How to get found in 2026: SEO, AI visibility and site speed

For fifteen years, "getting found" meant one thing: rank on Google. In 2026 it's three overlapping things — search engines, AI assistants, and the page speed that quietly gates both. Get one right and miss the others and you're still invisible: a beautifully written page a crawler can't reach, a fast site an AI can't understand, a perfectly optimised page nobody waits for because it takes six seconds to load.

Here's how the pieces actually fit together, and the order to fix them in. This is the map; where a section has a full deep-dive, I'll point you to it.

1. First question: can they even reach you?

Before anything can rank you or cite you, a crawler has to be allowed in and able to load the page. Sites fail this in two quiet ways.

A robots.txt that shuts the door. One line can tell a crawler to leave — and the most common version accidentally blocks the AI crawlers while letting Google through, so you vanish from ChatGPT and Perplexity without realising it. This is worth checking on purpose, not by accident — I covered exactly how a site goes invisible to AI and what to look for.

Content that only exists after JavaScript runs. If the words that matter are painted in by a script after load, a crawler reading the raw HTML often sees an empty shell. Whatever you need found — your headline, your copy, your product details — should be in the HTML the server sends, not assembled later in the browser.

2. Can they understand you?

Reaching the page isn't enough. Now the crawler — search engine or AI — has to make sense of what's there.

• Semantic structure. One clear <h1>, a logical heading order, real HTML elements. This is the skeleton both Google and an AI use to work out what the page is about.
• A unique <title> and meta description per page. Still the single highest-leverage on-page signal, and still the thing most sites get lazy about — one generic title copied across the whole site throws away free ranking.
• Structured data (JSON-LD). Schema markup tells a machine what the page is — an article, a product, an FAQ, an organisation — instead of making it guess. It's how you earn rich results and how AI answers pull clean facts about you.
• Content clear enough to quote. AI assistants cite sources they can extract a straight answer from. A rambling, vague page doesn't get quoted; a page that answers a real question plainly does. Write to be the sentence the AI lifts.

3. The speed that decides it

Speed isn't cosmetic. Google measures it on real visitors and uses it to rank you — and it's the same speed that decides whether the visitors you do get stick around long enough to convert. Slow pages rank lower and lose the people who arrive: the penalty compounds.

The three numbers to know are LCP, CLS and INP — what each means, what to aim for, and how to fix the usual culprits is its own piece: Core Web Vitals, the three numbers Google cares about.

4. The unglamorous foundations

Most sites half-do these. Finishing them is cheap and pays off:

• A sitemap.xml, and a robots.txt that points to it — so crawlers find every page, not just the ones they stumble onto.
• Canonical URLs, so the same content on two addresses doesn't split its own ranking.
• HTTPS everywhere — a baseline trust signal, and non-negotiable for ranking.
• Works on mobile — Google indexes the mobile version of your site first, not the desktop one.

5. What moves the needle — and what doesn't

A lot of visibility advice is noise. The short version:

• Myth: llms.txt gets you cited by AI. As of 2026, no major AI crawler officially uses it. Add it if you like — it's harmless — but it isn't the lever, and I explained why.
• Myth: more keywords, higher rank. Keyword-stuffing is a relic. Both Google and AI now reward clarity and usefulness, and penalise the opposite.
• What actually works is unglamorous and durable: be reachable, be understandable, be fast, and genuinely answer the question someone typed. That's the whole game — it just doesn't fit on a growth-hack thread.

How to check where you stand

You can walk all of this by hand — read your robots.txt, view source, check your titles, run a speed test. Or Reconvio checks the visibility signals (crawlability, AI-crawler access, meta tags, structured data) and the speed metrics (via Lighthouse) in one pass, and tells you in plain language what's missing and what it's costing you.

---

One caveat worth ending on: being found is necessary, not sufficient. A fast, visible site that leaks data or breaks compliance law is a bigger liability, not a smaller one — you've just made the problem easier to find. Get seen, but get the other two right first.