Schema and technical SEO for blog posts

Technical SEO makes sure a post can be fetched, parsed and trusted: a page that crawlers can't read doesn't exist for search or for AI. Get the structured data, the head, crawler access and speed right before the content can do its work.

Structured data

Structured data is a block of code (JSON-LD) that tells search engines what the page is, who wrote it and when.

  • Every post emits BlogPosting JSON-LD: an @id of the URL plus "#article", headline, description, inLanguage, datePublished, dateModified (the real update date, or the publish date), author as a Person, editor as a Person, publisher pointing at the Organization @id, mainEntityOfPage and image. The fact-checker goes on the page's WebPage node as reviewedBy, with lastReviewed. Escape "<" in the output.
  • Use only schema types that describe what is visibly on the page: BlogPosting, BreadcrumbList, FAQPage where a visible FAQ exists, VideoObject where a video is embedded. Never markup added only for search.
  • FAQPage only from the visible FAQ text, in every language. FAQ rich results now show mainly for well-known government and health sites (Google, August 2023), but the markup still describes the questions, and the visible text is what gets quoted.
  • BreadcrumbList mirrors the visible breadcrumbs.
  • About and author entities: AboutPage and ContactPage types, an Organization with founder and employee entries, Person schema on bio pages, bylines linking to the author's bio page, and sameAs only to accounts that exist. Author pages list the person's articles.
  • Never AggregateRating or self-serving Review markup for the business itself.
  • Structured data must match visible, verified facts, and stay in step with the copy and the site settings (address, hours, areas, phone). Google decides on a rich result; valid schema alone never earns one.

The other page types on the site carry their own markup.

  • For a local service business: the homepage carries Organization plus the trade type, with an @id of the domain plus "#organization", the address, areaServed (an administrative area where that's what it is), opening hours, price range and the telephone. Each service page carries one Service node with the provider @id, areaServed and the exact canonical URL.
  • For an online store: Product and Offer markup belongs on product pages. A post with product cards keeps BlogPosting and doesn't repeat the products' Product markup.
Structured data graph of a blog post: a BlogPosting node linked to its web page, breadcrumbs, publisher, author and editor, and a visible FAQ.
Every node describes something visible on the page; markup that describes nothing visible never ships.
Embed this image
<a href="https://rend.ee/en/blog/seo-blog-guide/technical-seo-for-blog-posts#fig-schema-graph"><img src="https://rend.ee/images/guides/seo-blog/en/blog-post-structured-data-graph.webp" alt="Structured data graph of a blog post: a BlogPosting node linked to its web page, breadcrumbs, publisher, author and editor, and a visible FAQ." width="1600" height="1117" style="max-width:100%;height:auto"></a>

Head and indexing

  • Every post has a self-referencing canonical, reciprocal hreflang alternates only for languages that exist, x-default pointing at the main language, og:type article, the published time, and a large-image social card.
  • Open Graph per page, with og:locale for each language and its alternates. og:image and twitter:image (with the summary_large_image card) point to the page's own 1200 × 630 share image, with width, height and alt. The image URL stays public even when the page is gated.
  • Keep the blog index "noindex, follow" until it has posts. The sitemap lists the index and the posts only once posts exist.
  • sitemap.xml lists language alternates and lastmod; use a sitemap index once a blog exists.
  • One trailing-slash policy per site, with a 301 from the other form.
  • A real 404 status, 410 for retired URLs, and a static 500 page.
  • Indexing doesn't cascade. Request indexing per URL in Search Console, which caps the requests per day, so key pages go first.
  • No crawl traps: clean canonicals with nothing missing or broken, and no faceted navigation that spawns endless URLs (a store's rules are in chapter 14).

Crawler access

  • Check crawler access before you touch content. A common reason a site never gets cited is a CDN or firewall rule silently blocking GPTBot, ClaudeBot, PerplexityBot or OAI-SearchBot. Check the firewall, bot protection and robots.txt across the whole hosting stack, not only the files you edit. A page nobody is allowed to crawl can't be the answer.
  • robots.txt lets in search and answer crawlers (Googlebot, Bingbot, OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-SearchBot) and decides on training tokens (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended) separately (chapter 11). It blocks only costly dynamic paths. Staging hosts send a noindex header.
  • Crawler access overrides bot protection. When a security setting gets in the way of search and AI crawlers, the crawlers win: they reach every page as easily as possible. Form protection never blocks crawlers.
  • Render on the server: prices, facts and FAQ answers are in the initial HTML. After publishing, fetch key pages without JavaScript, as GPTBot and ClaudeBot, to prove it.
  • When your site moves to a new domain, every SEO setting moves with it: URLs, redirects, canonicals, sitemaps and structured data.
  • Change a phone number everywhere at once: the copy, the menus, the footer and the telephone field in the structured data.
Grid of eight search and AI crawlers a blog's robots.txt should allow, from Googlebot and Bingbot to OAI-SearchBot, PerplexityBot and ClaudeBot.
A page an answer bot can't fetch can't be the answer; check the firewall and the CDN as well as robots.txt.
Embed this image
<a href="https://rend.ee/en/blog/seo-blog-guide/technical-seo-for-blog-posts#fig-crawler-access"><img src="https://rend.ee/images/guides/seo-blog/en/crawlers-robots-txt-should-allow.webp" alt="Grid of eight search and AI crawlers a blog's robots.txt should allow, from Googlebot and Bingbot to OAI-SearchBot, PerplexityBot and ClaudeBot." width="1600" height="771" style="max-width:100%;height:auto"></a>

Performance

  • Mobile Lighthouse performance 90 or higher, accessibility 100, layout shift under 0.05, and a hero image of 200 KB or less at phone width.
  • Page assets protect Core Web Vitals: explicit image dimensions, compressed and lazy-loaded images, minimal render-blocking code, and video embeds as a lightweight facade that loads the player on click. The progress bar and the sticky table of contents never shift the layout.
  • Self-hosted, preloaded fonts with the character subsets your languages need (Cyrillic, for example).
  • Cookieless analytics, so no cookie banner is needed.

Verify on the live page

A post is technically done when the live page proves it, not when the build passes.

  • Fetch the page without JavaScript and confirm the H1, the first sentence, the prices, the figures and the FAQ answers are in the HTML.
  • Run the structured data through a validator and compare every value with the visible text: the headline, the dates, the author and the FAQ wording match character for character.
  • Request indexing in Search Console for the new page and for the pages you changed to link to it, key pages first.
  • Open the sitemap and confirm the new address is listed with its language alternates and the right last-modified date.
  • Check the response headers: a 200 status, no noindex header left from a staging setup, and a cache rule that lets a fixed typo reach readers the same day.

Why the technical work comes before the text

A strong post on a page a crawler can't fetch loses to a weak post on an open one. So the order is fixed: access, structured data, the head, speed, and only then the text. The rest of the guide builds on this base. The internal links in chapter 9 do nothing if the crawler never reached the page, the language versions in chapter 10 work only with correct hreflang, and a citation in an AI answer (chapter 11) starts with the AI crawler fetching the page and finding the facts in its HTML.

Repeat this chapter's checks after every big change to the site: a new host, a new CDN, a new theme or a security plugin. Any of them can quietly block crawlers, and you only find out weeks later from falling impressions in Search Console. A short check on the day of the change costs less.

  • Andres Lepik, co-founder and head of growth at REND, with a short beard, smiling in a black REND cap and T-shirt against a light grey backgroundEdited by Andres LepikCo-founder & Head of Growth
  • Siim Saar, head of SEO at REND, with stubble, in a black REND cap and T-shirt against a light grey backgroundFact-checked by Siim SaarHead of SEO
Rasmus Kivi, co-founder and CEO of REND, smiling in a black REND cap and T-shirt against a light grey background

Post by Rasmus Kivi

Co-founder & CEO

Co-founded REND and still reads every client's monthly report himself before it goes out.