The AI Competitive Map
The AI does not use your battle cards. It has built its own picture of your market from everything it has ever read, and that picture is often quite different from yours. This chapter shows you how to find out what it looks like.
The short version
- The AI’s competitive map is not your competitive map. It is built from training data, not analyst reports or sales decks. The two usually look quite different.
- The gap between the two maps is the most important thing to understand first. Brands you have never heard of are often on the AI’s list. Brands you consider direct rivals are sometimes not.
- Build the map from what the AI actually says, not from what you assume. Let the data define the landscape.
- Unfamiliar brands outranking you is almost always a content and association problem, not a product problem. It is fixable once you know what is driving it.
Two chapters worth reading first
The competitive map measures how your brand appears in commercial AI responses. Before you start collecting data, it helps to understand two things: what kind of AI visibility you are actually measuring, and how to design the prompts that will give you reliable results.
Ch. 7 – AI Visibility Channels
Covers the difference between training data and grounded search. The competitive map measures commercial, grounded visibility. Understanding that distinction helps you interpret what you find.
Ch. 8 – Prompt Tracking
Covers how to design prompts that give you consistent, unbiased data. The quality of your competitive map depends almost entirely on the quality of the prompts you use to build it.
You do not need to read these first to get started. But if your results feel inconsistent or hard to interpret, these are the first place to look.
Why the AI’s map is different from yours
The AI did not go to your industry conference
Your competitive map was built from analyst reports, RFP shortlists, sales calls, and years of watching the same names come up. The AI’s map was built from training data: web pages, forums, review sites, articles, and anything else that was publicly available and crawlable at the time of training.
Those two sources produce very different results. A brand that was quiet during the AI’s training window will be underrepresented regardless of how strong it is today. A brand that published a lot of content, got cited frequently, and built strong topical associations will be well represented regardless of how it stacks up commercially.
The AI is not ranking competitors by quality, revenue, or market share. It is reflecting patterns in text. This is why the competitive map exercise is so often surprising. The brands that show up are the ones that were visible in the right places at the right time, not necessarily the ones that deserve to be there.
You might be a stronger product than every brand on the AI’s list. That does not matter until the model knows it. The competitive map is the starting point for understanding where you stand in the AI’s version of your market, which is increasingly the version that matters to buyers.
How to build your map
Start with prompts, not assumptions
The process is straightforward. You run category-level prompts across multiple AI models, log every brand that gets mentioned, and build a frequency table from the results. The key is to let the AI define the landscape rather than filtering it through what you already know.
Write category-level prompts
Write prompts the way a buyer who does not yet have a shortlist would write them. Think about the questions someone asks before they know which brand they want, not after. Do not mention your brand or any competitor by name.
Example prompts
“What are the best tools for [your category]?” / “Which platforms do companies use for [problem you solve]?” / “Compare the leading options for [use case].” / “What should I look for when choosing a [product type]?”
Run each prompt multiple times across at least two models
AI outputs are not fixed. Run each prompt five to ten times and record every brand mentioned across all runs. Do this across at least two models. ChatGPT and Gemini often produce different maps and both are worth knowing about.
Log everything, filter nothing
Record every brand that appears, including ones you do not recognise. The temptation is to skip past names you have never heard of. Resist it. Those are exactly the brands worth investigating.
Build a frequency table and classify each brand
Aggregate your results. How often does each brand appear across all runs and models? Then classify each one: known direct competitor, known adjacent player, or unknown. The unknown column is where the most interesting work begins.
Compare the AI’s map to yours
Lay the AI’s list next to your existing competitive map. What is on theirs that is not on yours? What is on yours that does not appear in theirs? Both gaps matter.
Reading the map
Three things to look for
Once you have your frequency table, you are looking for three things.
Brands you do not know
The most important finding. A brand you have never heard of appearing consistently in AI responses for your category is a signal worth taking seriously. It does not mean they are better than you. It means the AI has built stronger associations between them and your category topics than it has between you and those topics.
Brands you know but did not expect here
A brand you consider adjacent rather than competitive appearing on the AI’s map suggests the model sees the market differently than you do. Either the AI is wrong, which tells you something about how you have been positioned, or it is picking up on something real that you have not fully accounted for.
Your own position on the list
Where do you appear? How often? Are you consistently present or do you drop in and out depending on how the prompt is phrased? Inconsistent presence is a signal of weak topical associations. Consistent presence is a baseline to build from.
A brand that appears in 9 out of 10 runs is more embedded in the model’s understanding of your market than one that appears in 3 out of 10. Position within a single response matters less than consistency across many runs. Track both but weight frequency more heavily.
Why unfamiliar brands are outranking you
It is almost never about the product
If you find brands in the AI’s map that you have genuinely never heard of, the natural first reaction is confusion. How is a brand you do not recognise showing up more consistently than yours?
In almost every case the answer is the same: content, coverage, and association, not product quality. Here is what is usually driving it.
Training window presence
The brand was active and publishing during the period the model was trained on. Your brand may be stronger today but the model does not know that yet. The fix is not to catch up overnight. It is to start building coverage now so the next training cycle reflects your actual position.
Look at what they were publishing between 2022 and 2024.
Topical concentration
They have built strong associations with one or two topics that match the queries you are tracking. The model has learned to connect them to your buyers’ problems even if you would consider yourself a stronger fit. This shows up clearly in Topical Presence analysis.
Run a Topical Presence check on the brand. See Chapter 3.
Citation density
They appear frequently in sources the model treats as authoritative: review platforms, comparison sites, community forums, industry publications. Volume and source quality both matter here.
Run a citation audit. See Chapter 6.
Content structure
Their content is structured in a way that makes it easy for the model to extract clear, consistent associations. Clear headings, direct answers, consistent entity references. The model finds it easier to learn from well-structured content.
Look at how they structure their core product and category pages. See Chapter 5 (Entity Map).
Every one of these is fixable. None of them require a better product. They require a smarter approach to how you create and distribute content. The competitive map tells you who to study. The rest of the guide tells you how to close the gap.
What to do with the map
This is your starting point, not your answer
The competitive map is not the end of the analysis. It is the beginning. Once you know who the AI considers your competitors, you have a set of questions worth answering.
For each brand on the AI’s map that surprises you:
- What topics are they strongly associated with? See Chapter 3 (AI Topical Presence)
- Which sources are driving their citations? See Chapter 6 (Citation Data)
- How is their content structured compared to yours? See Chapter 5 (Entity Map)
For your own position on the map:
- Are you appearing consistently or sporadically?
- Which prompts surface you and which do not?
- What does the AI say about you when you do appear?
The map gives you a landscape. The other chapters in this guide give you the tools to understand it and change it.
Waikay builds your AI competitive map automatically by extracting every brand mentioned across your tracked prompts. Because it runs prompts continuously rather than as a one-off audit, you can see how the competitive landscape shifts over time and spot new entrants as they emerge rather than only when you think to look.
