How ChatGPT rewrote the search rulebook (and what you need to do about it)

How ChatGPT rewrote the search rulebook (and what you need to do about it)

How ChatGPT rewrote the search rulebook (and what you need to do about it)

Audrey Hepburn being surprised

by Ema Fulga

Ema is an AI Search Content Strategist and GEO (Generative Engine Optimisation) expert. She's also the founder of decipher., an AEO agency that helps brands appear where people are now searching: AI-powered platforms like Perplexity, ChatGPT, Gemini, and others. With a background in copywriting and creative strategy, she’s on a mission to turn messy messaging into clear and structured content that helps brands get mentioned and cited in AI searches.

Connect with Ema.

Last updated: 07.09.2026

TL;DR: ChatGPT quite literally rocked the world. Not just from the user’s perspective, who now has the power of AI at their fingertips, but also for brands who had been using traditional search to reach customers for years. OpenAI has pointed out multiple features of their system which means their shiny new version of search doesn't work like Google. It rewrites what people type before it ever goes looking for answers. It won't touch your site unless you let its crawler in. It hands out inline citations instead of blue links. If a user has memory switched on, it personalises the entire search before your brand even gets a look-in. Old SEO instincts won't save you here. New rules, new game.

For twenty-odd years, "getting found" in search meant one streamlined process: rank for the right keywords, land on page one, hope for the click. Now your traditional Google search isn’t the only option. Search is happening more than ever before on AI tools. ChatGPT was the one that kicked off this whole revolution. Along the way, ChatGPT search has quietly binned the rulebook and written a new one. Where the opportunity lies is that most brands haven't noticed this yet.

OpenAI's own help centre lays out, in plain terms, how ChatGPT search actually works behind the scenes. Read between the lines and it's basically a checklist of everything we didn’t have to think about with traditional SEO and a lot of what we need to do differently for GEO.

Here's what changed, and what to do about it.

1. Your visibility starts or ends with crawlability

One way OpenAI is the same as Google is the slyness. They're not giving it all away. They have said there's no way to guarantee top placement in ChatGPT search, full stop. But before you can even be in the running, your site needs to let OAI-SearchBot crawl it, and your host or CDN needs to allow traffic from OpenAI's published crawler IPs.

Translation: if your firewall, CDN, or a forgotten robots.txt rule is quietly blocking OAI-SearchBot, you're not being overlooked by ChatGPT. You're invisible to it. This is the unglamorous, unsexy foundation nobody wants to talk about, but everyone needs to do.

Quick check:

  • Is OAI-SearchBot allowed through your CDN and firewall?

  • Does your robots.txt actually welcome it, or quietly wave it off?

  • Are your published IP allowlists up to date?

Sort this first. Everything else in this article is wasted effort otherwise.

2. Query rewriting just killed keyword-matching

I, for one, am so grateful for this! The amount of time I spent trying to shoehorn phrases into content. No more! ChatGPT doesn't search for what your user typed word for word, overlooking any variations. It searches for intention, i.e. what it thinks they meant.

OpenAI confirms that ChatGPT typically rewrites a person's question into one or more targeted queries before sending them off to search partners. Then, if needed, it will fire off follow-up, more specific queries based on what the first round turns up. So a vague, conversational question gets translated into something much sharper before it ever touches a search index.

That's a fundamentally different game to classic SEO, where you could reverse-engineer a keyword and build a page around it. ChatGPT is optimising for intent, not playing snap, which means your content needs to answer the underlying question thoroughly, from multiple angles, rather than chasing the exact phrase someone might type into a search bar.

Keyword research isn't dead. But keyword worship is. Your content needs to align with queries, both how your customers would word the question and all the questions AI will ask in order to answer it. 

3. Inline citations are the new page one

In classic search, visibility meant a blue link. In ChatGPT search, it means being the source the model chose to name.

Responses that use search can include inline citations you can hover over or click, and if those aren't shown directly, there's a "Sources" panel sitting right below the answer. This list of links is noticeably shorter than Google search results. There's no more "we rank #4, that's still fine." You're either the source ChatGPT trusted enough to name, or you don't exist in that answer at all.

That raises the bar considerably. To earn a citation, your content needs to look and read like something worth citing: clear sourcing, well-structured facts, credible authorship, and none of the vague fluff that search engines used to tolerate. AI doesn't cite content that sounds unsure of itself.

4. Memory means results are personalised

This is the one most brands haven't clocked yet. If a user has ChatGPT's memory feature switched on, it can fold what it already knows about them into the search query itself, like location, preferences, past context, before the search even runs.

OpenAI's own example is telling. Someone with memory turned on who's vegan and based in San Francisco asking for "restaurants near me that I'd like" doesn't get a generic local search. They get "good vegan restaurants San Francisco" rewritten and filtered before results are even fetched.

Stretch that logic across every category, and the implication is groundbreaking (and not in a florals in spring kind of way). There is no longer one "top answer" for a given topic. There’s a top answer for each individual based on what they've told ChatGPT about themselves, and what it's inferred along the way. Ranking for a single, generic query is a shrinking prize. Being the obvious, well-documented answer across a range of specific, differentiated angles is where the money is made.

So what does this actually mean for your brand?

Put the four things together and what do we get? Technical access gets you in the room, intent-matched content gets you heard, citation-worthy structure gets you named, and range gets you found across the different versions of the question people are actually asking.

None of that happens by accident. AI is quoting reliable sources, not keyword lists, which means the brands that show up are the ones that sound like something worth quoting.

If you want to know exactly where your brand is falling through these cracks, whether it’s crawl access, query coverage, citation-readiness, or the lot, book a free discovery call and we'll show you.