Explanation of dimensions and methodology

🧢 Coach T’s Recap

A brand dimension is an attribute, quality or theme consumers care about when evaluating your category, like taste, value for money or trust. Some dimensions may be similar to category entry points. For each dimension, we write a set of unbranded questions and ask them on each AI platform every week.

Your AI visibility on a dimension shows how readily AI answers bring your brand up when a buyer cares about that attribute. It measures association in AI answers, not how often people ask, and it counts mentions rather than whether they’re positive.

What is a brand dimension?

Brand dimensions are the attributes, qualities and themes consumers care about when evaluating your category. Some may resemble category entry points, but not all dimensions are category entry points. We build them for each category from Tracksuit’s own data.

In soft drinks, they might be taste, low sugar, value for money and natural ingredients. In everyday banking, they might be trust with money, fees, rewards and help managing spending.

Your results are shown by dimension, so you can see which attributes AI answers link with your brand and which they link with your competitors. Different brands may show up more strongly on different dimensions.

How we build the questions

We start with your brand dimensions. Each one reflects an attribute, quality or theme consumers care about in your category.
We write questions for each dimension. Each dimension gets its own set of questions, framed around buyer needs so that we’d typically expect a brand mention to be positive. “I’m cutting back on sugar but still want something that tastes great. Which soft drinks should I try?”
We keep questions unbranded. The questions don’t name a brand, so a mention means AI brought it up without being prompted to name it. We count mentions whether they’re positive or negative.
We test the questions. A question only counts if AI answers to it name brands from your competitor set.
We ask every question on each AI platform, every week. We record which of your tracked brands each answer mentions.
We calculate your metrics. Your AI visibility on a dimension is the share of answers to its questions that mention you. Your relative share on the dimension is your share of mentions among the brands in your comparison.

Questions are generated through the same process for every category. Note: the example question above is illustrative, not from a real category.

Why we do it this way

Here’s the thinking behind the method:

Brand health, not search rankings: we don’t pick questions by how often people search for them. The questions are designed to probe how strongly AI links each brand to an attribute, the way a survey probes what people think.
Unbranded questions measure real salience: asking “What do people think of [brand]?” would always mention that brand. Unbranded questions show whether AI brings you up when you’re not named.
Breadth across questions: AI answers vary, so we look across many answers. We prioritise asking a broad set of questions across topics and dimensions rather than asking a narrower set more frequently, such as daily. This gives you a wider view of brand salience across your category.
The same questions each week: we ask the same set every week, so a change in your results reflects a change in AI answers, not in what we asked.

What this means when you read your results

Association, not demand: how many questions a dimension has, or how often they’re asked, says nothing about how much that dimension matters to buyers.
Mentions, not verdicts: a mention counts whether the answer recommends your brand or warns against it.
Modest bases: one dimension’s questions on one platform in one week is a small base, so look for patterns across weeks and platforms.
Uneven is normal: brands are known for some things more than others. A health-led soft drink often mentioned on low sugar but rarely on taste may simply reflect its positioning. It’s only a gap if the buyers it wants choose on taste.
Size matters: bigger, better-known brands tend to be mentioned more on most dimensions. How you do for your size says more than where you rank.
Dimensions don’t add up: a question can belong to more than one dimension, so dimension results never add up to your overall number.

Reading dimensions next to your survey

Many brand dimensions have a counterpart in your survey, such as a brand statement or an imagery theme. When the two agree, that’s a consistent picture of your brand. When they don’t, it’s a question worth asking:

“People link us with quality, but AI answers rarely mention us on it.” One possibility is that your story lives in people’s experience but isn’t yet reflected in what others write and say about you.
“AI answers link us with low prices, but people don’t.” It may reflect how reviewers and communities talk about you, or an older framing that has stuck.

LLM Brand Health is collected weekly, while survey results are monthly and usually shown as a 3-month rolling average. When you compare them, pick survey months that cover the weeks you’re reading.

💡 Tip

Use the Dimension filter on the Visibility page to see your AI visibility on one dimension.
Then open your Statements page and find the statement that matches it, to read what people think beside what AI answers say.

A few things to keep in mind

The dimensions are set by Tracksuit: you can’t add or change dimensions or questions yourself, but you can filter out dimensions that aren’t relevant to your brand or your analysis. If there’s a dimension you’d like us to track, tell your Brand Champion.
One category at a time: dimensions and questions are built per category, so don’t compare a dimension across categories or markets.

Questions? For questions, ideas or feature requests, reach out to your Brand Champion at brandchampion@gotracksuit.com.

FAQs

Can I see the questions you ask?

Yes! Open the LLM Brand Health tab and go to the Questions page, or click View all tracked questions on the Visibility page.

Can I add my own questions or dimensions?

Not currently. We build every category’s questions the same way, so results stay consistent. You can filter out dimensions that aren’t relevant to your analysis. If something important is missing, let your Brand Champion know.

Why don’t your questions match what people actually type?

They aren’t meant to. We design questions to understand how strongly AI links each brand with the dimensions of the category, rather than to reproduce what people actually ask.

Are dimensions the same as category entry points?

Not always. Some dimensions may resemble category entry points, while others describe attributes, qualities or broader themes consumers care about when evaluating a category.