SEO Interview Questions for Experienced Candidates (3+ Years)

In order: (1) Verify it’s real — check GA4 for tracking breakage and confirm the drop appears in Search Console clicks too. (2) Check for announced causes — GSC manual actions and security issues first, then Google’s Search Status Dashboard for a confirmed update rollout during the drop window. (3) Segment the loss — GSC compare mode: is it site-wide or specific pages/queries? Brand or non-brand? A device or country? (4) Match the pattern: site-wide + update timing = algorithmic reassessment; specific pages = re-check those SERPs (new competitor, intent shift, AI Overview appearing); sudden overnight cliff = technical (accidental noindex, robots.txt, hosting). (5) Only then act — and if it’s mid-rollout of a core update, the correct action is documented patience, not panic changes.

That’s keyword cannibalization — two pages competing for one query, splitting signals so neither wins. I’d confirm it in GSC by filtering the query and switching to the Pages dimension. The fix depends on the pages: if both genuinely serve the same intent, merge the best content into the stronger URL and 301 the other; if they serve different intents that got blurred, differentiate them — retarget one to a distinct cluster, adjust titles and internal anchor text so signals point consistently. Prevention is a keyword mapping sheet: one cluster, one page, enforced at planning time.

Rising impressions mean growing visibility — falling CTR alongside it has three usual causes, distinguishable by looking at the actual SERP. (1) An AI Overview now appears and absorbs clicks — check whether we’re cited in it; if not, that’s the work. (2) We gained impressions for new marginal queries at low positions, mathematically diluting CTR — segment by query to confirm; this is actually healthy pipeline. (3) Our title/snippet is losing the click to better-packaged competitors — a title rewrite test. The key point: this pattern is a diagnosis trigger, not automatically a problem.

I tell them the truth calmly: a confirmed update is rolling out, there’s no penalty (verified in GSC — penalties come with messages), best practice is not to make changes mid-rollout, and we’ll assess when it completes. What I actually do: benchmark exactly which pages and queries lost, compare against what now ranks instead (the gap is usually depth, first-hand experience, or trust signals), then run a quality-focused improvement sprint on the highest-value hit pages. I also set expectations honestly: recovery from core updates typically shows at subsequent updates — months, not weeks — and the work is real content improvement, not a checklist tweak.

Week 1: access and baselines — GSC, GA4 with key events for their actual conversions (calls, WhatsApp, forms), a full technical crawl, and the business conversation: what do they sell, what’s a lead worth, who are their real competitors. Week 2: keyword research and mapping — clusters mapped to existing pages, gaps identified, prioritised by business value not volume. Week 3: fix the blockers — indexing issues, cannibalization, broken redirects; these gate everything else. Week 4: the roadmap — quick wins first (position 8–20 pages, title CTR fixes), then the content and authority plan by quarter, delivered with the measurement framework we’ll report against. The principle: diagnose and prioritise before producing.

Impact versus effort, with one override: blockers first — anything preventing crawling or indexing gates all other work, so it jumps the queue regardless of effort. After that, I score by expected business impact (traffic × conversion value of the affected pages) against implementation cost, which naturally surfaces the classic quick wins: pages ranked 8–20 for valuable queries, high-impression/low-CTR titles, and internal linking fixes. Big-effort items (site restructures, content programs) get scheduled, not skipped. And I re-prioritise monthly against actual GSC data rather than following the original plan off a cliff.

Before: full URL inventory (crawl + GSC + analytics top pages), a one-to-one 301 redirect map to the closest equivalent — never everything to the homepage — and benchmark rankings/traffic for comparison. During launch: redirects live at cutover, internal links updated to final URLs (not routed through redirects), canonicals and sitemaps pointing to new URLs, GSC change-of-address for domain moves, and both properties verified. After: monitor crawl stats and indexing of new URLs daily, watch the old URLs drain, fix redirect chains and 404s from the live error list. Expectations set in advance: some turbulence for weeks is normal even in a clean migration; disaster only comes from missing or lazy redirects.

Quarterly, from GSC data rather than instinct. Existing pages get priority when they’re declining (period-over-period compare), sitting at positions 8–20, or earning impressions with poor CTR — an existing page has indexation, links and history; it usually only lacks a current, complete answer, so refreshing it is the shortest path to movement. New content is for genuine cluster gaps in the keyword map. And some pages deserve neither: overlapping pages get merged with 301s, and dead-weight pages get pruned — a decaying tail drags the site-wide quality assessment down.

Earned-first: create genuinely citable assets — original data (even small surveys), tools, definitive guides — and put them in front of the writers who cover the space; supplement with guest contributions on real publications and expert quotes via journalist request platforms. Safety comes from one filter applied ruthlessly: would this link exist without payment or exchange? If payment is the only justification, it’s a bought link whatever it’s packaged as, and I decline — including the “guest post packages” that dominate outreach spam. I count referring domains over raw links, judge sites by real audience over metrics, and I don’t do disavow theatre — without a manual action, junk backlinks are noise Google already ignores.

E-E-A-T isn’t a score to hack — it’s proxies to earn, so I work the proxies. On-site: named authors with real bios and credentials on every piece, a substantive About page, working contact routes, sourced claims, and content that demonstrates first-hand experience (original photos, real numbers, practitioner details) rather than claiming it. Off-site: build the author and brand entities — consistent identity across profiles, Person/Organization schema with sameAs, and independent validation through press mentions, expert quotes and reviews. For YMYL topics specifically, visible qualified review of content isn’t optional. The one-line version: make it effortless for both a human rater and a machine to verify who’s behind the content and why they’re credible.

URL Inspection in GSC first — it names the reason. Then down the pipeline in order: Is it blocked in robots.txt? Carrying a noindex (often a page-builder or plugin toggle)? Canonicalised to another URL — and did Google choose a different canonical than declared? Is it in the sitemap and internally linked, or an orphan? If it’s “Discovered — currently not indexed” at scale, that’s usually a quality/crawl-priority verdict, not a technical bug — the fix is improving or consolidating thin pages, not resubmitting them. I’d also check whether the content requires JavaScript rendering to exist, since that adds delay and risk.

First, field data over lab data — the CWV report in GSC and the real-user section of PageSpeed Insights, because that’s what counts for ranking; Lighthouse lab scores are diagnostic only. Then match fix to metric: poor LCP is usually hosting response time, render-blocking assets, or an unoptimised hero image; poor INP is heavy JavaScript (plugin bloat on WordPress); CLS is images without dimensions and injected banners. On WordPress specifically, the highest-ROI sequence is decent hosting, a caching setup, image compression/WebP, and plugin audit. And I frame expectations honestly: CWV is a threshold, not a race — once page groups pass “Good,” further speed chasing has diminishing SEO returns.

The types that earn something: Organization/LocalBusiness (entity clarity and local SEO), Article with author linkage (supports E-E-A-T verification), Product with price and reviews (rich results that move CTR), FAQ and HowTo where genuinely appropriate, and Person schema with sameAs on author pages for entity building. Implementation on WordPress is mostly Rank Math/plugin-generated with manual JSON-LD for the custom cases. Validation: Rich Results Test for eligibility, the Schema.org validator for syntax, and GSC’s Enhancements reports for ongoing monitoring — plus the honesty rule: markup must describe what’s visibly on the page, or it’s a spam risk rather than an asset.

Business outcomes first: key events from organic — enquiries, calls, WhatsApp clicks, orders — trended and compared, because “traffic up 30%” means nothing to an owner and “47 enquiries from organic, up from 31” means everything. Under that: traffic and visibility (GSC clicks/impressions, brand vs non-brand split), meaningful ranking movements with causes for the losses, work shipped that month, context annotations (confirmed updates, seasonality), and next month’s committed plan. Losses get reported with the same clarity as wins — that honesty is what keeps the relationship through the inevitable bad month.

Three concrete changes. Content: even more answer-first and extractable — self-contained passages with specific facts, because AI answers are assembled passage by passage and cite the sources with concrete substance. Measurement: I read impressions-up/CTR-down against AI Overview presence before diagnosing content failure, and I’ve shifted reporting weight from raw clicks toward enquiries and brand demand, since cited visibility increasingly converts without a click. Scope: I now treat citation in AI answers as a target alongside rankings — including making sure the site is indexable by Bing (ChatGPT’s retrieval) and AI crawlers aren’t accidentally blocked. What hasn’t changed: the fundamentals feed both systems — commodity content just got even more worthless.

With the honest competitive picture, respectfully. I’d show them the actual SERP: who ranks, their authority and content depth relative to ours, and which SERP features (ads, local pack, AI Overview) sit above position 1 anyway — often “position 1” gets fewer clicks than they imagine. Then I redirect the conversation to what we can win and what it’s worth: the cluster of adjacent and long-tail queries where we’re competitive now, and the realistic path and timeline to the head term. Most importantly, I anchor on their real goal — they don’t want position 1, they want customers — and show where those are actually coming from.

By making their work easier, not by throwing audits over the wall. With developers: tickets, not PDFs — each issue with the specific URLs, the expected behaviour, the business reason, and a priority label; batched sensibly; and I test in staging with them rather than filing and vanishing. With writers: briefs that respect their craft — the target cluster, the questions to answer, the intent, real examples to include — never keyword-density demands; and I share the GSC results of their pieces, because writers who see their article hit position 3 become allies permanently. In both cases the currency is credibility: a few early wins delivered together buys every future request.

Have a real one ready — the structure matters more than the story: the decision and why it was reasonable at the time, the honest outcome with numbers, the diagnosis of what was actually wrong, and the specific change in your process since. Example shape: “I scaled location pages off one template across 20 areas; they never ranked and indexation stalled — Google treated them as doorway pages. Now every location page passes a simple test: could this only have been written by someone who actually works in that area? Three real pages outperformed the twenty thin ones within a quarter.” Candidates who can’t name a failure read as either inexperienced or dishonest.