National AI visibility score: what is it and how do you improve it?
The National AI visibility score is an index between 0 and 100 that measures how visible your brand is in AI answers within your sector. Search engines are turning into assistants, and in those answers only a small handful of names gets mentioned. If you're not among them, you're technically findable but in practice you don't get chosen.
Why a National AI visibility score is needed
Classic SEO metrics tell you how you score in the search results: positions, impressions and clicks. What those figures miss is the AI layer that now sits on top of them. Panels with AI answers are increasingly taking over the top of the page. As a result, we come across companies that see their positions holding up while enquiries decline, because the answer at the top doesn't mention their name. We don't put a figure on how often that happens: there is no public measurement of it that you can open yourself, and an estimate of our own would here create the impression of a figure we don't have.
The distribution in those answers is also more skewed than in the classic search results:
The top of a sector
gets repeated more often in AI answers than in the classic results, where ten blue links sit side by side. We don't put a percentage on that: we haven't found a measurement you can open yourself, and our own AI Index NL measures nine sectors and not the whole market.Outside that leading group
a brand simply doesn't get mentioned in many answers.Most brands
don't yet measure this in a structured way.The score is therefore not about whether you're findable, but about whether AI trusts you as an answer.
What exactly is the National AI visibility score?
The National AI visibility score is an index between 0 and 100 that shows how visible your brand is in AI answers within your sector. It rests on three pillars.
Mentions
how often your brand gets mentioned.Context
whether you appear as a recommendation, as an example or as a passing mention.Position
whether you're among the first names in the answer or only listed as an additional option.Here's how to read the outcome:
- 0 to 20: invisible to AI.
- 20 to 40: occasionally mentioned.
- 40 to 60: recognisable in your niche.
- 60 to 80: structurally present.
- 80 to 100: category leader in AI.
In practice, the National AI Index shows, for example, how often a brand like Amazon appears in AI answers, how the score per brand develops over time, and which ten brands in a sector are mentioned most often.
Three scenarios for your brand in AI answers
Scenario 1: you're not mentioned anywhere
Your score often sits below 20. The solution starts with a basic story that AI can understand and repeat: clear positioning, consistent content and a profile that's easy to cite.
Scenario 2: you're occasionally mentioned
Your score sits between 30 and 50. The focus now is on building recognisability and authority, so you move from occasionally to structurally mentioned.
Scenario 3: you're the category leader
Your score sits above 70. The strategy is about protecting your position, experimenting with new questions and making sure your information always stays better than everyone else's.
Step-by-step plan to improve your AI visibility score
Step 1: determine your core questions
Don't start with keywords, start with the questions customers would ask an AI assistant. That's the starting point of every measurement.
Step 2: check which brands are named now
Test your core questions in different AI assistants and note which brand names keep coming back. That way you see straight away who currently owns the moment of decision.
Step 3: map your current score
Use our National AI Index as a baseline measurement. If you want us to set this up structurally for you, that falls under AI and automation. If you first want to know whether you appear in those answers, that starts with having your AI visibility measured.
Step 4: make your brand easier to cite
AI would rather give advice based on brands with a clear profile. Make sure your brand is recognisable in a couple of sentences: what you do, for whom and why you.
Step 5: optimise content for AI answers
Not everything has to be long, but it does have to be extremely clear. Build content blocks that answer your core questions directly. While you're at it, also take a critical look at the structure of your product pages, because AI cites pages that answer quickly and concretely.
Step 6: make your site AI-ready with llms.txt
With an llms.txt file you set out a curated table of contents of your brand information, category pages and advice guides. It doesn't fix anything and doesn't force anything: no major AI provider has committed to fetching it from external sites, and Google says Search ignores it. This belongs to a healthy technical SEO NL foundation, just like schema markup in JSON-LD with your organisation and language.
Step 7: measure and update every quarter
AI changes faster than traditional SEO reporting. Repeat the measurement every quarter and adjust based on what you see.
Why AI visibility now matters just as much as positions
More and more choices are made in the AI answer itself, before anyone clicks through. If you're missing from that moment of decision, you're still technically findable but practically invisible. The National AI visibility score fills exactly that gap.
These are questions AI already answers for your customers today:
- Which provider is reliable for this product or service?
- Which webshops are safe and deliver quickly?
- Which provider suits a specific type of customer?
- Which brands score well on price, quality or sustainability?
Prompts to test your own AI visibility
You don't have to wait for a full measurement. With these five prompts you can test today, yourself, in ChatGPT, Gemini or Perplexity, where you stand:
General comparison
You are an assistant helping Dutch consumers. Name a maximum of 5 reliable providers for: [product or service].Specialist question
I'm a customer looking for: [specific situation]. Which providers best suit this situation and why?Checking your own brand
You know the Dutch market for [your category]. In which situations would you recommend [your brand]?Competitor perspective
Which providers are most often mentioned as an alternative to [your main competitor]?Areas for improvement
You are a critical advisor. Look at the online presence of [your brand] and describe 3 areas for improvement.For each prompt, note which brands are mentioned and in what order. That gives you your first rough baseline measurement.
Why the same question gives a different answer twice
Before you do anything with a score, you need to know why it wobbles. Anyone who asks the prompts above twice in a row often gets two different lists of names back. That's not a glitch, and it doesn't mean the measurement is nonsense either; it's part of the instrument. It means you should never read a single outcome as your standing.
Four reasons the answer shifts
The model doesn't always choose the same thing
Language models generate word by word with a certain degree of variation. The same question asked twice therefore produces a slightly different wording each time, and sometimes a slightly different order of names.There's a live search happening
Assistants that consult the web fetch pages at the moment you ask. If something changes in what's online that day, the answer changes with it.Your own history counts
If you're logged in, in a conversation where you've already talked about your own brand before, or with a memory feature switched on, you're not measuring the market but your own trail. This is the most common mistake and also the easiest to avoid.The wording does more than you think
Reliable provider, best provider and cheapest provider are three different questions that produce three different lists. Anyone who tweaks their wording between two measurements is measuring their own choice of words.How to still make something useful out of it
By repeating rather than refining. Ask each question five times in a new, logged-out conversation and only note whether you're mentioned, not how nicely. Being mentioned five times is different from being mentioned once, and that difference is stable enough to steer by. Then fix the questions in a file and don't change anything about them again, not even if you think of a better wording. A yardstick you keep adjusting along the way measures nothing.
Look further at the list as a whole rather than at your own place in it. If, in three months, two names have disappeared that always used to be there, that's more valuable information than whether you moved from spot four to spot three.
What you can actually steer here
You can't change anything about the model itself. You can't flip a setting, buy a spot or submit a request. What you can change is the material such a system draws on, and most of that material isn't yours. That insight determines where you put your time.
What you have in your own hands
Your own words, in plain sentences
Put two or three sentences somewhere on your site saying who you are, what you do, for whom and where. No jargon, no slogan. A system can lift that exact sentence word for word. A brand whose sentence appears nowhere has to be summarised from scattered fragments, and what comes out is something you never wrote yourself.Answering the question itself
A page that asks the question and then answers it is more useful than a page that writes around the answer. That's the same shift as the structure of your product pages: the answer first, the reasoning after.Being findable and readable
Anything behind a login, anything that only exists in a PDF, and anything that only appears after scripts have run, doesn't count. This is the part that overlaps with ordinary technical SEO NL.What others have in their hands
- What gets written about you. Mentions on places that aren't yours carry more weight than your own page, for the same reason a recommendation from someone else carries more weight than a recommendation from yourself.
- Your reviews, and especially the text in them. Not the average score but the words. Reviews in which customers describe their problem supply the words that belong to you. A wall of fine, would recommend delivers none.
- The lists you're on or not on. Comparison pages, sector overviews and registers are exactly the material a list of five names gets built from. If you're missing from all of them, that's often the whole explanation.
The conclusion is uncomfortable but useful: most of your AI visibility gets decided outside your own site. Your own pages make sure it's correct once you're mentioned. Whether you're mentioned depends on what exists about you elsewhere.
Four ways to make this measurement worthless
A measurement that's set up wrong is worse than no measurement, because you'll still act on it. These are the four pitfalls we come across most often.
Only asking about your own name
Ask what do you think of this brand and you'll always get an answer, because you supplied the name yourself. That measures nothing. The question that counts is the one without your name in it: who would you recommend for this problem. Only when your name comes up then are you visible.Polishing the prompts along the way
Sharpening the wording a little each round feels like improving, but in reality it throws away your comparability. Fix the questions and accept that they aren't perfect.Taking one assistant and calling it the market
The assistants draw on different sources and behave differently. Test at least two and keep the outcomes separate, because an average across two dissimilar systems is a number without meaning.Turning it into a target figure
As soon as someone is held accountable for a score, people start steering towards the score rather than the visibility. That goes wrong especially fast here, because with a handful of prompts you can nudge things towards your own answer without anything in the market having actually changed. Use the figure as a thermometer, not a thermostat.What remains if you avoid those four is modest and useful: you know whether you're mentioned for the questions your customers ask, you know who else is on the list, and over the quarters you see whether that moves. This subject promises no more than that, and anything that promises more is a sales pitch.
What we do and don't claim about this
A lot gets said about this subject with a lot of confidence, so here is what we do and don't substantiate.
What we don't say.
We don't name a percentage for how often the leading group recurs in AI answers, because we haven't found a measurement for that which you can open yourself. We don't promise a position in an AI answer, because no provider can force that; anyone who does promise it is selling something they don't own. And we don't call this a quality mark, because nothing is certified: it's an index we compile ourselves.
What we do say.
Our own AI Index NL measures nine sectors and not the whole market, and how that measurement works is stated openly alongside it. That's as far as it reaches, and no further. Within that scope, the outcome is repeatable and you can check it yourself.
Why this distinction matters.
A score in this field is an estimate of something nobody can see in full, and that is exactly why it's so tempting to attach an impressive number to it. Anyone who does that is selling a certainty that doesn't exist. We would rather state the scope alongside it and let you judge for yourself whether the figure carries enough weight for your decision.
Frequently asked questions
Is AI visibility the same as SEO ranking?
No. It deliberately looks beyond classic positions. You can rank highly and still barely appear in AI answers.
Does it make sense if my SEO isn't perfect yet?
Yes, but it's wise to have the basics in order. A brand that AI can cite well is usually also solid technically and in terms of content.
How often should I measure?
For most companies, measuring once a quarter is a good rhythm. More often is fine, especially around major AI updates.
Can I set this up myself?
Yes, the test prompts above will get you a long way. Our full National AI Index adds extra layers, such as trends per brand and comparison within your sector.
What's the first concrete step?
Start with your core questions and test them with the prompts. Then request a sector analysis.
How to use this article in your content strategy
This article can serve as a central pillar page around AI visibility. Use it as a hub that other articles and landing pages link back to. Think, for example, of a page about AI traffic loss, an explanation of llms.txt, and a case page showing how your score has improved.
From this page you can link through to deeper material, such as a separate article on AI traffic loss, a comparison of heatmap tools, or a page about your own AI index. That builds a network of pages that's logically connected, both for visitors and for AI assistants.
Is the National AI visibility score the same as a regular SEO ranking
No. The score deliberately looks beyond classic positions alone. You can have a good SEO ranking and still have a low AI visibility score. The score measures how often you're mentioned in AI answers and how prominent that mention is. That's a different layer than organic positions alone.
How often should I measure my National AI visibility score
For most companies, measuring once a quarter is a good rhythm. The AI layer changes faster than classic SEO reporting, but we see patterns take clear shape over a period of a few months. For major campaigns or a big repositioning, an extra measurement in between can be worthwhile.
Can I set up a National AI visibility score myself
You can already gain a lot of insight yourself by systematically testing questions in AI assistants and tracking which brands get mentioned. TheSEO's full National AI Index adds extra layers on top of that, such as sector comparisons, trend data and AI traffic loss estimates. That combination makes it easier to make decisions.
In short: the National AI visibility score
The score is especially relevant for:
- Brands that depend on organic traffic.
- Webshops, service providers and SaaS companies.
- Teams with a mature SEO approach who are still seeing enquiries or revenue decline.
What the score measures is the step from being findable to being chosen
It combines how often you're mentioned, how prominently that happens and how well that matches the intent behind the question.
Three steps you can take today:
- Map out your core questions.
- Test which brands are mentioned right now.
- Make your brand story easier for AI to cite.
Learn more about how AI is changing search
In the knowledge base you'll find more articles about AI visibility and SEO.
Start with five questions and an empty list
The smallest usable step is already in this article: write down five questions the way your customer would ask them, put them to two assistants while logged out, and note which names come up. That takes half an hour and gives you something you don't have today: a baseline measurement you can compare something against in three months. Everything after that, from your citable brand sentence to your mentions elsewhere, only becomes meaningful once you know where you started.
If you can't work out which five questions belong to your market, or you'd like someone to join you for the first round and tell you honestly whether this subject is urgent for your business or not yet: put that question to us. Tell us your sector, your main service and the two competitors you always run into, and we'll go through it together. That can be done through the contact page, where you pick a moment in the calendar yourself. If you'd rather have the measurement run structurally instead of doing it yourself every quarter, that falls under AI visibility and the rates are on the pricing page NL.
You're still in the top ten
You still see your brand in the normal results. Positions wobble a little, but nothing dramatic shows up in your tools. Yet traffic and revenue are declining.
AI gives ready-made advice
The AI layer immediately gives an overview with a handful of brands. That's where the main attention goes. If you're not among them, you're invisible in the new layer of search.
From feeling to measurable visibility
Instead of guessing whether you're mentioned, the National AI visibility score gives a number between 0 and 100 that shows how often AI chooses you relative to your competitors.
Classic SEO mainly looks at positions, impressions and clicks. That's valuable, but it misses the most important layer that has been added in recent years. AI answers in search engines and external assistants such as ChatGPT, Perplexity and other tools that are taking over search queries.
In our research we see that for many search queries, the click is now only one of the possible endpoints. The rest gets handled in AI overviews, summaries or complete recommendation lists. And the same set of brands keeps getting mentioned in them. So the question isn't only whether you're findable, but whether AI trusts you as an answer.
Three out of five companies see enquiries and revenue decline while their positions change only slightly. As soon as they check their main keywords in AI answers, they see the same competitors coming back every time.
The National AI visibility score was developed to make that part visible. Not just visibility in Google, but also in AI assistants that can name your brand or skip over it.
In conversations we keep hearing the same sentence back. My traffic is dropping, while my keywords still look fine on paper. The fault isn't in your gut feeling, it's in the measuring point. You're measuring the old layer, while a new layer has since been placed on top of it.
The National AI visibility score is designed as an extra layer on top of your existing SEO reporting. You keep measuring positions and clicks, but you add an AI index that shows how often you're chosen as the answer.
The National AI visibility score is an index between 0 and 100 that shows how visible your brand is in AI answers within your sector. The score combines three components. How often you're mentioned, how prominent that mention is and how well you match the intent behind the question.
A score of 20 doesn't mean you're doing nothing right. It means that right now you're only getting a small share of the AI attention in your category. A score above 70 means you're structurally present in most of the relevant answers.
So the score isn't an ego number. It's a way of seeing whether AI knows you, understands you and is willing to name you when your ideal customer asks for advice.
The three pillars of the score
- Mentions how many times your brand is named in AI answers to questions that are relevant to you.
- Context whether you're mentioned as an example, as a recommended party, or merely as a passing name without a clear reason.
- Position whether you consistently appear among the first names or panels, or only get mentioned as an additional option.
Taking those three layers together produces a score that goes beyond simply counting links. It's about trust and recognition in the eyes of AI assistants.
In the first National AI Index we see that only a very small number of brands in the retail and electronics category score above 70. Most companies stay stuck below 30, even when they have a strong classic SEO strategy.
Does it make sense to work on my AI visibility if my SEO isn't perfect yet
Yes, but it's wise to have the basics in order. Strong content, a technically healthy site and a clear brand story matter for both. See the National AI visibility score as a next step once you no longer just want to compete on classic positions, but also want to be chosen in AI panels and assistants.