TOFU: Does AI Even Know You Exist?

At the top of the funnel, the buyer doesn't know who you are and isn't looking for you. They are looking for a category. "What are my options for X in Y?" Whatever AI names in that answer becomes the starting universe for everything that follows. If you are not in it, you are not in the comparison, the shortlist, or the decision.

What AI is doing at this stage

AI is building a map. It works out what the category is, who operates in it, which of those players serve this buyer's region and audience, and which ones are mentioned often enough to be safe to name.

That last part matters. At TOFU, AI favours brands that show up consistently across many independent sources. One strong homepage is not enough. AI is looking for corroboration: the same brand, described the same way, in the same category, across directories, lists, reviews, and articles. Brands with thin or inconsistent footprints get left off, even when they are a perfect fit.

The prompts buyers actually type

These are the five TOFU templates SurfaceMap runs for every brand. None of them name your brand, so every mention you earn here is organic.

ID	Shape	Question	What it tests
T1	Superlative	What are the best (Category) options for (Target Audience) in (Region)?	Whether you make the "best of" list, the single most common real-world discovery query
T2	Directory (general)	In (Region), which companies operate in the (Category) space for (Target Audience)?	Whether AI recognizes you as a player in the category at all
T3	Directory (use-case narrowed)	Which companies provide (use case) for (Target Audience) in (Region)?	Whether AI associates you with the specific thing you sell, not just the broad category
T4	Problem-framed / educational	What should (Target Audience) look for when choosing a (Category) provider in (Region)?	Whether your content shapes the buying criteria, and whether you are cited as an example
T5	Reputation-based directory	Who are the most recommended (Category) providers for (Target Audience) in (Region)?	Whether third-party reputation signals are strong enough to get you named

Fan-out queries for this stage

When a model with search enabled gets a TOFU question, it rarely runs that exact question. It breaks it into several narrower searches, called fan-out queries, and builds its answer from what comes back. At TOFU, the high-likelihood fan-out patterns look like this:

• best (Category) (Region) (current year)
• top (Category) companies (Region)
• (Category) providers for (Target Audience)
• (use case) companies near (Region)
• (Category) directory (Region)
• (Category) reviews (Region)
• how to choose a (Category) provider
• (Category) buyer's guide
• (Category) reddit recommendations

Two things stand out. First, almost none of these contain your brand name, so your visibility depends entirely on whether the pages that rank for them mention you. Second, the year-stamped and "reviews" variants mean AI is actively looking for fresh, third-party lists, not your homepage.

SurfaceMap captures the actual fan-out queries each model generates for your prompts, so you can see which of these patterns are in play for your category and which sources answered them.

Where AI looks for answers

TOFU is the stage where AI has the most third-party material to draw from and relies least on your own site.

Typical sources at this stage:

• "Best of" listicles and roundups
• Industry and local business directories
• Review platforms and aggregators
• Trade publications and news coverage
• Forums and community threads
• Wikipedia and knowledge-base style entries
• Competitors' own "alternatives" and comparison pages

The questions are broad and the constraints are few (usually just category, region, and audience), so plenty of independent sources can answer them. Your own website still matters here, mainly for one job: stating clearly and consistently what category you are in, who you serve, and where. If your homepage calls you one thing, your directory listings another, and your LinkedIn a third, AI has a harder time placing you on the map.

This changes quickly as buyers move down the funnel. See MOFU for where third-party coverage starts to thin.

What failure looks like

These AI Mention States show up most often as a TOFU problem. Read the 12 AI Mention States for the full framework.

• Invisible Brand: no mentions anywhere, starting here.
• Category Ghost: AI doesn't associate you with the category you actually compete in.
• Traffic Donor: your content gets used or cited to explain the category, but your brand isn't named as an option.
• Customer Secret: customers know and like you, but that reputation never reaches the sources AI reads at discovery.

How to measure it

• Mention rate across T1 to T5, per model. ChatGPT, Claude, Gemini, and Perplexity often disagree at this stage because they draw on different indexes.
• Position in the list. Being named fifth of eight is different from being named first.
• Description accuracy. Is AI describing your category, audience, and region correctly?
• Cited domains. Which sources did the answer come from, and are you present on them?
• Trend over time. TOFU tends to move slowly, so a locked query set and a regular cadence matter more than any single run.

How to fix it

On your own site

• State your category, audience, and region in plain language on the homepage and key service pages. Use the words buyers use, not internal jargon.
• Publish content that answers the T4-style question ("what should (Target Audience) look for in a (Category) provider?"). This is how you shape the criteria AI applies later.
• Add Organization structured data that clearly identifies what you are, what you offer, and where you operate.

Off your site

• Get listed, accurately and consistently, in the directories and review platforms that show up in your cited domains.
• Earn placement in the "best of" lists AI is actually citing for your category. The fan-out and cited-domain data tells you which ones.
• Make sure your category description is identical everywhere. Consistency is a corroboration signal.

See where you stand at TOFU

• Run a free audit to see whether ChatGPT, Claude, Gemini, and Perplexity name you when buyers ask about your category.
• Take the 2-minute diagnostic to find your AI Mention State.

Next stage: MOFU: Are you in the comparison?

FAQ

What is TOFU in AI visibility? TOFU (top of funnel) is the Awareness stage of the AI buyer journey, where buyers ask AI broad questions about a category and AI decides which brands exist and are worth naming.

Why doesn't AI mention my brand when buyers ask about my category? Usually because AI can't find enough consistent, independent sources that place your brand in that category, region, and audience. At this stage, corroboration across third-party sources matters more than your own website.

What are fan-out queries? Fan-out queries are the narrower searches an AI model generates on its own when answering a question. The model builds its answer from what those searches return, so they reveal which sources it is actually reading.