Jana Legaspi

Technical SEO for the AI Era You want your content to show up inside AI answers. Cool. Then we need to talk about the least sexy part of SEO… the part that quietly decides whether you exist: Crawling + indexing + structured data. Because here’s the truth: AI answers don’t “discover” your site. They retrieve from indexes. And indexes only contain what bots can crawl, render, understand, and store. The new game isn’t “rank #1.” It’s “be eligible to be used.” AI features in search engines are still built on the same foundation: crawling, indexing, and serving. If you’re not crawled and indexed correctly, you’re not eligible to be cited: no matter how good your content is. Even worse: AI-style retrieval often fans out across related subtopics, which means your supporting pages and your internal linking matter more than ever. Part 1: Crawlability: Can the bot even get in the building? If crawling is blocked or hindered, everything else is theater. Crawlability failures are usually self-inflicted: robots rules, fragile servers, infinite URL traps, or content locked behind logins. Crawlability checklist (the boring stuff that prints money) Robots.txt isn’t sabotaging you (and you’re not blocking CSS/JS your site needs to render). Your server isn’t screaming “go away” (watch 5xx errors and timeouts in logs). Your important content isn’t behind a login or paywall the bots can’t access. You’re not generating infinite URL garbage (facets, parameters, calendars, session IDs). Part 2: Indexability: Even if crawled, will it be stored? Indexing is where search engines decide what your page is about, whether it’s a duplicate of something else, and which version becomes canonical. Crawled does not automatically mean indexed. Indexability checklist No accidental noindex (meta tags, HTTP headers, CMS defaults). Canonicals aren’t lying (don’t canonicalize everything to the home page; don’t point to the wrong URL). Duplicate versions are handled intentionally (HTTP/HTTPS, www/non-www, trailing slash, parameters). The important content is actually present as text (not only in images; not hidden behind broken JS). Part 3: Freshness:  AI answers punish stale pages quietly If you want to be cited, you want the engine crawling the current version, not last month’s snapshot. Faster discovery of updates can matter, especially on engines that support rapid URL submission. The fastest freshness lever many sites ignore: IndexNow IndexNow is a ping that tells participating search engines a URL was added, updated, or deleted, so they can recrawl it sooner. It doesn’t guarantee ranking, but it can shrink the “found it later” delay. Basic idea: Generate an IndexNow key (and host it on your site). When a URL changes, ping the endpoint with the updated URL (or submit a batch list). Use it for additions, updates, and deletions, especially if your site changes often. What about Google’s Indexing API? Google’s Indexing API is not a general purpose “index my blog faster” button. It’s intended for specific page types (notably job postings and live streams). For most sites, you still win with clean architecture, strong internal linking, sitemaps, and technical health. Part 4: AI crawling isn’t one bot: it’s multiple bots with different goals In the AI era, you’re not choosing “block bots or not.” You’re choosing what kinds of bots you allow, and for what purpose (search visibility vs training). Different companies publish different user agents and controls. Practical robots.txt pattern (example): show up in AI answers, don’t feed training Strategy example (adjust to your legal/commercial preferences): User-agent: OAI-SearchBot Allow: / User-agent: GPTBot Disallow: / User-agent: Googlebot Allow: / User-agent: Google-Extended Disallow: / Why this pattern exists (high level): Allow search-focused crawlers so your pages can be retrieved/cited. Block training-focused crawlers if you don’t want your content used for model training. Remember: blocking in robots.txt can prevent a crawler from seeing your noindex/meta rules, so choose intentionally. Part 5: Structured data, label the world so AI doesn’t guess Structured data doesn’t guarantee special treatment, but it reduces ambiguity. It helps machines connect your pages to entities (brand, authors, products) and extract key facts without guessing. The structured data stack that tends to matter most Organization (or LocalBusiness): name, logo, URL, sameAs profiles. WebSite + WebPage: connect pages back to the site and publisher. Article/BlogPosting: headline, author, publish/modify dates. Product + Offer (ecommerce): price, availability, identifiers (GTIN) when available. BreadcrumbList: reinforce site structure. Simple JSON-LD pattern (example) <script type=”application/ld+json”> {   “@context”: “https://schema.org”,   “@graph”: [     {       “@type”: “Organization”,       “@id”: “https://example.com/#org”,       “name”: “Example Co”,       “url”: “https://example.com”,       “logo”: “https://example.com/logo.png”,       “sameAs”: [         “https://www.linkedin.com/company/example”,         “https://x.com/example”       ]     },     {       “@type”: “WebSite”,       “@id”: “https://example.com/#website”,       “url”: “https://example.com”,       “name”: “Example Co”,       “publisher”: { “@id”: “https://example.com/#org” }     },     {       “@type”: “WebPage”,       “@id”: “https://example.com/ai-seo/#webpage”,       “url”: “https://example.com/ai-seo/”,       “name”: “Technical SEO for the AI Era”,       “isPartOf”: { “@id”: “https://example.com/#website” },       “about”: { “@id”: “https://example.com/#org” }     }   ] } </script> Part 6: “Structured data + clean HTML structure” is what gets you quoted If you want to be cited, make your pages quote-ready: clear headings, scannable lists, and tables where appropriate. Don’t bury key facts in UI elements that fail to render for crawlers. See Also: Structured Data for AI Answers: Entity Hygiene & JSON-LD Patterns Part 7: Measure AI visibility like an adult (not with vibes) Traditional SEO diagnostics still matter (index coverage, crawl errors, canonical issues). On top of that, track referral traffic from AI surfaces and watch citation features in webmaster tools where available. Action plan Today (60-120 minutes) Check robots.txt for accidental blocks (especially resources needed to render). Pick 5 target pages → verify 200 status, indexable, canonical correct. Add/clean Organization + WebSite + WebPage schema on core templates. This week (half day) Fix duplication/canonical clusters that split signals. Improve internal linking to the pages you most want cited (support pages matter). Implement IndexNow if you publish/refresh frequently and care about Bing/Copilot discovery. This month (1-2 days) Add structured data for your content type (Article/Product/etc.) and validate it. Create quote-ready sections: headings, bullets, tables; make key facts obvious. Set up measurement for AI referral traffic and any available citation reporting. See Also: Measuring AI Visibility: Crawls, Indexing & AI Citations Conclusion You don’t “optimize for AI” by stuffing prompts into HTML. You optimize for AI by making sure bots can crawl you, engines can index you correctly, your pages are eligible to be shown, and your structured data matches reality. Further reading: Google Search Central: How Search works  Google Search Central: AI features and your website  Google Search Central: Intro to structured data  Google Search Central: Structured data policies  Google Search Central: Google common crawlers (Google-Extended)  OpenAI: Robots.txt and crawlers (GPTBot / OAI-SearchBot)  OpenAI: Publishers & developers FAQ  IndexNow IndexNow documentation Bing blog: Introducing Copilot Search in Bing  Bing blog: AI Performance in Bing Webmaster Tools OpenAI: Introducing ChatGPT search  Recent coverage The Verge: Google’s AI search results will make links more obvious  The Guardian: Google puts users at risk by downplaying health disclaimers under AI Overviews  Wired: How to hide Google’s AI Overviews from your search results