How to Read Your Prompts Overview Stats in AmICited

The row of stats at the top of the Prompts page is your monitoring at a glance: how many questions you track, how many earn citations, and how much work AmICited is doing behind the scenes.

At a Glance

• The Prompts page stats bar shows six numbers: Tracked, Cited, Not ranked, Avg rank, Runs, Engines.
• Tracked is your prompt library size; Cited vs. Not ranked is your citation scoreboard.
• Avg rank shows how well you rank when you are cited; Runs and Engines show data depth.
• Watch Cited climb and Not ranked fall together, that’s the clearest sign of GEO progress.
• Drill into the Tracked prompts table below to see which specific prompts are driving the numbers.

What is a prompt, and why track it?

A prompt, in the context of AI search visibility , is a real question someone might type into ChatGPT, Perplexity, Gemini, or Google AI Overviews: things like “best project management software for small teams” or “who are the top GEO tools in 2026.” Prompt tracking is the practice of monitoring a defined set of these prompts on a recurring schedule, running them against multiple AI engines, and recording whether your brand shows up in the answer, where it ranks, and what it’s cited alongside. It’s the AI-era equivalent of keyword rank tracking, except the unit of measurement is a natural-language question rather than a search term, and the “result” is a generated answer rather than a list of blue links.

When an AI engine answers a prompt, it doesn’t always name a source. When it does (mentioning a brand, linking to a page, or attributing a claim), that’s a citation. An AI citation is the core signal GEO tools measure, because it’s the closest thing to a ranking position that generative engines expose. Unlike traditional search, there’s no fixed results page to check; the same prompt can return different answers on different days, across different engines, and even across different accounts. That variability is exactly why prompt tracking exists: a single manual check tells you almost nothing, but running the same prompt set repeatedly, across engines, over time, turns noisy individual answers into a reliable trend.

This matters for AI search visibility because prompts are the demand side of GEO. Your content, schema, and entity signals only translate into business value if they surface at the moment someone asks a relevant question in an AI assistant. A well-built prompt set mirrors the actual questions your buyers ask (comparison questions, “best X for Y” questions, troubleshooting questions) so that your citation data reflects real opportunity, not vanity metrics. The Prompts overview stats bar is where AmICited condenses all of that activity, across however many prompts you’re tracking, across every engine, across every run, into six numbers you can read in a few seconds.

Where to find it

It’s the pill-shaped stats bar directly under the Prompts page title, above the Brand voice chart. It’s visible as soon as you land on the Prompts page, no filtering or drill-down required, because it’s meant to be the first thing you check before you go looking at individual prompts or engines.

What each stat means

• Tracked (e.g. 114): the total number of prompts you’re monitoring. This is the size of your prompt library: the full set of realistic buyer questions AmICited is running on your behalf. A bigger, more representative library means your Cited and Not ranked numbers reflect the real shape of your market rather than a narrow slice of it.
• Cited (e.g. 15): how many of those prompts your brand gets cited for at least sometimes. This is your raw citation count, the precursor to a citation rate : the share of tracked prompts where an AI engine has named or linked your brand in its answer, even if only occasionally rather than every single run.
• Not ranked (e.g. 99): how many you’re never cited for yet. These are your opportunities: prompts where competitors may already own the answer, or where no one has cracked it yet, and where new content, better schema, or stronger entity signals could move the needle.
• Avg rank (e.g. #2.3): your average citation position across the prompts where you do appear (lower is better). This is the closest analogue to a traditional SEO rank tracker’s average position metric, adapted for AI answers where citation order still carries meaning: being cited first or second in an answer typically gets more attention than being mentioned last.
• Runs (e.g. 606): the total number of times your prompts have been executed across all engines. Every run is a fresh check of a prompt against a specific engine on a specific day, which is what gives you a defensible trend line instead of a one-off snapshot.
• Engines (e.g. 4): how many AI platforms your prompts are checked against. Visibility on ChatGPT alone doesn’t tell you much about Perplexity, Gemini, or AI Overviews, since each engine retrieves, weighs, and cites sources differently, so tracking the same prompt set across multiple engines is what turns this into a genuine AI rank tracker rather than a single-platform check.

How to use it

1. Watch Cited climb and Not ranked fall. That movement is the clearest sign your AI-visibility work is landing. Because both numbers are drawn from the same Tracked total, they move in tandem: every prompt you newly win moves directly from one column to the other, which makes this pairing a simple, honest scoreboard for GEO progress over a quarter or a campaign.
2. Sanity-check Tracked. Too few prompts and you’re not seeing the whole picture; add more to cover your space. A prompt library that only covers your brand name and a handful of obvious product terms will overstate your visibility, because it skips the comparison and “best of” questions where most citation battles are actually won or lost. If you’re not sure which questions to add, it helps to systematically find which prompts trigger AI brand mentions for your category before expanding the list.
3. Use Avg rank as a quality gauge. Being cited is good; being cited near the top (a low Avg rank) is better, since citation position tends to correlate with how much weight an AI answer gives a source: a brand cited first in an AI Overview is doing more commercial work than one buried as a fourth mention.
4. Cross-reference Runs and Engines against your monitoring cadence. If Runs looks low relative to how long you’ve been tracking, or Engines hasn’t grown even though new AI assistants have gained adoption, that’s a sign to revisit your monitoring setup rather than your content.
5. Treat the stats bar as your entry point, not your endpoint. It tells you that something changed; the Tracked prompts table tells you which prompts changed and why.

For the detail behind these totals, drop down to the Tracked prompts table and filter by Cited or Not ranked. That table breaks the same numbers down prompt by prompt, so you can see exactly which questions you’re winning, which engines are citing you on each one, and where a Not ranked prompt might just need one well-targeted page to flip into a citation. If you want a more structured way to work through that list, rather than scanning it prompt by prompt, it’s worth learning how to measure AI search visibility using the fuller set of metrics and KPIs behind these headline numbers, including how citation rate, share of voice, and average rank relate to each other.

Once you’re comfortable reading the stats bar, the next useful step is usually improving what feeds it. That means treating your prompt set as a living asset: periodically reviewing it against new competitor launches, seasonal questions, and shifts in how your buyers phrase things, the same way you’d maintain a keyword list for traditional SEO. Teams managing this across multiple client accounts or brands tend to lean on AmICited’s agency workspace to keep each portfolio’s Tracked, Cited, and Avg rank numbers separate and comparable side by side. And if the Not ranked column is stubbornly large, that’s less a monitoring problem than a content and structure problem, worth pairing with a look at how competitors are winning those same prompts before you book a demo to see how AmICited’s full prompt and citation workflow fits your reporting cadence.