How Foreground works

What is AI search visibility?

More people now ask ChatGPT, Gemini, and similar tools questions like "what's the best accountant near me" or "who does custom furniture in Rotterdam" instead of typing them into Google. Those AI answers don't return a list of ten blue links you can scan for your own name — they return a short, confident paragraph that either names your business, names a competitor instead, or names neither.

AI search visibility is whether your business shows up in those AI-generated answers, for the customers you actually want. Think of every answer as having a foreground with room for only a few names — Foreground finds out whether yours is one of them, niche by niche, instead of guessing from one generic "about your category" question.

Pick your customer niches

A scan doesn't start from your category in general — it starts from the specific customers you want. Give Foreground your site, and it reads it and suggests 3 customer niches: who the customer is, what they're looking for, and where — each with an honest note on whether your own site currently supports that niche at all. You confirm 1 niche for the free check. Once a niche is saved, its questions are fixed — to change them, archive the niche and create a new one.

How a scan works

For each active niche, Foreground sends 3 questions — phrased the way that specific customer would actually ask, not just "tell me about {your brand}" — to 4 AI models (ChatGPT, Gemini, Claude, and Grok), for 12 answers per niche. If the niche is local, Foreground also checks 3 answers from Google Maps, shown separately as "on Maps: X of 3."

Every answer is grounded in a live web search rather than only the model's static training data — each model is queried directly against its own provider's API, using that provider's own web-search tool. See "Limitations, honestly stated" below for what this does and doesn't cover.

Separately, Foreground also asks 3 questions that do name your business directly (a comparison with competitors, "is this a good choice?" and "what do people say about it?"). We use those answers internally and don't show them in the app yet. They don't count toward "named in X of 12," since an AI repeating a name it was just given doesn't show it would bring you up on its own.

How "named in X of 12" is counted

For each of a niche's 12 answers, Foreground checks whether the answer's text names your business — by name, a known alias, or your domain. For the Maps check, it's whether your business's listing appears among the places returned. The count is exactly that: how many of 12 (or, for Maps, how many of 3) name you. There's no decaying credit for position and no single 0-100 score — just the count, and where you rank against everyone else who got named.

If an AI gives no answer in a scan (an outage or a usage limit on its side), its questions are left out of the count rather than counted as "not named": the count then reads "named in X of 9", and the report says which AI didn't answer. When two scans had different AIs answering, changes are compared only on the AIs that answered both times.

Foreground also shows whether you're in the shortlist for a niche: named in at least a quarter of the answers that came back (3 of 12, or 3 of 9 when one AI didn't answer), and never on a single answer. Not clearing that bar isn't a failure — it means you're not in the shortlist yet, and your report shows exactly how far off you are and who's ahead of you.

AI answers move between runs, even with nothing changed on your end. Between two single scans, a move of fewer than 3 (out of 12) isn't meaningful — only a difference of 3 or more counts as real. And "you improved" is something Foreground only says once that holds across 3 scans in a row, not after one. Your report states this range plainly next to every number, not just here.

Who gets named instead, and the pages AI reads

Every niche result also lists the other businesses AI named across those 12 answers — counted the same way as your own mentions, so you can see exactly who's ahead and by how much.

Separately, Foreground records every web page that ChatGPT, Gemini, Claude, and Grok's searches returned or cited while answering your niche's questions — up to 15, most-read and most-cited first — and checks whether each one actually mentions your business. A page these models keep citing that doesn't mention you is a concrete, fixable gap, not a guess.

The action plan

Once a niche has a completed scan, you can generate an action plan: specific page copy, directory listings, and a review request, built from your own site and that scan's results. Every factual claim in it is checked against your site or the pages AI actually cited before it's shown to you — a second pass removes anything that doesn't hold up, rather than softening it. You can generate an action plan for any niche with a completed scan, including with the free check. Rather have us do the work? The done-for-you fix does it for you.

What to do if you're not named yet

Not being in the shortlist means AI isn't naming you for that customer niche yet. A few directions that tend to help:

• Check what these models are actually reading. Since answers are grounded in live web search, models tend to lean on sources that are easy to find and clearly written — review sites, directories, and your own site's content. If those sources don't mention you clearly, an AI model has nothing to cite. Your "pages AI reads" list shows exactly which ones.
• Make your own site answer the question directly. Content that plainly states what you do, who it's for, and how you compare to alternatives is easier for a grounded search to pick up than vague marketing copy.
• Look at who gets named instead. That's a direct signal of who AI currently considers the default answer for this customer niche — and which sites it trusts enough to cite.
• Generate your action plan. It turns the gap into specific, fact-checked things to publish — or we fix it for you.
• Track it over time, not just once. A single scan is a snapshot; scanning again later shows whether a change you made actually moved the count, using the honest 3-of-12 / 3-scans-in-a-row rule above.

Limitations, honestly stated

AI model outputs aren't fully deterministic or fully within anyone's control, including ours. A result reflects a real check against real model outputs at a real point in time — it's a strong directional signal, not a certified measurement, and it can shift between scans even with nothing changed on your end (see "How 'named in X of 12' is counted" above for exactly how much of a shift we treat as real).

This measures the API, not necessarily the consumer app. ChatGPT, Gemini, Claude, and Grok are each called directly against their own provider's API, grounded using that provider's own web-search tool — so this measures each model in its own real product, not a third party's approximation of it. It's still not identical to what a customer sees opening the ChatGPT app or the Gemini app: those apps can add things like business listing cards or their own Maps integration that a plain API call doesn't use. We report what the grounded API answer says, not what any specific consumer app's interface shows.

One asterisk worth naming: Gemini's citations come back as Google's own grounding-redirect links rather than the original source URL, so while Gemini's answer text is checked the same way as the others, we can't always tell exactly which page it read. Foreground also doesn't currently check Google's AI Overviews/AI Mode or Microsoft Copilot — AI-answer surfaces this scan doesn't cover. We'd rather state all of this plainly than let "4 AI models" imply more coverage than what's actually being measured.