AI Isn't Ranking Your Brand. It's Forming An Opinion About It.
For over twenty years, SEOs have thought about visibility as a search problem. We learned about how search engines work, we optimized pages, built links and focused on rankings.
The last few years have seen an acceleration of technology that is impacting how users search. SEO and its foundations are still important, but the visibility layer with our audience and the channels they use for research are shifting.
Google is not the singular destination it once was and is now part of a blended user journey to solve their problems and answer their questions.
I spoke to Duane Forrester on IMHO recently, and he summarised it well:
“Do not approach AI and your desire, your need, to be a part of that answer in the same way you approached being a part of the answer in Google. They’re very mechanically different things.”
Some SEOs still have a mindset to do keyword research, rank and measure it, but SEO in 2026 requires a shift to understanding how your brand is being represented within machines and LLMs. And it’s reputation and not ranking that counts.
AI Is Building A Story About Your Brand
Most brands still think about visibility in terms of content. Publish more, create more resources, build topical authority. The assumption is that if you produce enough of the right material, AI will understand who you are and reflect that back.
But as Duane explained, that’s not how these systems work.
“Those LLMs are not just taking your word for it. They are also incorporating every instance where you are talked about or mentioned on the internet. So every time your competitor says something about you, every time a reviewer says something about you, all of that is being included. It’s a very large landscape and surface area that an SEO has to manage today.”
Your reviews matter, your media coverage matters, Reddit threads about you matter and what your competitors say about you matters.
In traditional SERPs, a negative article about your brand could sit alongside positive ones. And it was far easier to simply push one or two pages you didn’t want surfacing down page one. Reputations could be sculpted.
LLMs don’t do that. They synthesise and they form a consensus view, and that consensus is what gets returned as the answer.
That’s both useful and uncomfortable. Brands with strong reputations built over years now have an opportunity to benefit from all those positive signals being connected together, but it goes both ways:
“Every time your competitor says something about you, every time a reviewer says something about you, all of that is being included.”
A competitor no longer needs to outrank your content, they just need to influence the consensus. AI isn’t asking which page should rank first, It’s increasingly asking, “Do I trust this brand?” That trust assessment is being made about your brand right now, whether you’re contributing to it or not.
The Most Valuable AI Skill Isn’t Prompting
Recently, Duane has been writing about the difference between the execution layer and the judgment layer.
Execution = prompt in + output out + go act on it.
Most people are still stuck operating in this execution layer. They’re generating content, writing code, creating summaries, or building workflows, which are all useful, but it isn’t a differentiating factor.
Duane said, “If you’re using these systems, you have to be an excellent product manager, program manager, project manager, you have to excel at those skills. You being a programmer is a lower-needed skill for these platforms.”
In short, technical skill used to be the differentiator, but now it’s a baseline.
What matters is the ability to hold the totality of a project in your head, catch when an output doesn’t match what was agreed, and recognise when an answer is incomplete.
“The judgment layer is you expressing your judgment about this based on experience.” Dan explained.
It’s harder to teach because it depends on knowing what you don’t know. That’s the part nobody wants to talk about because it isn’t as exciting as the latest AI tool or prompt framework, but it’s also the part that matters most.
“The judgment layer is knowing when to say, hold on a second, this is not the complete answer. I actually have to dig deeper.
The quest for speed strips that away in so many instances. People think, ‘Oh, if I just write a better prompt, I’ll get the answer.’ And it’s never that simple or straightforward.”
The people getting the best results from AI aren’t necessarily the people writing the cleverest prompts. They’re the people who know when the answer is wrong, incomplete, or missing something important.
We Have The Data. We Don’t Have The Understanding Yet.
I asked Duane whether the industry actually has what it needs right now to make good decisions about AI visibility. His answer was yes and no.
Yes, because there are now platforms emerging that can help us understand how brands are appearing in AI systems.
Yes, because we have access to measurements that simply didn’t exist a few years ago.
Data that was never available before is now accessible. And in Duane’s view, the industry moved faster to build this tool infrastructure than it did in the early SEO days: “What took us, I would say, 10 years to actualize, the beginning of the SEO industry, probably took about 3 months to actualize with the beginning of this industry.”
But also no. Many people are making the same mistake they made at the beginning of SEO. They’re focusing on the numbers before they understand what the numbers mean. They’re taking old keyword research thinking and applying it directly to AI measurements. They’re looking for shortcuts. And whenever a new technology arrives, shortcuts are usually where people get into trouble.
“They are taking the idea of what traditional keyword research was and did and they are blanket applying it to the new measurements that we have and they are saying, ‘I was guessing before. Now I have a number. All I have to do is optimize to that number.’ And they have no idea what’s actually behind the number and what it means in the real world.”
That gap, between data availability and genuine understanding, is where a lot of the current AI visibility work is going wrong.
The Real Shift
The biggest changes in the industry right now are not simply AI visibility, or a need for GEO/AEO. The shifts required are to understand that we have moved from a place of success being driven by isolated page optimisation to judgement and reputation.
AI visibility isn’t a content problem, it’s a trust problem, a knowledge problem, and a systems problem. And those three things require a different kind of thinking than most SEO work has demanded up to now.
The brands that figure this out won’t necessarily be the ones publishing the most. The practitioners who do well won’t necessarily be the best prompt writers. They’ll be the ones who understand how AI forms its conclusions, and recognise that AI isn’t simply retrieving information about their brand, it’s forming an opinion about it.
Watch the full video interview with Duane Forrester here:
Thank you to Duane for sharing your IMHO.