How to Find Prompts Where You're Not Ranked in AmICited

The Not Ranked counter is arguably the most actionable number on the dashboard, because it’s not really a metric, it’s a list of missed opportunities with your name on them.

At a Glance

• Not Ranked is the last counter card on the Dashboard: a headcount of prompts where your brand has never been cited.
• A falling trend (e.g. ▼ -15.0) is good news; a rising one usually means new prompts were added, or a competitor pulled ahead.
• Sort the Prompts section by visibility to turn the counter into a prioritized to-do list.
• Open a not-ranked prompt to see which sources are winning it instead of you.
• Generate and publish content targeting that exact prompt, then watch the counter shrink over time.

What does “not ranked” mean in AI search?

When people talk about ranking in AI search, they mean something different from classic Google rankings. There’s no position-one, position-two ladder to climb. Instead, an AI assistant like ChatGPT, Perplexity, Gemini, or Google’s AI Overviews reads a question, synthesizes an answer from the sources it trusts, and either names your brand in that answer or doesn’t. If it names you, you’re “ranked” for that prompt, cited, mentioned, recommended, whatever word you want to use. If it doesn’t, you’re not ranked, full stop. There’s no page two to fall back on. You’re either part of the answer or you’re invisible.

This binary nature is what makes AI citation tracking so different from traditional SEO. A keyword you rank #8 for on Google still sends some traffic and still signals relevance. A prompt where you’re not ranked in AI search sends you nothing, no traffic, no mention, no brand exposure, because the AI simply picked other sources to build its answer from. Multiply that across the dozens or hundreds of prompts real buyers ask about your category, and “not ranked” stops being an abstract label and becomes a map of exactly where your competitors are winning attention you don’t even know you’re losing. It’s why an AI rank tracker that checks each prompt on a recurring cadence is the only reliable way to see the gap, rather than spot-checking a handful of questions by hand.

This is also why AI visibility monitoring tools track prompts rather than keywords. A prompt is the actual question someone types or speaks to an AI assistant, “what’s the best project management tool for a 10-person agency,” “how do I choose a GEO platform,” and each one either surfaces your brand or it doesn’t. Prompt tracking runs the same set of real-world questions against AI engines on a recurring schedule and records, prompt by prompt, whether you showed up. The Not Ranked counter is simply the running total of prompts where the answer, every single time it’s been checked, has been no.

Understanding this distinction matters because it reframes the fix. You can’t “optimize” your way into an AI answer the way you might tweak a meta description for Google. You have to give the AI engine a reason to consider you a trustworthy source for that specific question, which almost always means publishing content that directly answers it, backed by the kind of structured, citable detail these models pull from. That’s the work the rest of this guide walks through.

Where to find it

It’s the last of the four counter cards below the headline metrics on the Dashboard, labelled NOT RANKED. In the example it shows 114, “never cited yet”, with a ▼ -15.0 change versus the previous period. Because it sits at the far right of the counter row, it’s easy to skim past in favor of the more flattering numbers to its left, but it’s usually the one worth opening first.

What it measures

Every prompt in your tracked set falls into one of two buckets: prompts where you get cited, at least occasionally, and prompts where you never do. Not Ranked is a count of the second bucket at the moment the dashboard last refreshed. It isn’t an average, a percentage, or a weighted score; it’s a plain headcount of dead zones in your AI presence.

The trend arrow next to it tells you which direction that headcount is moving. A change like -15.0 is good news here: it means 15 fewer prompts are shut out than in the previous period, so you’ve started appearing for questions you previously missed entirely. If the number is rising instead, it usually means one of two things: you’ve added new prompts to your tracked set (which temporarily inflates the count until you catch up on them), or a competitor has out-published you on topics you used to at least partially cover. Reading the Not Ranked trend alongside your AI Visibility Score gives you a fuller picture: the Visibility Score tells you how you’re doing on the prompts where you do appear, while Not Ranked tells you how many prompts you haven’t even entered the race for yet.

Why it matters

These prompts are your clearest, lowest-ambiguity opportunities, for a few concrete reasons:

• They’re real questions people are already asking AI engines about your category, not hypothetical keywords, but demand that’s provably there because you’re tracking the prompt in the first place.
• Someone is being cited for each one. The demand is being met; it’s just not being met by you. That means there’s a known, visible bar to clear rather than an unproven market to create from scratch.
• Winning even a handful of them moves your numbers in two places at once: it shrinks Not Ranked and it lifts your overall Visibility Score, since a previously silent prompt starts contributing citations instead of zeros.

There’s also a competitive-intelligence angle worth not overlooking. Not Ranked prompts are effectively a live feed of where your share of voice is currently zero. Comparing that list against which competitors do appear on those same prompts tells you, topic by topic, who is currently defining the category in the eyes of AI assistants, and gives you a prioritized list of exactly where to contest that.

How to use it

1. Treat it as a to-do list, not just a stat. Head to the Prompts section and sort by visibility so the not-ranked prompts rise to the top. This turns an abstract counter into a concrete, ordered backlog you can work through.
2. See who’s winning them instead. Open a not-ranked prompt to view the actual AI answers and the sources being cited. This is the fastest way to reverse-engineer what “good” looks like for that question, the format, depth, and specificity the AI engine currently trusts enough to cite. If you want a repeatable process for this step across your whole prompt set, running a proper content gap analysis , comparing every prompt you track against what’s currently ranking on it, turns one-off spot checks into a systematic backlog.
3. Close the gap with content. Each not-ranked prompt is effectively a ready-made content brief : you already know the exact question, and you already know which competing sources currently answer it better. Generate an article in AmICited targeting that specific prompt, publish it, and let the model re-crawl and reassess on the next tracking run. AmICited’s SEO agent can take this a step further by drafting AI-optimized pages directly from a Not Ranked prompt rather than starting from a blank brief.
4. Watch the trend, not just the snapshot. A steadily shrinking Not Ranked number over several tracking cycles is one of the clearest, least ambiguous signals that your AI-visibility work is paying off, more reliable, in the short term, than watching your overall Visibility Score alone, since that score can be noisy while individual prompt-level wins accumulate underneath it.

If you’re building out this workflow for the first time, it helps to work backwards from the prompt set itself: a guide on how to find which prompts trigger AI to mention your brand walks through building and refining the list that Not Ranked is drawn from, so the counter reflects the questions that actually matter to your business rather than a generic starter set. And if Not Ranked consistently outnumbers the prompts where you do appear, that’s usually a sign it’s worth running a dedicated content gap analysis for AI search visibility rather than tackling prompts one at a time, it lets you batch the highest-impact gaps into a single publishing push instead of chipping away indefinitely. Either way, the counter itself is the diagnostic; the fix is always the same combination of understanding what the AI is currently citing and giving it a stronger, more specific reason to cite you next time.