As AI search evolves, tracking brand visibility across platforms like ChatGPT, Gemini, and Claude has become critical but complex. Existing tools offer limited accuracy and consistency, making it difficult to trust reported data. A key shift is from simple citation to meaningful recommendation—where brands are trusted as solutions to user problems rather than just appearing in search results. Tools such as NIMPT help by generating contextually relevant prompts and showing how a brand performs across different AI surfaces, revealing both visibility gaps and opportunities. The most effective strategy involves understanding conversational search behavior, identifying real knowledge gaps in AI training data, and creating authoritative content that supports trust and intent. This doesn’t mean producing large volumes of repetitive content, but instead focusing on high-value, user-intent questions. Brands must also consider geographic and contextual relevance, as users often omit location details in queries. Crucially, visibility isn’t about ranking but about being cited or recommended—where content adds genuine value to AI responses. Success depends on interpreting data accurately and building a knowledge base that AI systems naturally adopt, not just mirroring existing content. As AI models evolve and rules shift (e.g., Reddit’s sudden drop in citations), ongoing monitoring and adaptive content strategies are essential to staying visible and trusted in the emerging AI search landscape.
Which are the best tools to track, whether a brand is being sighted across all of the
different LLMs, like chatGPT, Claude, perplexity, Gemini, AI mode, co-pilot, etc.
It's going to be one of the questions that get asked, an awful lot over, you know, the
next year, probably before Christmas and we're sitting here in, or I'm recording this in
October, 2026. So where we are, there's a plethora of tools and as it stands, there are options
for every single budget out there at the moment. My question is, do many of these tools report
data correctly and it feels a bit like we are in the Wild West at the pavement. And by
that I mean, if you go back to the early days of SEO, lots of tools sprang up, the best known
of which now would be, for example, SEMRUSH or HREFs and there are tons and tons of different
tools. Whether clients use them as much as agencies is another question. So me personally
as an agency, as question.marketing, I use tools to report and I'm going to go on to show
you one of those and how you can get some understanding as to what your visibility is
in terms of, do you show four questions? I want to say questions, I want to talk about
questions that convert. So there is a huge difference developing at the moment between
whether you're cited and clicked and that leads to an inquiry or a sale or somebody is
asking a question to find out an answer and that then satisfies what they were trying
to do and that therefore doesn't then transpose or move forward into a click onto a site
and or an inquiry. We are as we know with AI search or AEO moving into a world where you
know, just getting cited, that's that's great. You're getting your brand recommended, talked
about, trusted and also, you know, referrals coming from that but a lot of that no click
marketing. So we're in a world where a lot of people are not then going on to click
onto a website, you're going to have to rely on an awful lot of recommendation and people
taking action themselves. So I believe as a lot of you will know if you listen to my
podcast, you'll know I think we move into voice search, we move into a world where we
all engage with our agentic tools, you know, be that muse charm in the future or just
talking more to Siri or, you know, Alexa was your first ever sort of interactive speaker
although it didn't really do what it could have done. I'm sure Jeff Bezos is sat back there
bashing somebody over the head saying, you know, maybe we could have done a better job
with this JAPS. He'll come back, I'm sure it will at some point. But if we look at the
tools that are out there at the moment, there's very, you know, light touches because if
you want to run a visibility report and you want to ask a question or create a prompt
that is relevant to your business, every time a prompt is run or ran, every time it runs
on an LLM, that consumes credits. Now if you're using a tool to do that and for example,
you want to track 30 different questions or prompts and you want to track that across
six or seven of the different surfaces, that's a lot of prompts, that's a lot of searches.
So if you then run that every day, you end up with consuming an awful lot of credits.
If you then want to compare where you sit, so if you want to take a prompt that you think
your business should come up for and you want to compare it to however many other competitors
that you might have, then you multiply that up exponentially again. So think of a scenario
where you've got 30 prompts, 10 competitors, seven surfaces and you multiply that up and
run it for 30 days in a month, it starts to get pretty expensive. Where that data comes
from is also important. So as we move into a world where we will, I'm certain, move
a lot more to localised search, you will want a tool that talks about is your, for example,
is your brand mentioned for your town. So one of my clients is a business called decoded,
they are based in Manchester and they are a recruitment specialist. Now I wouldn't
want to run their reporting from just an IP address based in Doncaster in the UK because
although I will be mentioning the brand and for example, Manchester in the prompt, you
would hope that that would give more visibility to that prompt and therefore get them cited
in results more but the chances are the people in their area, they know they're in Manchester
right. So if somebody is searching for where can I quickly find a contract engineer that
can work on Azure and my AI projects blah blah blah whatever that question or prompt is,
they are probably more likely not to then add in the word Manchester at the end of it.
So these are all the challenges that we're going to have to look at. You also want to
look at how fan out queries contribute to a search. So how is that search by any particular
LLM constructed and how is the result that's given to you produced. All of the different
surfaces are working on different models, pulling data from different sources and therefore
giving different results. So if you want to know clearly where you sit, you need to have
one of the more advanced tools. There are quite a few that are springing up that will give
you an idea. As I say, we're in a we're in a period where no one necessarily knows which
tools are working properly and which ones to trust because the data is so different.
If you're already paying for SEO, you might look at the semrush toolkit and it might cost
you something like $99 a month and it will give you an idea where you sit on chat GPT,
Plexity, Google AIO, Gemini and it will give you a broad brush and a lot of these will
give you a broad brush. HREF's brand radar, that will give you a lot more detail and there's
other tools out there. I'm going to go on to show you a tool called NIMPT and no disguising
the fact that I do like this tool. I'm going to show you the insides but what I thought
I would do is show you me setting up my own brand on NIMPT so that that then gives you
an idea of what you might expect and how you could use a tool like this. So if I want
to check my brand so I'm going to put in question dot marketing into the tool and I'm going
to check it in the UK so I'll then select a country. If you are just listening to the
podcast I'll do my best to talk you through this. Obviously you can watch it on my usual
channels like YouTube or on my blog or visit the website so I'm going to select the United
Kingdom. If I can spell it that is. It does help, right? So we'll put in UK and I'm going
to do English. So obviously if you want to test in other countries, love to hear from
people if they're running this in foreign languages. I'm currently not. I've got enough
on my plate trying to run it in English for UK clients and US clients at the moment.
So tools like this will then typically go out, research your site and give you a set
of prompts. In my example here it has come back with it knows I talk about answer engine
optimization a lot. So it has asked the question how does answer engine optimization work
for a small business. Sounds like a decent question that somebody might ask. It has also
produced is question.marketing a good AEO consultancy for UK service businesses. Somebody
might ask that question as part of their follow up research if they have previously discovered
me now I think we'll see.
we're gonna find that I rebranded question.marketing.
So it was previously my agency called Institution.
We were a full service agency.
I specialize now in AEO.
Before somebody says, well, why do you do that Simon?
I have been talking about Alun Sir engine since 2019.
I feel like I'm becoming a little bit
of an authority in the space.
I talk about it on my blog and my podcast, et cetera, et cetera.
So I am focusing on that as a service for clients now.
So it has then come up with things like,
who are the best AEO consultants in the UK
for an e-commerce brand?
Now, I pretty much know I won't come up for that
because the question.marketing website
has only been up for three months.
So it has replaced Institution,
but the challenge for me then is,
how do I rank, do I show up?
Can I get listed in any of the LLMs?
Now, this has then produced.
The tool has produced a ton of recommendations
all around different topics.
So one of the things we do are audits.
So one of the questions it's come up with is,
what does an AI visibility audit involve
for a bit small business?
Which UK agencies offer AI visibility audits
to show which competitors ChatGPT recommends instead of me?
Now, bear in mind, someone may say
that a lot of these questions or prompts are very detailed.
We have to start thinking about how people are gonna talk
and you're not trying to rank for keywords now.
You're trying to rank in conversational search
and it doesn't mean you have to go and build
hundreds and hundreds of pages
around slight variations of how people are gonna ask questions.
It's that's not the deal here at all.
Some SEO companies might think,
well, let's just go find out all the questions
that are on a subject from people also asking Google
and just build a whole ton of content around there
and answer all the questions
and therefore publish that.
That's not how it works people.
What we need to do is understand a fan out query
behind a prompt, how does an LLM come to the decision it makes
to give you the information it is going to display
or talk to you about.
So I'm gonna tell NIMT that I'm happy
with what it's come up with.
We could spend a lot more time on this subject
but I wanna get through a demonstration
and talk to where my brand sits.
I'm imagining and I haven't preempted this by running reports
in advance I am doing this live or recording it live
to see what comes through and I'll talk you through
what we are going to then show.
So it's put me up against some competitors.
It's found a business called HydroCreative Colface AI.
It is obviously saying that chat GPT perplexity,
applaud Gemini etc are effectively competitors to what I do
and they're not.
What it's doing is understanding that when somebody asks
a question that I'm trying to answer a lot of the time
the AI will be the answer.
So if somebody asks chat GPT how do I rank in the answers
within chat GPT, chat GPT will tell you its own answer
and it will therefore be the recommended answer itself.
So I'm going to deselect the actual LLMs as competitors
because they are not.
What I want to understand is what are my competitors
and I would have to teach it a bit more in this scenario.
Now if I go to my share of voice and this is still running
as we speak, what you can see is I've got
and for those that are listening branded search
as we'll be with probably the first time 99.9% of you guys
put your own brand into one of these tools.
If you stick in branded terms within there
you should obviously find yourself as the answer hopefully.
If you can't do that, that's probably the first misstep
that you've got because it would mean that the LLMs
are reading your site completely incorrectly.
So you should find you are able to be cited
for your own brand at least.
Now for those of you watching again,
I am now going to enable chat GPT, co-pilot, AI mode,
complexity and Claude within this tool and this is one of the
interesting things around NIMT is inside the tool,
you're able then to collect data across all of the surfaces
or all of the important surfaces as it stands.
Meaning you can get a much better idea of where you sit
to begin with.
Now remember once you sign up for a trial
and these guys do have a trial,
so please do go ahead and put your details in.
If you've had an outreach message,
perhaps from me, you may well get directed into this tool
to get a free audit of your current position.
So if you then go and find am I listed,
you'll be able to see the gaps and you'll be able to see
where your brand sits, where you're mentioned,
which prompts you are showing up for and which you aren't.
I'm not going to go into too much detail
in this session around how you then get listed,
but what we don't want to do is produce AI slop.
So the temptation for a lot of people is to go produce
a ton of content, try and mirror what other people have done.
Now the overriding principle here is don't use AI
to produce content, to fill in your own site
with information that already exists within LLMs.
My thesis and the way I think a lot of people in this space
are moving forward to look into how to rank
across the different LLMs.
And I say rank, I don't actually like the word rank
because there is no real source of the truth
in terms of how you get ranks across all of the LLMs
in one go at the moment.
I'm sure if you've got a huge brand
and you're, let's say, HubSpot
and you're trying to rank for CRM
and there's certain things within that that you've got,
you've got a very, very good chance
because so many third-party sites already trust your brand,
they already talk about you,
but if you look at the gaps of knowledge within queries,
that's the bits that you wanna fill in.
So the interest for me, and it should be for you guys as well,
is comparing what different surfaces say about your brand,
what they say about your industry,
are the answers correct across all of those surfaces
and is there an opportunity, more importantly,
is there an opportunity for you to provide some data
inside knowledge that will satisfy that LLM,
that you are trusted, you are also an authority
and you are therefore cited as an answer
or an authority in your space.
We're finding that, for example,
perplexity only referencing 13%
of the top-sighting Google in results.
Now, even to find that that's different, obviously,
Google AI overview and AI mode will then, therefore,
probably cite far more of its own listings
or its own knowledge base or data set.
We've found recently and follow me along
on some of these other case studies
that I'm running at the moment.
If, for example, you use Google Notebook LM
to produce video, then publish that on YouTube,
talk about it, write about it, share it,
and get people engaged with that content.
That is very quickly picked up by Google listed
and we're seeing the ability then to see content
come into the top of main Google search
within 48, 72 hours.
And I've got some case studies around this
that show this happening and demonstrated.
And this is now, well, it's not science,
but there's nothing to say that this method
or these methods won't change quickly.
Things like where we saw Reddit being cited
and used it across a lot of the surfaces,
particularly chat GPT, and then suddenly the rules changing.
So there's nothing to say that these rules are in stone.
There's no methodology that anybody can say
is 100% correct.
Now that we've sent out our queries,
the tool now will come back and give us some sort of level
of visibility.
Now, we can see now, if you're watching this on YouTube, you can see that question.marketing,
the question was, is question.marketing a good choice for an AI visibility audit?
Now in Google AI mode, apparently 60% of the time for my business is cited on AI overviews,
I'm top, I'm number one, apparently mine, this is one run, one run on one day, at one particular
time, from me running that, albeit in the cloud, on my computer. Now, there's nothing to
say, it may well be, it may well be just the fact that it knows me, right? I don't know.
You can see, in Claude, I am referenced, I'm found, chat GPT, I'm found, I'm top, but
that's again, it's a branded search, so I would hope, albeit that that website's only
been up for three, four months, maybe I should reference the exact date that the public
it was published, so maybe my AI will go and correct this piece of the video for me
or add a date in. Claude, can you please do that in the final publication of this video?
But again, more branded search, yes, I'm showing, and there are various results, there's
a ton of data here, right? I could pull all of this out. Again, branded search, I'm showing,
now, when we get to another one, who are the best AI implementation consultants for
South Yorkshire? I didn't check that prompt, it's put it in itself, it's probably found
it in my website, I am on chat GPT for that, interestingly enough, good to know. Which companies
in South Yorkshire can build an AI system? I am referenced in chat GPT, okay, interesting.
I run a small firm in Donkaster, which company can set up AI to take care of our, I don't
know what the additional part of that search is, but it says I am referenced in chat GPT.
Now what I want to do now is look for something that is a bit more general. So if we get down
to how does answer engine optimization work for a small business, that was one of the queries
that we looked at originally. Now I am not currently referenced, but the majority of those
answers are getting served by the LLMs themselves, so that's teaching me if I really want to try
and rank for that, I've probably got no chance, right? Only for the fact that if you want
to try and rank out rank chat GPT on chat GPT, does somebody want to tell me how you think
that might go? And if I achieved it, well, I think I might get Sam Altman on the phone
and say, come on Simon, what's going on here? But it's not going to happen. Here's a more
interesting one. I run an account and see firm and want to show up in AI answers, which UK
AEO consultancy should I use, right? I'm not currently listed, but we're starting now
to see some SEO/AEO companies getting listed. So there are articles on here from professional
trade bodies. There's one from the ACCA, which is the professional body of accountants. Now
that's in Claude. I can pretty much guarantee that that article is not talking about UK AEO
consultants for accountants. And I'd be very surprised if there is very much knowledge already written
into any of these LLMs designed to satisfy that exact query. So by that I mean, has somebody
written a piece of content that says they are an AEO consultant, they work specifically with
accountants and businesses, and they help them show up in AI search. Opportunity maybe, maybe I want
to write some content around that, probably not, because I'm not going down the route of trying to write
very long tail prompt research around content that is, well, I'm more focused on trying to help
small businesses, not necessarily going to try and write it just for an accountant.
Somebody's put, or one of the queries is, I run an e-commerce brand, which company can set up
ongoing tracking of our visibility in AI search. Now, again, we're seeing the actual LLMs listed,
probably extremely hard to be able to be found. Again, I run an e-commerce brand,
who can audit my visibility in AI search? Things like Shopify getting listed.
There is a business Sussex SEO. Be interesting to go and see if they
have written some content specifically to pull into that query. I'd imagine they probably have.
Scrolling down further. What does an AI visibility audit involve for a small business?
Now, what we've got to ask ourselves here is if I showed up for what does an AI visibility audit
involve for a small business, one, can I outrank any of the LLMs, probably not? But if I did show up,
what percentage of the people that see the answer if I was cited? Now, the difference being cited
or recommended is huge. So, I could well be cited. I could build some content around that
prompt, right? And I could be cited. The chance for people then clicking through to read
the piece of content I'd written to satisfy that query, probably already small,
landing them on a page and then telling them what they've probably just been told,
because I wrote the article and they land on something that looks similar. Follow me here.
Again, then the chance to get an inquiry from that very, very low, right? The prompt that's
below that is a little bit more interesting. Where can I book an AI visibility audit in the UK
to see if my business appears in an AI search? Now, in AI mode, very little showing,
there's a business called Kangaroo UK showing in AI overviews, Optimus consulting,
we're starting to now get into a recommendation. So, instead of a citation, could we get recommended?
I think this is probably the point of this video to show you the tools and show you the difference
between being cited as in your content is being used to pull in information into the LLM
to answer a question, or can you be recommended as a business to help somebody with a problem
that they may well have or are your products able to then be recommended as the solution to what
somebody's looking to buy? So, have a think about that. Quick, very, very quick tool through NIMPT
today. One of the tools that I like, it will show you rankings, it will show you where you sit,
it will show you where you rank in, you've got to be able to interpret the answers and I think
this is where I like to get involved with marketing teams and brands to help them understand
how to interpret this data, how to work with it and how to build the content in a way
to then become the answers and the recommendations, not just the citations. So, don't think of it in a
way of just getting picked up. You can do it through building your own knowledge bases and publishing
content correctly that adds to the knowledge corpus that is out there, right? So, if you want to learn
more, if you want to understand more about AO, GEO, being listed as the answer, all the recommendation,
I would love to speak to you. Hopefully that's been useful today. Please subscribe, watch, continue to
like and comment and I will see you in the next episode guys. Thank you.
Podcast Summary
Key Points:
Tracking brand visibility across LLMs like ChatGPT, Gemini, and Claude is challenging due to inconsistent, often unreliable reporting from current tools.
Many tools lack accuracy in data collection, especially when it comes to conversational or AI-driven search queries, creating uncertainty in results.
True visibility depends not just on being cited, but on being recommended as a trusted authority or solution in AI-generated responses.
Tools like NIMPT offer deeper insights by generating relevant prompts and analyzing how brands appear across multiple AI surfaces and languages.
Content strategy must shift from keyword stuffing to addressing real user intent in conversational search, focusing on knowledge gaps in AI training data.
Brand visibility in AI search is highly context-dependent—local relevance, prompt specificity, and content authority significantly influence rankings.
Some LLMs (like Google AI Mode) prioritize their own knowledge bases, while others (like Perplexity) rely more on external data, leading to uneven visibility.
Success lies in identifying content opportunities where brands can provide authoritative, unique information that AI systems naturally reference or recommend.
Summary:
As AI search evolves, tracking brand visibility across platforms like ChatGPT, Gemini, and Claude has become critical but complex. Existing tools offer limited accuracy and consistency, making it difficult to trust reported data. A key shift is from simple citation to meaningful recommendation—where brands are trusted as solutions to user problems rather than just appearing in search results.
Tools such as NIMPT help by generating contextually relevant prompts and showing how a brand performs across different AI surfaces, revealing both visibility gaps and opportunities. The most effective strategy involves understanding conversational search behavior, identifying real knowledge gaps in AI training data, and creating authoritative content that supports trust and intent. This doesn’t mean producing large volumes of repetitive content, but instead focusing on high-value, user-intent questions.
Brands must also consider geographic and contextual relevance, as users often omit location details in queries. Crucially, visibility isn’t about ranking but about being cited or recommended—where content adds genuine value to AI responses. Success depends on interpreting data accurately and building a knowledge base that AI systems naturally adopt, not just mirroring existing content.
, Reddit’s sudden drop in citations), ongoing monitoring and adaptive content strategies are essential to staying visible and trusted in the emerging AI search landscape.
FAQs
Tools like NIMPT, SEMRUSH, and HREFs offer visibility tracking across major AI platforms. NIMPT provides detailed insights by analyzing prompts and responses, while SEMRUSH gives a broad overview of brand mentions in AI search results.
Data accuracy varies across tools. While some platforms like NIMPT offer in-depth analysis, others provide broad, approximate results. There's currently no single 'official' source of truth, and results can change rapidly due to evolving AI models and data sources.
Ranking in AI search is not a traditional SEO process. Brands are typically cited in responses based on content that has been indexed or referenced by AI systems, not through keyword targeting. True visibility depends on authoritative, relevant content.
Being cited means your content is directly used to answer a question. Being recommended means your brand is suggested as a solution to a problem, often based on trust or expertise, which is a stronger signal for conversions.
Fan-out queries show how AI models generate answers through multiple related questions. Tracking them helps identify which topics or prompts are most relevant and where your brand could add unique, valuable insights.
Local search performance depends on whether prompts include location-specific terms. A brand in Manchester may not appear in searches from Doncaster, as users often omit location details when asking broad questions.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.