Open ChatGPT and ask it something. Look at the answer. Notice what's missing?
No list of ten blue links. No, you're number three. No weekly score to obsess over.
For many years, marketing ran on that score. You picked your keywords, watched your position on Google, and tried to climb. Straightforward enough.
AI search blew that up. And a lot of sharp people are still trying to play the old game on a board that no longer exists.
Why Google felt easy to track
Google is predictable. Ask the same question today and tomorrow; you get the same results in the same order. Number one stays number one. Because it held still, you could measure it, report on it, and fight for it. That's why rank-tracking tools became a whole industry.
AI search has none of that predictability.
What an AI actually does when you ask it something
When you type a question into an AI, it doesn't grab the top Google result and hand it back. It does something messier.
It quietly breaks your question into several smaller ones. Ask, "What's the best tool for a small team?" and behind the scenes, it might also be exploring "tools for startups", "affordable options for small teams", and a few others it invented on its own. It finds decent answers to each of those, then blends everything into one reply and names a handful of sources.
Here's what that means for you: you can be number one on Google for the exact phrase someone typed, and the AI still won't mention you because it's off answering questions it generated itself.
There's something else worth knowing. The AI writes a fresh answer every single time. Ask the same question twice and you can get two different answers with different brands named. That's not a glitch. That's just how these tools work.
The proof there's no ranking
A research team led by Rand Fishkin at SparkToro tested this properly. They ran nearly three thousand questions through ChatGPT, Claude, and Google's AI – with real people asking the same things.
The result: it was rare for two separate runs to name the same brands, and almost unheard of for those brands to appear in the same order.
If the list keeps changing, there is no "spot" to hold. So if anyone promises you a fixed ranking position inside an AI answer, be sceptical. The AI doesn't produce one.
What it produces is more like a batting average. Your brand might appear in 7 out of 10 answers, while a competitor appears in 4 out of 10. That gap is real and worth tracking. It's just not a rank; it's how often you show up.
What to measure instead
Mention rate. Pick a set of questions your real customers ask. Run each one many times, at least twenty runs, before you take the number seriously. One run tells you nothing because the answers shift constantly. Check weekly. Daily is too noisy; monthly and you'll miss what's actually moving.
Share of the answer. Did the AI pull one line from you, or did half the response come from your content? There's a big difference. This is one of the clearest signals of how much an AI actually leans on you, and almost nobody checks it.
Brand recognition. Does the AI know who you are? Not just your website, but your presence across the sources AI tools tend to trust: Wikipedia, review platforms, news coverage, and industry directories. When those all describe you consistently, you start getting mentioned more, even if you haven't touched your own site.
Competitor comparison. Showing up in 4 out of 10 answers sounds decent, until you find out your main rival is in 7 out of 10. This number tells you whether you're winning or just present.
Tone of mention. Appearing is good. Being praised is better. If the AI keeps saying "They're fine for small teams but not as strong as the others," that quietly puts buyers off even though you got a mention. When that happens, trace where the negative framing comes from (usually an old review or a comparison article) and fix it with clearer, stronger content.
A few terms you'll keep seeing
GEO, AEO, LLMO, and AI SEO; they all mean roughly the same thing: getting AI tools to mention your brand when customers ask questions. Don't let the jargon trip you up.
The one phrase actually worth knowing is 'your question set'. That's the list of real customer questions you track over time. If that list is full of your own brand name, your results will look great and mean nothing. Use the questions real buyers ask before they've ever heard of you.
Why chasing a "rank" tends to backfire
You end up promising something you can't deliver. You report a position, it shifts the following week for no obvious reason, and people stop trusting your numbers. Most teams give up right before this kind of work actually starts paying off.
You also end up doing the wrong work, optimising for questions where you already appear instead of the ones where new customers might find you for the first time.
And you misread what the game even is. On Google, there's one number one, and someone has to lose their spot for you to take it. In an AI answer, several brands can be named in the same reply. You and a competitor can both show up together. You're not trying to knock anyone off a pedestal. You're trying to become a name that the AI reaches for naturally, across many different questions. That takes time and consistency, not a single clever move.
How to actually track this
Start by writing down thirty to fifty questions your customers genuinely ask, in their own words, not your brand name. Good places to find them: the "People also ask" section on Google, Reddit, and Quora.
Run each question at least twenty times before drawing any conclusions. This is the step most people skip, and it's why most reports end up misleading. You can use tools for this or a simple spreadsheet. Check weekly.
Then connect it to things you already trust. When you start showing up more in AI answers, you'll often see more people searching your brand name directly on Google. They heard about you from the AI and wanted to know more. That's your signal, it's working. You can also look at whether visitors arriving from AI tools convert better than your usual traffic. They often do, because the AI already recommended you before they got there.
On timing: quick wins and clearer answers, a solid FAQ section can show up within a month or two. Getting AI tools to genuinely recognise your brand takes longer, somewhere between three and six months. Track AI numbers alongside your normal results, not separately from them.
The bottom line
In AI search, there's no ladder to climb. There's only a reputation to build.
Either AI tools see your brand often enough, across enough credible sources, to confidently bring you up when someone's asking, or they don't.
So the question worth asking isn't "Where do I rank?" It's "How often does the AI trust me enough to mention me when my customers are looking?"
Build that now, and you'll be the brand AI keeps recommending for years.
Digital MindFlow helps brands move from guessing to growing, with clear, data-led AI and digital strategy.