, indexed, and traffic-producing counts so URL growth cannot stay invisible.
• New families launch behind sampled crawl and index quality gates rather than becoming indexable by default.

I would encode states around user value and replacement intent, not apply one status code to every unavailable product.

• Temporarily unavailable products stay live when the offer is likely to return and the page still helps users.
• Permanently retired products redirect only to a close substitute or stable successor; otherwise they return a genuine 404 or 410.
• Sitemaps, internal links, canonicals, and structured data change with the state so search signals do not contradict the response.

I would whitelist indexable facet combinations from proven demand and suppress the rest from crawlable discovery.

• Approved combinations receive stable URLs, distinct landing value, self-canonicals, internal links, and sitemap inclusion.
• Unapproved sorting and filter states do not emit crawlable links or indexable pages, while users can still apply them in the interface.
• The rules cap facet depth and conflicting combinations, then logs verify that bots spend time on the intended inventory.

I use verified bot requests to compare actual crawl allocation with the URL families the business wants discovered.

• I aggregate hits, response codes, recrawl intervals, and wasted parameter requests by template rather than inspect isolated URLs.
• I compare those patterns with internal links, sitemap membership, change frequency, and index outcomes to find the controllable cause.
• I change architecture or generation rules first, then confirm in later logs that useful families gained crawl without just increasing total requests.

I would use a rules engine when URL states are generated systematically and cannot be maintained safely as individual mappings.

• Deterministic rules cover normalization, known legacy patterns, and entity successors, while exact exceptions remain versioned and auditable.
• The engine must prevent chains, loops, broad redirects to irrelevant destinations, and conflicts between redirects and canonicals.
• Before release I test representative and adversarial URLs, then monitor destination status and rule-match volumes for unexpected reach.

I model pages as nodes and crawlable links as edges, then ask whether authority and discovery flow toward priority entities.

• Depth, orphan components, and inlink concentration show where navigation or template links create structural advantages and dead ends.
• I weight links by placement and relevance rather than assume every edge passes equal context or value.
• Changes are tested at template level because adding one useful module can reshape thousands of paths more reliably than manual links.

I would keep the smooth interface but expose a deterministic paginated series through crawlable URLs and links.

• Each page has a stable server-resolvable URL, unique item slice, and self-canonical so content does not depend on scrolling or interaction.
• The interface can append results progressively, but normal anchor links provide discovery and a usable fallback.
• I keep ordering stable enough to limit duplication and verify deep items through rendered crawls and search-engine inspection.

I would block scale until the template proves unique user value, valid demand, and operational maintainability.

• Every page needs a defensible intent match, reliable source data, and substantive variation beyond swapped place or product names.
• Automated checks reject empty states, duplicate titles, invalid canonicals, broken structured data, and pages below the template's content threshold.
• A sampled launch must earn indexing, engagement, and conversion signals before the generator expands the inventory.

I would stop generation and discovery first, then map only URLs with a genuinely equivalent destination.

• Valuable or externally linked pages receive one-hop redirects to the closest successor, not a blanket redirect to the homepage.
• URLs without replacements return 404 or 410 and disappear from links, sitemaps, feeds, and canonical references.
• I monitor bot requests and index counts until the family decays, then remove temporary routing rules when they no longer earn their cost.

I turn critical SEO behavior into versioned template contracts with automated acceptance checks.

• Contracts cover response status, indexability, canonical output, headings, crawlable links, structured data, and required content by page state.
• Continuous integration tests fixtures for normal, empty, filtered, and retired states before a template can ship.
• After release, synthetic monitoring and sampled crawls compare the deployed output with the contract so broad failures surface quickly.