Practical guides to technical SEO, content optimization, structured data, and AI search visibility for websites that want more organic discovery.
Featured
Recovery plan
Rebuild the weak parts of the site before you reapply
Audit the whole site, remove or rebuild weak inventory, add first-hand value, fix AI-assisted pages, and reapply after the public site has changed.
Read the field guide50-question series
Focused answers for crawling, indexing, snippets, structured data, AI citations, and Next.js discoverability.
Name the task, the object being checked or created, and the page's useful distinction. Keep the title aligned with the visible H1 and the tool's real output.
Define one reader question, the decision behind it, the evidence needed, and a distinct result. Give the writer an answer map instead of a keyword quota.
Group queries by reader and decision, then assign each group to one canonical page. Use close wording inside the page and create a new URL only for a distinct task.
Track each material claim with its primary source, supported passage, scope, date, and review status. Keep observed AI citations in a separate log.
Name the entity, its category, function, scope, and nearest confusable alternative in two or three sourced sentences. Reuse the same public name across the page.
Keep source and referrer data, landing-page visits, assisted actions, observed citations, and brand searches in separate views. Document attribution limits.
Find every dependent claim, compare old and new guidance, revise the answer and conditions, record the change, and retest any cited passages or structured data.
Verify business identity, service and area pages, contact paths, crawl signals, local proof, and conversion tracking. Rank defects by customer and search impact.
Check audience fit, category quality, moderation, current activity, cost, link and profile treatment, measurable traffic paths, and ongoing maintenance before submitting.
Review audience, editorial purpose, category context, listing accuracy, ownership, destination path, cost, and measurable referral value before treating the link as useful.
Compare reader, question, decision, and answer across existing URLs before drafting. Merge close variants when the same searcher would choose either page for the same task.
Crawl the site, group pages by reader task, and inspect incoming links to each important URL. Fix broken paths, orphan pages, vague anchors, and misplaced links first.
Choose the URL that best represents the shared reader task, merge unique material into it, update internal links, and redirect each retired page straight to that destination.
Put the entity, direct answer, scope, condition, date, and source in one compact passage. The excerpt should remain accurate when read away from the rest of the page.
Define the compared entities, criteria, date, units, and source in visible text. Keep each cell specific and explain conditions that a compact table cannot carry.
Use a fixed question set, save complete outputs and citations, then compare the extracted claim with the page's entity, scope, conditions, date, and source.
Document the question, setup, inputs, date, environment, procedure, output, and limits. Separate what you observed from the broader recommendation.
Create a location page only when the business has distinct service facts, proof, constraints, or customer decisions for that area. Otherwise use one strong regional page.
Choose directories used by your buyers, prepare one accurate product record, tailor each submission, and track review status, referrals, qualified actions, and maintenance.
Keep the canonical article as the complete source, then adapt one useful idea for each channel with a clear reason and link to the full task when appropriate.
Check access, indexability, page identity, answer quality, and internal links in that order. The result is a short evidence-backed fix list, not a generic score.
Summarize the product's task, audience, and concrete output in one or two sentences. Use visible page facts and give the searcher a sound reason to visit.
Diagnose the loss by query, page, country, and date before rewriting. Update stale evidence, repair intent gaps, and preserve the URL when it still serves the same task.
Track eligibility, impressions, qualified visits, reader actions, and observed citations as separate stages. Use a fixed question set and dated change log.
Log the prompt, answer, cited URL, quoted or paraphrased claim, date, and product. Check support, conditions, visibility, referral path, and business fit separately.
Use real follow-up questions that change the reader's action. Give each a self-contained sourced answer and remove questions already answered by the main sections.
Keep the customer's decision at the centre, verify the answer with primary evidence, and add the example, conditions, and next action needed to finish the task.
Name the service, area, customer problem, process, proof, and next step. Answer cost and timing factors with concrete conditions instead of generic sales copy.
Build paths from regional hubs to services and qualified locations, then link between pages only where a customer needs the next detail. Use descriptive anchors.
Organize the launch around a working product path, clear positioning, proof, analytics, owned pages, selected channels, support, and a dated follow-up review.
Render ordinary anchor links to stable heading IDs, then add scrollspy as progressive enhancement. Keep navigation usable without JavaScript and below sticky headers.
JavaScript can alter the DOM, but using hydration to repair a wrong or missing canonical creates conflicting signals. Emit one correct canonical in the server response.
A CSP can block scripts, styles, images, or connections and break hydration, but server-rendered text may remain visible. Test both raw HTML and the working experience.
Describe the information represented by the preview image without repeating the full title mechanically. Keep essential guidance in page text, not only inside artwork.
The robots metadata route often builds Sitemap and Host from a deployment URL. Use one validated public origin and keep preview environments protected or noindexed.
Using new Date() during sitemap generation records deployment time, not meaningful page modification. Source lastModified from content data or omit it.
The route often returns 200 with a missing-content message or generic shell. Call notFound for absent records and make error pages clearly distinct from valid content.
No. It chooses paths to pre-render, but unknown dynamic routes still need explicit lookup and notFound handling. Otherwise a generic 200 page can become a soft 404.
Relative metadata URLs resolve against metadataBase. A localhost, branch, or deployment hostname there can leak into canonical and social tags across the entire site.
Yes. Canonical, title, description, robots, and social metadata should be available in the server response so crawlers and preview clients do not depend on hydration.
Compare the generated claim with the exact cited passage, then score entailment, scope, freshness, and source quality instead of counting the link as automatically correct.
Bing reports grounding queries generated during AI answer creation, not necessarily the user's complete prompt. Treat them as retrieval evidence, not keyword rankings.
A citation can satisfy the informational need inside the answer, appear for low-intent queries, or link to a page with no compelling next step. Diagnose the full path.
Track every consequential claim, its primary source, checked date, applicable scope, page location, and next review so citations remain verifiable as facts change.
Keep the claim, eligibility, time frame, exception, and source in one compact passage so extraction does not turn a qualified answer into a universal promise.
Google says nosnippet applies to AI Overviews and AI Mode, but it also removes ordinary text snippets. Use narrower controls when only part of a page needs exclusion.
No. Google says Google-Extended is not used to control inclusion or ranking in Google Search features such as AI Overviews. Standard Search controls apply.
OAI-SearchBot supports ChatGPT search discovery, while GPTBot relates to potential model improvement. Publishers can set separate robots policies for each crawler.
A CDN, WAF, bot manager, rate limit, or origin rule can deny requests after robots permission. Diagnose the complete request path and verify with current OpenAI ranges.
Google says no special AI file or markup is required for its AI search features. Indexing, snippet eligibility, clear evidence, and ordinary SEO remain the controls that matter.
Stable identifiers let Article, Organization, WebSite, breadcrumb, and software entities refer to one another without creating contradictory duplicates.
Real first-party testimonials can appear visibly, but self-serving review markup for your own organization is generally ineligible for Google review stars.
Validation confirms format and supported fields, not indexing, policy compliance, page quality, feature availability, or selection for a specific search.
Google can process JavaScript-generated structured data, but server-rendering JSON-LD is simpler to fetch, test, cache, and keep aligned with the initial page.
Use sameAs for authoritative profiles or identifiers that represent the exact same person or organization, not for every link associated with the brand.
Use Dataset markup when the directory exposes a coherent, documented collection of data. A list of marketing cards without data provenance is not a dataset.
A browser tool can use SoftwareApplication markup when the visible page describes a real application, its category, operating system, and offer accurately.
Use FAQ markup only for genuine visible questions and answers, not as a traffic shortcut. Google generally limits FAQ rich results to authoritative health and government sites.
Breadcrumb JSON-LD should describe a navigation path users can understand on the page. Hidden, invented trails create weak and potentially misleading markup.
Article markup can remain valid without an image, but omitting a representative image reduces completeness and can limit how the page is presented in search.
The two titles can vary for social context, but they should identify the same page and promise. Keep the subject stable and test generated previews before publishing.
Use Person when a named individual wrote the article and Organization when the editorial organization owns the byline. Keep visible and structured authors identical.
Keep the original publication date and update dateModified only after a meaningful content change. Align visible dates, Article JSON-LD, and sitemap lastmod.
The data-nosnippet attribute can exclude selected text from Google snippets. Apply it to supported elements, keep it stable in rendered HTML, and preserve answer conditions.
Tabbed content can be indexed when it exists in rendered HTML, but important answers should remain easy to find, deep-link, and read without fragile interaction.
Use anchor text that names the destination and fits the surrounding sentence. Contextual links help readers and crawlers understand why the article is the next useful page.
Two equally prominent headings can make the main page title unclear. Give one visible H1 stronger hierarchy and make secondary promotional copy visually subordinate.
Duplicate descriptions do not create a standalone penalty, but generic copy makes similar pages harder to distinguish. Generate accurate summaries from page-specific facts.
Google often builds a query-specific snippet from visible page content. Give the tool a clear server-rendered answer and write metadata that accurately summarizes that content.
Google builds title links from the title element, visible heading, prominent text, links, and other signals. Align those sources around one accurate page promise.
Sitemap lastmod should reflect the page's last significant content change, not the latest deployment. Stable dates make the signal useful and keep freshness claims accurate.
A sitemap can expose an orphan article URL, but contextual internal links show readers and crawlers where the page belongs. Add links from relevant hubs and adjacent guides.
Every old article URL should reach its closest replacement in one permanent redirect. Chains add latency, complicate crawling, and preserve migration mistakes longer.
A route can return HTTP 200 while its visible content looks like an error, empty shell, or missing resource. Give valid tool states a complete server-visible purpose and response.
A live noindex directive takes effect after a crawler can fetch and process the page. Check access, response headers, canonical signals, last crawl time, and duplicate URLs.
Robots.txt can prevent crawling, but Google may still know and index the URL from links. Use noindex, authentication, removal, or a proper status based on the real goal.
Keep useful filters available to visitors while consolidating duplicate parameter combinations. Decide which URLs deserve indexing, then align links, canonicals, and crawl controls.
Google may choose another canonical when duplicate pages send conflicting or incomplete signals. Align redirects, canonical tags, sitemap URLs, and internal links.
A sitemap helps Google discover a canonical URL, but it does not require Google to crawl or index it. Diagnose access, duplication, page value, and internal links separately.
An indexed page can show zero impressions when it is eligible for Search but has not matched a measured query. Check eligibility, intent, demand, and evidence in that order.
