100 SEO + GEO questions, 100 concrete answers

SEO (Search Engine Optimization) is the set of techniques that make your site readable and desirable to Google. Three pillars: technical (Google's crawler can reach it), content (what it finds deserves to rank), authority (other sites validate yours). Results are measured in SERP positions.

SEO = free organic traffic, slow but durable (3-9 months to rank). SEA = Google Ads, paid traffic that's instant but stops the second you cut the budget. They complement each other: SEA for fast tests, SEO for long-term asset value.

Plan 3 to 6 months for first long-tail keyword rankings, 6 to 12 months for competitive head keywords. Speed depends on your domain authority, technical quality, and the competitiveness of your sector.

Yes, but it's evolving. Google still holds 90 % of global searches and LLMs (ChatGPT, Claude, Perplexity, Gemini) rely heavily on web content indexed by search engines. SEO is the foundation, GEO (Generative Engine Optimization) the new layer.

On-page = everything you control on your site: titles, content, meta tags, schema, URL structure, speed. Off-page = what happens elsewhere: backlinks, brand mentions, external authority signals (Wikipedia, press, GBP).

The 2024 leaks confirmed 14,000+ signals. The most impactful: content relevance vs intent, domain authority (backlinks), user engagement (CTR, dwell time), freshness, Core Web Vitals, HTTPS, technical structure. E-E-A-T weighs heavily on YMYL topics.

It's the work of identifying the queries your audience types on Google, their monthly volume, competition difficulty, and intent (informational, transactional, navigational). Free tools: Google Suggest, AnswerThePublic. Paid: Ahrefs, Semrush.

It's the real goal behind a query. Four types: informational ("what is"), navigational ("facebook login"), commercial ("best smartphone 2026"), and transactional ("buy iphone 15"). Aligning content to intent is the fastest lever to rank.

The title tag is Google's #1 signal for what a page is about (heavy ranking weight). The meta description doesn't rank directly but impacts SERP CTR. Good length: 50-60 characters for title, 120-160 for meta description.

Short, readable, lowercase URLs with target keywords separated by hyphens. Avoid random numbers, ?id=123 parameters, uppercase, and special characters. Good: /seo-audit-saas. Bad: /page?id=42&cat=SEO_2025.

GEO is the set of techniques to get your site cited by generative AI engines (ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews). Key lever: structure your content into extractable passages with Schema.org structured data and authority signals.

ChatGPT cites sources via Bing Search (web mode) or its training corpus. To get cited: complete Schema.org Organization + Article, Q&A-format content, llms.txt, external authority signals (Wikipedia, Wikidata, press mentions). Bing Webmaster Tools is its primary source.

llms.txt is an emerging standard (backed by Anthropic, Vercel, Stripe) that guides LLMs to the key sections of your site in Markdown. Format at root: /llms.txt. No direct ranking but a strong AI authority signal and cleaner extraction.

AI Overviews is the generative AI answer Google displays at the top of results for ~30% of queries. It synthesises multiple sources and cites domains. Appearing inside has become a major organic acquisition channel in 2026.

Perplexity favours sources that answer the question in a clear passage, with authority (backlinks, brand signals) and freshness. Sites with FAQPage schema, complete Article schema, and Q&A-formatted content get cited more often. More purely source-driven than Google.

Yes, it's one of the strongest GEO levers. The most impactful types: Organization (brand identity), Article (authorship), FAQPage (extractability), Product (e-commerce), BreadcrumbList (navigation). Without schema, AIs hesitate to cite you as an authoritative source.

E-E-A-T = Experience, Expertise, Authoritativeness, Trustworthiness. Google and LLMs use it to evaluate whether a page deserves to be cited. Concrete signals: named author with bio, publication/updated dates, cited sources, linked Person + Organization schema.

Not killed, transformed. 58% of searches under 13 words end without a click in 2026. SEO must aim to: be cited inside AI Overviews / featured snippets (free brand awareness), and win the remaining clicks on transactional queries.

Emerging tools: Profound, AthenaHQ, AlsoAsked AI. Manual method: run 5-10 business queries on ChatGPT, Claude, Perplexity and note your presence (cited, mentioned, ignored). SeAudit includes an estimative test based on on-page signals.

Yes, and it's win-win: most GEO levers (Schema, Q&A content, llms.txt, E-E-A-T) also reinforce traditional SEO. Reversing the order would be risky. Start by fixing technical SEO, then stack GEO on top.

An SEO audit is a full diagnostic of your site: crawl, content, authority, performance, AI visibility. It quantifies what works, what blocks, and prioritises actions by impact × effort. Typical output: PDF with score, prioritised issues, 30/60/90-day plan.

Agency: €1,500 to €4,000 depending on scope. Senior freelance: €800 to €2,000. Automated tools like Semrush/Ahrefs: €99-449/month (but you need expertise to interpret). SeAudit: €97-247 for an audit delivered in 48-72h.

Full audit: at least once a year, once a quarter for e-commerce. Mini-audit after every major change (redesign, migration, semantic overhaul). Continuous monitoring of key positions: monthly.

Crawl: Screaming Frog (£89/yr, gold standard). Visibility: Ahrefs / Semrush (both premium). Performance: PageSpeed Insights (free), DebugBear. GEO: Profound, AthenaHQ. Free: Google Search Console + Bing Webmaster Tools (essentials).

Yes: Google Search Console (essential), Google PageSpeed Insights, Bing Webmaster Tools, Screaming Frog free version (500 URLs), Schema.org Validator, Mobile-Friendly Test. SeAudit also offers a free /100 score with no signup.

Yes for 80% of signals: external crawl, meta tags, schema, mobile performance, sitemap, robots.txt, backlinks, brand signals. Only Google Search Console and Analytics data are missing. That's what enables instant auditing from a single URL.

Audit = snapshot of current state and identification of issues/levers. Rank tracking = movie of how rankings evolve on target keywords over time. They're complementary: audit to understand, tracking to measure progress.

SeAudit benchmarks: <40 critical (blocking issues), 40-69 average (big levers to activate), 70-84 decent (polish + authority push), 85+ solid. A perfect score doesn't exist: it's a relative benchmark to sector and competition level.

Three absolute priorities, in order: (1) technical blockers (noindex, robots, sitemap, HTTPS), (2) titles and meta descriptions of strategic pages, (3) Schema.org Organization + Article. The rest comes once the basics hold.

An actionable PDF. Inside: overall score and per-axis sub-scores, Top 10 actions prioritised by impact × effort, technical / content / GEO / backlinks sections, 30-60-90-day plan. Avoid raw, uncommented CSV reports: that's data, not an audit.

robots.txt tells crawlers (Googlebot, Bingbot, GPTBot, ClaudeBot…) which parts of the site they may or may not explore. File at the domain root. Doesn't block indexing on its own (use meta robots noindex for that).

Yes, as soon as your site exceeds twenty or so pages or if your internal linking leaves orphan pages. The sitemap lists URLs to prioritise for indexing, with last-modified dates. Submit it via Google Search Console.

It tells Google which URL is the "official" version when multiple URLs serve similar content (pagination, parameters, http/https, www/non-www). Avoids cannibalisation and authority dispersion. Tag <link rel="canonical" href="..."> in the head.

Each page must list all its language versions via <link rel="alternate" hreflang="en-GB" href="..."> + an x-default version for fallback. Symmetry is mandatory: if A points to B, B must point back to A. Classic mistake: forgetting the return link.

First check: meta robots noindex present? robots.txt blocking? page accessible (200 OK, not 404)? canonical pointing elsewhere? duplicate content? Search Console > "URL Inspection" gives the exact diagnosis. Request manual indexing after fixing.

Yes, since 2014 it's an official ranking factor. In 2026, a pure HTTP site is flagged "not secure" by all browsers and demoted by Google. Let's Encrypt provides free certificates — no excuse not to enable it.

Google renders JS via headless Chromium but with a delay (deferred indexing). Other bots (Bing, GPTBot, ClaudeBot) handle it poorly or not at all. Solution: SSR or pre-rendering (Next.js, Nuxt, Astro). Client-only content = invisible to LLMs.

301 = permanent, passes 99% of SEO juice to the new URL. Use for migrations, redesigns, definitive moves. 302 = temporary, keeps the old URL in the index. Common bug: using 302 by default "to not break anything", which kills the SEO transfer.

Three levers: properly set canonical tag, 301 redirects for multiple versions (http/https, www/non-www), URL parameters filtered in Search Console. Content-wise: don't copy supplier product descriptions, rewrite them.

Yes, via Core Web Vitals (LCP, INP, CLS) which are official factors since 2021. Direct impact moderate, indirect impact huge: a slow site repels users, which degrades behavioural signals (bounce, dwell time) that weigh heavily.

No official minimum but observed benchmarks: 800-1,500 words for a standard informational article, 2,000-3,000 for a pillar guide. More important than length: cover the full intent and beat ranking pages on answer depth.

Low individual volume but massive combined sum (~70% of search traffic), low competition, often stronger transactional intent. Ideal strategy for young sites or niches. Example: "seo audit saas b2b" beats "seo audit" at the start.

Regularity > volume. Better: 1 quality article per week for 6 months than 5 rushed articles in one month. Google rewards both freshness AND depth. For pillar topics, plan 2,000+ words, real sources, named author.

A broad pillar page (e.g. "Complete SEO Guide") surrounded by 5-15 deep satellite articles that link to each other and back to the pillar. Boosts topical authority perceived by Google and eases extraction by LLMs. HubSpot method, popularised in 2017, still effective.

Evergreen for the foundation (80% of long-term traffic): guides, definitions, comparisons. News for brand + spikes: industry news, trends. Mixing 70% evergreen / 30% news is a good ratio for a brand built to last.

One H1 per page (main topic). H2 = main sections. H3 = sub-sections. No skipping (H1 → H3 without H2). Include target keywords in H1 and H2 when natural. LLMs rely on this hierarchy to extract passages.

Factual description of what's shown, tied to the page context. 5-15 words typically. No "image of" prefix. No keyword stuffing. If the image is purely decorative, alt="" (empty string). Both an accessibility AND SEO criterion.

Three rules: link from high-authority pages to your target pages, use descriptive anchor text (not "click here"), keep a max depth of 3 clicks from home. Internal linking distributes PageRank and clarifies structure for Google.

Yes, often the most cost-effective lever. Identify pages ranking 4-15 (close to top), rewrite them to better match current intent, update the date. Google loves refreshed content. Plan 2-4 weeks to see ranking improvement.

Not penalised as such since Google's 2023 clarification. What matters: user value. Raw AI content copy-pasted without review ranks poorly because it's repetitive and shallow. AI + human edit + sources = winning strategy.

Three official Google metrics: LCP (Largest Contentful Paint, < 2.5s = good), INP (Interaction to Next Paint, < 200ms = good), CLS (Cumulative Layout Shift, < 0.1 = good). Measurable in PageSpeed Insights, real data in Search Console.

Top 5 actions: preload the hero image (rel="preload"), serve WebP/AVIF images with srcset, defer non-critical third-party scripts, use a CDN close to users, set up SSR or ISR. Target < 2.5s on mobile in field data.

INP (Interaction to Next Paint) measures the latency of ALL user interactions (taps, clicks, key inputs), not just the first one like FID. Became an official Core Web Vital in March 2024. Target: < 200ms. Typical INP killer: heavy JS at mount.

CLS = layout jumping after load. Main causes: images without width/height set, web fonts loading with shift (FOUT/FOIT), banners/ads injected dynamically, content inserted above what the user is already reading. Always reserve the space.

Since 2020, Google crawls your site's mobile version first to decide rankings. If your mobile version hides content present on desktop, that content is ignored. Golden rule: what you want to rank must be present and accessible on mobile.

No, AMP lost its SEO advantage in 2021 (removal of the mandatory Top Stories carousel). Today, better invest in healthy Core Web Vitals on your own stack (Next.js, Astro, static) than maintain a parallel AMP version.

No, never on the hero image (above-the-fold) because it degrades LCP. Lazy-load only images below the fold. Native loading="lazy" attribute is enough, no heavy JS lib needed. Works in all browsers since 2020.

AVIF compresses 30-50% better than WebP at equivalent quality, but encoding is slower. 2026 strategy: target AVIF with WebP fallback via <picture>. For image-heavy sites, this can halve bandwidth.

Mobile > 80 (lab data) with green field data on LCP/INP/CLS. Hitting 100 is rarely worth it beyond a teaching moment. Field data (CrUX, Search Console) counts 100x more than lab data for Google. Measure what matters.

Excellent: native SSR/SSG/ISR, App Router with generateMetadata, Image component that enforces width/height (anti-CLS), optimised routing. Paired with Vercel or Cloudflare Pages, it's one of the strongest SEO stacks in 2026. Manage server components carefully for prerendering.

A backlink is a link from another site pointing to yours. One of the 3 major ranking factors: Google treats it as a vote of confidence. Not all links are equal: one link from The Guardian > 10 links from junk sites.

Dofollow = passes SEO juice (PageRank). Nofollow = signal to the engine not to follow/weight the link (comments, sponsored, user content). Both have value in 2026: Google treats nofollow as a "hint" rather than a hard directive.

0-100 metrics invented by third-party tools (Moz DA, Ahrefs DR, Semrush AS) to estimate domain strength. Not an official Google signal. Useful internally for comparison / prospecting, but Google doesn't use it. Real signal: the actual backlink profile.

Five effective levers: link-worthy content (study, dataviz), digital PR (HARO, niche press), editorial guest posting, healthy link swaps with partners, unlinked brand mentions you turn into links. Avoid PBN backlink purchases.

Rarely useful in 2026. Google auto-ignores most low-quality links since Penguin 4.0 (2016). Only disavow in case of documented negative SEO attack or manual penalty received in Search Console.

External backlinks = major weight (authority), internal links = distribution + structure. Both matter. For a young site, prioritise clean internal linking (quick win). For a mature site, prospect external backlinks (long-term win).

Vary: target keywords, brand, naked URL, long descriptive phrase. A 100% exact-match anchor profile = manipulation signal. Good ratio: 30% brand, 30% URL/generic, 20% exact match, 20% contextual long-tail.

Three pillars: entity recognised on Wikidata with a Wikipedia page ideally, complete Organization schema (sameAs to GBP, LinkedIn, Crunchbase), consistent press mentions. Patience: 6-18 months after setup for a Panel to appear on brand SERP.

The Wikipedia link is nofollow so no direct PageRank juice. But it's a major indirect authority signal: LLMs rely heavily on Wikipedia, and an article that cites you passes your entity into the future training corpus. Big GEO lever.

Yes, Google can recognise unlinked brand co-citations and uses them as an authority signal since Hummingbird (2013). In 2026 it's even more important for GEO: LLMs rely on brand + topic co-occurrence in their corpus to cite.

Local SEO optimises your presence on geolocalised queries ("hairdresser Bordeaux", "plumber near me"). Three pillars: optimised Google Business Profile (formerly GMB), consistent NAP citations, regular customer reviews. Critical for brick-and-mortar businesses.

Five quick wins: fill 100% of fields (categories, hours, attributes), add 10+ up-to-date professional photos, reply to ALL reviews within 24h, post weekly content (offers, news), encourage customer reviews with a direct Google link.

NAP = Name, Address, Phone. Strict consistency of these 3 pieces across all your presence points (site, GBP, directories, social) is a major trust factor for Google Local. A single comma difference can dilute your authority.

Register in quality directories for your sector: Yelp, Yellow Pages, Foursquare, plus sector-specific directories (TripAdvisor for restaurants, Healthgrades for health, etc.). Verify NAP is strictly identical everywhere. Tools: BrightLocal, Whitespark.

LocalBusiness with a specialised @type if possible (Restaurant, Dentist, HairSalon…). Required fields: name, address (PostalAddress), telephone, openingHoursSpecification, geo (latitude/longitude), priceRange. Add aggregateRating if you have ≥ 5 reviews.

Massively. Three key metrics: review count (volume), average rating (quality), recency (freshness). A business with 50 reviews at 4.8★ beats a competitor with 10 reviews at 4.9★. Aim for > 4.5 stars and > 30 reviews to rank in the local pack.

Set up GBP in "service area" mode without public address, precisely list covered cities (max 20), create local pages per zone ("Plumber in Mérignac", "Plumber in Pessac"…) with unique content, photos, local testimonials. ServiceArea schema.

One local page per location with NAP, photos, map, testimonials, local offers. One GBP per location. LocalBusiness schema per location with unique @id. Internal linking: national page → local pages. Avoid duplicate via genuine per-zone content.

Three pillars (Local Search Ranking Factors 2024): relevance (optimised GBP), distance (can't be influenced via SEO), prominence (reviews, citations, site authority). Of the 3, prominence is the only real medium-term lever.

Reply within 24h, publicly, without aggression: acknowledge, explain, propose a solution. Invite private follow-up to resolve. The reply matters as much as the review itself: future customers read it. Never delete a real review.

Five priorities: (1) Product schema with price, availability, rating, (2) unique product descriptions (never copy supplier), (3) semantically optimised category pages, (4) breadcrumb schema, (5) strict variant handling via canonical to avoid duplicate.

Decent by default, excellent with polish. Strengths: solid technical structure, OK speed, native Product schema. Weaknesses: forced URL structure (/products/, /collections/), limited canonical control, dependence on third-party apps for advanced optimisations.

Complete Product schema: name, image, description, sku, brand, offers (price, priceCurrency, availability, url), aggregateRating + review if you have them. Brand as nested Brand or Organization schema. Verify in Google's Rich Results Test.

No, e-commerce anti-pattern. Questions are generic per category, not per SKU. Better: FAQPage schema on category/collection pages (sizing, returns, shipping) and keep pure Product schema on product pages. Avoid dilution.

Editorial 200-400 words above or below the grid (no hiding), canonicalised filters (one indexable URL), clean pagination (rel="next" deprecated, rely on sitemap), breadcrumb schema. Category pages are often more profitable than product pages.

One canonical URL for the master product, variants accessed via URL parameters (?variant=) or JS. Never index one URL per variant: creates massive duplicate. Product schema with variants via offers[] if volume justifies it.

Never 404 or redirect to home. Keep the page accessible with Schema availability="OutOfStock", suggest alternatives, "notify me when back" form. If permanently out: 301 to parent category or successor product.

Product title with brand + model + key attribute (max 150 chars), keyword-rich description (5,000 chars possible), GTIN + brand mandatory for Performance Max, high-resolution photos without watermark. Sync availability + price in real time.

Yes, it's a critical top-of-funnel capture channel. Strategy: 1 article per week on informational intents ("how to choose", "what is the best"), internal linking to categories. Winning ratio: 70% editorial / 30% transactional.

Both depending on target ROI. Google = relevant for top-funnel acquisition + brand. Amazon = relevant for bottom-funnel conversion on mature products, but margin crushed by fees (15 %+) and dependency on opaque algos. Ideally: presence on both + direct site.

SeAudit is the automated SEO + GEO audit that delivers an actionable PDF in 48-72h from a single URL, at €97 instead of €1,500-2,000 in agency. Built for founders and marketers who don't have time to brief an agency.

You enter your URL, we calculate a free /100 score across 5 axes (technical, content, GEO, perf, trust) in 30 seconds. For the full PDF, we generate via Claude (Anthropic) from a deep crawl: 25 to 50 pages, schemas, mobile perf, GEO signals.

The full PDF audit starts at €97 (launch tier, limited slots), then €147, then €247. No subscription, no commitment. À-la-carte add-ons: SEO article, meta rewrites, schema pack, monthly AI position monitoring.

Free /100 score: 30 seconds. Full PDF audit: 48 to 72 hours after payment, by email. Express 24h option available as add-on. All audits are manually reviewed before sending to guarantee relevance.

Same method, same PDF deliverable. The difference: no 2h brief, no 6-week wait, no €2,000 invoice. We work for founders who want a clear diagnosis without the sales dance. The recommendations are implementable in-house.

Semrush/Ahrefs are excellent raw data tools but require SEO expertise to interpret and cost €99-449/month. SeAudit delivers the interpreted summary + prioritised action plan in one shot. And we include GEO, which Semrush doesn't yet measure properly.

Yes, that's our differentiator. The PDF's GEO section covers: detected Schema.org inventory, LLM citability format, llms.txt presence/conformity, E-E-A-T signals, estimative visibility test in ChatGPT/Claude/Perplexity on 3 business keywords.

Yes, 14-day no-questions-asked satisfaction guarantee. If the audit doesn't seem relevant to you, we refund 100 % via Stripe. Our commitment: the PDF must deliver clear value or it's not owed.

Yes, natively in French and English. The scoring and crawl engine is language-agnostic, and the PDF is written by Claude who masters both languages perfectly. Editorial tone is direct second-person in both languages.

Yes as add-on: monthly re-audit + diff vs previous month at €49/month, or SeAudit Plus bundle (re-audit + 1 SEO article + 1 rewrite/month + AI position tracking) at €129/month. No commitment, 1-click cancellation.