# Where buyers meet the answer

> How buyers arrive at each AI surface, why that decides how tracked prompts are phrased, where a shortlist forms, and what the dark funnel hides.

## Reading brief

- Decision: Which AI surfaces a brand's buyers actually meet, and how their questions should be tracked on each.
- Evidence: How each surface is reached, from Google's own descriptions, and what that implies for prompt phrasing.
- Action: Track search-box phrasing for AI Overviews and full questions for chosen surfaces, and weight the questions where shortlists form.

Buyers meet AI answers in a few different places, and they arrive at each one differently. How they arrive decides what they typed, and what they typed decides how a brand should phrase the questions it tracks.

Most writing on this subject restates the buyer journey with AI added, often with adoption figures that disagree from one survey to the next. This chapter skips both and works from how each surface is actually reached.

## How does a buyer arrive at each surface?

| Surface | How the buyer gets there | What the buyer typed |
|---|---|---|
| AI Overviews | Shown above some Google results without being asked for | An ordinary search-box query |
| AI Mode | Chosen, as a separate tab in Google Search | A longer question, then follow-ups |
| AI assistants such as ChatGPT, Claude, Perplexity, Gemini, and Copilot | Opened for the purpose | Full sentences, often with the constraints that matter |
| AI agents | Handed a task to complete | A goal rather than a question |

AI Overviews appear on a Google search the buyer was going to run anyway. Google states that they are [shown only when its systems judge them additive](https://developers.google.com/search/docs/appearance/ai-features) and that they often do not trigger. Each comes with supporting links to the pages behind it.

AI Mode is a choice. Google describes it as a [separate tab built for follow-up questions](https://blog.google/products/search/google-search-ai-mode-update/), and the same page notes that AI Mode and AI Overviews can show different responses and links. The assistants work the same way: the buyer opens them deliberately and asks in sentences, a style called conversational search.

Agents are the newest arrival and the least visible. Chapter 7 covers them.

## Why does the surface change how prompts are written?

A prompt set written only in full sentences misses AI Overviews, where buyers type short queries. One written only in keywords misses the assistants, where buyers describe their situation in detail.

A useful prompt set carries both forms. Search-box phrasing comes from real search demand, the grounding [chapter 3](/guides/aeo/measuring-ai-visibility) describes. Full-sentence questions come from how buyers describe their problem on sales calls, including the constraints that decide a purchase.

One buyer need, written for each surface, might read:

- **Search box, for AI Overviews:** expense management software for small business
- **Assistant or AI Mode:** A 40-person company is moving its expenses off spreadsheets. Which tools handle card reconciliation and connect to an existing accounting system?
- **Follow-up:** Which of those costs least for under 50 users?

The second and third forms carry the details that decide the purchase, and they are where answers start mentioning specific vendors.

Both forms start from the same place: the category entry point, the situation or need that makes a buyer start looking. A question written from that situation reads the way a buyer would ask it, on any surface.

## Which surfaces should be tracked first?

The surfaces worth tracking first are the ones a brand's buyers already use. Sales calls are the quickest evidence: buyers often say where they did their research.

AI Overviews need one extra reading. Because Google shows them only on some searches, whether an overview appears at all for each tracked query is worth recording. A category where overviews rarely appear puts more weight on the surfaces buyers choose.

## Where does the shortlist form?

A shortlist is the small set of vendors a buyer decides to evaluate seriously. In an AI answer, it can form inside a single reply: a category question returns a few brands, and those become the shortlist.

That makes category questions the part of the buyer journey where AI visibility matters most. A brand missing from those answers is missing from the shortlist before any website visit or sales conversation.

For measurement, it means category questions deserve the closest reading. Mention rate on those questions shows whether the brand makes the list, and average position shows where it sits on it.

## What does the dark funnel hide?

The dark funnel is the part of a buying journey that analytics cannot see. AI answers add to it: a buyer can read an answer, form a view, and arrive later through a branded search, with nothing linking the two visits.

Dark social widens it further. Buyers paste AI answers into messages and email threads, where any link they share arrives with no record of where it came from.

The answer itself is the one part of this that can be measured. Tracked answers show what buyers are told, while what they do afterwards stays mostly out of view.

Chapter 3 explains why the two layers are reported side by side and never used to prove each other. [The dark funnel in AI search, measured](/blog/dark-funnel-ai-search) looks at how the engines themselves talk about the idea.

## What does this chapter not cover?

Knowing where buyers meet the answer says which surfaces and questions to watch. Measuring a brand's place in those answers is the subject of the next chapter.