Ranking #1 Used to Mean Something. In 2026, It Doesn’t

I’ve read about forty “state of SEO” articles this year. They all say the same three things. AI is changing everything. Write helpful content. Build your brand. Then a list of tips you’ve seen a hundred times.

Skip all that. I want to tell you about the parts of search that quietly stopped working the way you were taught, and nobody seems to want to explain in plain words. Some of this will annoy you if you’ve been doing SEO a long time. Good. The stuff that annoys you is usually the stuff you need to hear.

Let’s start with the thing that broke the most.

Being number one doesn’t win the way it used to

For twenty-five years there was one scoreboard. Where do you sit in the list of blue links. Top of the list got the clicks. Bottom got scraps. Everybody clawed toward the top because the top was the whole game.

Here’s what barely anyone says out loud. That list isn’t where a lot of answers come from anymore. When someone uses Google’s AI Mode, or just sees that AI summary box at the top, the answer isn’t lifted off “the page that ranks first.” It gets built out of bits and pieces of loads of different pages. And the pages it borrows from are very often not the ones sitting at the top of the normal results.

A study late last year checked which pages actually got quoted in AI answers. About 68% of them were sitting outside the top ten. Sit with that. Two out of three sources the AI picked were pages that, under the rules you know, you’d have called losers. Page two. Page three. Nowhere near the top. The machine grabbed them anyway.

If you’re new, this is great news. You don’t have to knock the giant brand off position one to get seen. If you’ve been doing this for years, it should bother you a bit, because that rank tracker you’ve stared at every single morning is now measuring half a game. You can sit at number three and be completely missing from the AI answer. You can sit at number fifteen and get quoted in it. Those are two different fights now, and most people are only tracking one.

So why would a machine skip the top results and reach down to page two? You have to see how these things actually go looking for stuff. It’s weirder than people think.

One question turns into about a dozen

Type a question into an old-fashioned search engine and it hunts for pages with your words on them. One question in, one list out. Simple.

AI search does something else. The nerds call it query fan-out, which sounds fancy but isn’t. You ask one big question. Behind your back, the system chops it into a handful of smaller, sharper questions, runs all of them at the same time, grabs the answers, and glues them into one reply.

Think of asking a mate who happens to be a brilliant researcher: “what’s the best face wash for a teenager with oily skin?” An old search engine goes looking for that exact phrase. The AI doesn’t. It quietly fires off a little cluster of questions you never see. What ingredients are good for oily skin. What’s gentle enough for a teenager. What do the dermatologists say. What do the reviews actually report. It answers each one, then stitches it all together into a neat little reply for you. People reckon one proper question can split into roughly eight to twelve of these hidden ones.

That’s the trick behind the surprise above. Your page might be average at the big, popular question everyone’s fighting over. But if it happens to be the single best answer on the whole web to one of those small hidden questions, the machine takes it. Doesn’t matter that it never ranked near the top for the main term. You didn’t win the headline fight. You won one of the twelve quiet ones. That was enough.

Now think about what that does to your strategy. The old habit was to dump everything into one page, polish it to death, aim it at one juicy keyword. Fan-out makes that look daft. No single page can be the best answer to twelve different questions at once. What wins now is range. A bunch of genuinely useful pages that, between them, cover the smaller questions a topic naturally breaks into. Knowing a subject properly beats one trophy page buffed to a shine.

And here’s the bit that should make you a little uneasy, in a useful way. You can’t see the twelve questions. Google won’t show you. You’re optimising for questions you have to guess, which means you have to know your subject so well that you’ve already answered what a curious person would ask next. That’s not a keyword trick you pull on a Tuesday afternoon. That’s actually knowing your stuff, which is dead hard to fake, which is exactly why it works.

The old deal between sites and bots is dead

Here’s a thread almost nobody outside the technical crowd is talking about, and it changes the maths of running a website.

For the whole history of search there was a handshake deal. We let Google’s bot read our pages. Google sent us visitors back. The reading cost us a sliver of server effort. The visitors paid us back many times over. Fair. Everybody got it, even if nobody said it.

AI bots ripped that deal up. The numbers are brutal. Cloudflare, the company a huge chunk of the web runs through, looked at this. One big AI bot was reading websites somewhere north of a thousand times for every single visitor it sent back. Another was far, far worse than that. The bots hoover up piles of your content to build their answers and send you next to nothing in return, because the answer gets handed to the user inside the AI, where your page never gets the click.

If you run a small site, you feel this somewhere specific. Your bill. Every one of those bot visits burns your server. When swarms of them batter a site, it slows down for the actual humans trying to use it, and a slow site is a worse experience and a weaker performer in Google’s own eyes. So now you’ve got a decision that didn’t even exist a few years back. Which bots do you let in?

That’s why a sleepy little file nobody thought about is suddenly causing arguments. It’s called robots.txt, a plain text note at the root of your site that tells bots where they can and can’t go. It’s always run on trust. A polite bot obeys it. A bot that fancies ignoring it just reads your site anyway. For normal search that was fine, because Google played nice and paid you in traffic. With AI bots that take loads and give back crumbs, trust starts to feel a bit thin, and site owners are getting dragged into questions that used to belong to the security team. Who’s allowed to read our stuff, and what’s it worth to let them?

There’s even a proposed new file, llms.txt, meant to tell AI systems what to do. Whether it ever matters is a proper row, and hardly anyone uses it yet. I’m not going to tell you to drop everything and add it. The file isn’t the point. The point is that, for the first time, people who run websites are treating AI readers as a separate audience with their own rules, apart from human visitors and apart from old search. That’s a real change in how people think, even if that one file turns out to be nothing.

Your product spreadsheet is now your shopfront

Here’s one that sounds boring and absolutely isn’t, especially if you sell things.

For years the spreadsheet that feeds Google Shopping, the titles, the descriptions, prices, categories, what’s in stock, got treated as somebody else’s job. It lived with the paid ads team. The SEO lot wrote the nice category pages and never went near the feed. Two separate worlds that didn’t talk.

That wall’s coming down fast. As AI starts recommending products straight to people, the question stops being “does my category page rank.” It becomes “when a machine builds a list of products to suggest, am I even allowed in the room?” And whether you’re allowed in comes down to that dull little feed. Missing titles. Sloppy descriptions. Broken categories. A price in the feed that doesn’t match the price on the page. Every gap is a reason the machine leaves you off the list when it builds its recommendation. The feed went from back-office plumbing to the thing that decides whether you show up in AI shopping at all.

If you’re a beginner selling online, the lesson’s almost too simple. Before you fret about blog posts, make sure the basic facts about your products are complete, correct, and the same everywhere they appear. If you run a bigger team, the lesson’s about people. If your SEO crew and your feed crew still don’t talk, that silence is now a visibility problem, not just a messy org chart.

A strong brand is the closest thing to insurance there is

Every few months a Google update rolls through and the forums fill up with the same wail. My traffic’s gone and I don’t know why. I’ve watched this cycle for years. There’s a pattern in it that the panic tends to bury.

The sites that get hammered worst are nearly always the ones with loads of search traffic and nothing else. No visitors who come direct. Nobody typing their name into Google on purpose. No audience that would even notice if they vanished overnight. They built the whole thing on ranking and nothing else. So when the algorithm shifts, there’s nothing holding them up, and down they go.

The sites that ride it out share one thing. People actually look for them by name. They’ve got an audience that turns up direct, a reputation that exists off their own website, other people mentioning them and linking to them without being asked. Google reads all of that as proof you’re a real thing people care about, not just a clever page that talked its way up the list.

This puts the oldest advice in the business in a different light. “Build your brand” normally sounds like a motivational poster in a meeting room. Through the lens of updates it’s something much harder-edged. A strong brand is the cushion that takes the hit when the algorithm changes its mind. The good news, if you’re starting out, is that you don’t build that by fooling anyone. You build it by being good enough that people remember you and come back. If you run a budget, it’s a straight trade. Some of the money you’d have spent squeezing out one more ranking is better spent making people want you specifically, because that’s the one asset an update can’t easily strip off you.

So what is SEO now, really?

For a long time SEO was basically a game of matching keywords and climbing one list. That game still exists, and it still matters. It’s just been quietly knocked down from “the whole thing” to “one part of a bigger thing,” and a lot of people haven’t updated their habits to match.

What wins now is harder to fake, and I think that’s the whole story. Real depth on a topic, because the fan-out is hunting for the best answer to small questions you can’t even see. Clean, correct, well-organised facts about who you are and what you sell, because machines read your site now to decide whether to put you forward. A reputation that lives beyond your own pages, because that’s what’s still standing after the next update. And a bit of technical nous about which automated readers you let through the door and what your content’s worth to them.

None of this is a hack. It’s closer to just running a decent business and making sure the machines can tell that you do. That cited-page study was really pointing at one thing. The shortcuts are dying off, and being properly useful is becoming the most reliable optimisation left.

I know that’s not what everyone wants to hear. If you built your whole approach on tricks, this is a rough read. If you’ve been doing the real work for years and watching thinner sites outrank you, it’s about to get fairer. I’d take that trade in a heartbeat.