Questions

• A check asks up to 3 of your tracked questions. Agent Search Visibility asks each one 3 times on each engine; the apps in AI Chat Visibility are asked each question once.
• Questions for your buying situations go first, one per situation, then questions about your category, highest-ranked first.
• A question that names your brand tests recognition, not recommendation. It is asked only when there is room, and it never counts toward the minimum.
• A check needs at least 3 questions about your category. Below that it won't start, and the Questions tab shows “Needs review”.
• For a local or regional business in the Netherlands, questions are in Dutch, and each check adds one English question on top so you can compare: the Questions tab labels it “English cross-check”. A plan keeps one English question; it never takes a Dutch question's place. Everyone else is asked in English.

Right after you add a brand, S6S reads its website in the background and writes the questions from what it finds. Until that's done (usually about a minute) the Questions tab shows “Preparing your questions…” and Run Check waits, so the questions you review are the ones your first check asks. Nothing is charged while you wait.

At the top of the Questions tab, S6S asks “What do people come to you for?” and lists the situations it found on your site. Say yes or no to each, or add your own, up to 8. Only the ones you say yes to are used.

When you next generate questions, you get a question for each confirmed situation, and checks ask those first. The Report then shows how often you were named for each situation and for your category in general.

• Generate new questions: S6S looks at what people search for in your market and drafts questions. This is free and takes about half a minute.
• Review: untick any question you don't want to track. Drafts that wouldn't measure your visibility reliably are listed under “Left out”, each with the reason. Questions S6S isn't confident about can't be selected.
• Save: nothing changes until you save. Questions you picked from question research stay as they are.
• Saving starts a new baseline: checks after it are not compared with earlier ones. Your next Run Check asks you to confirm this.
• Don't ask this: turn off one question from the list. Checks stop asking it, and generating new questions never brings it back (“Ask this again” does). It starts a new baseline too, and it isn't possible when a check would be left with fewer than the minimum.

You edit the business name, category, keywords and competitors in brand settings. If your brand details change (name, website, category, keywords, competitors, markets or buying situations), the tab shows “Brand details changed”. Your current questions still run until you save new ones.

Each question also says whether people actually search for it, for example “People search for this: … · about N searches a month”, or “We haven't seen people search for this yet.” Search data is available for a set of supported countries; elsewhere questions are planned from your brand details only, and the preview says so.

In the Fix Plan tab, question research collects questions from Google “People also ask”, Reddit and search suggestions. A question with no real page behind it is labelled “Suggested (not observed)”: a hypothesis about what buyers ask, not evidence of demand.

Who AI named

Under each question you see which companies the Agent Search Visibility answers named in your last check, and in how many answers you were named. If no answer named any company, the question may read as a request for advice rather than for recommendations.

Why a check skips a question

Skipped questions sit under “Not used” with the reason in plain words, for example:

• Names your brand, so it can't show whether AI recommends you on its own.
• Asks about the whole world, but your brand targets specific markets.
• Too similar to another question.
• Ranks below the questions a check looks at. It moves up if higher-ranked ones are removed.
• Not phrased the way a buyer would ask.

A dot on the Questions tab tells you when something needs a look: red for “Needs review” (checks can't run), amber for changed brand details or for questions a check skipped or that named nobody.

• A brand can have up to 10 target markets, written as “City, Country” (for example “Almere, Netherlands”) or as a country or region.
• A local brand needs at least one target market. It is also what lets the Local & Maps check run.
• Each question is asked in one market: its own, or your first market. Adding markets doesn't multiply the questions or the price.
• A question names your market only when it is a real place. “Global”, “Worldwide” or “Online” are a scope, not a place, so questions never read “… in Global?”.
• In Agent Search Visibility, ChatGPT also receives the market as a location setting; Perplexity gets it in the question text. App answers are collected for the market's country, in supported countries; a brand with no market is checked from the United States.