Track: AEO & GEO · Format: text lesson · Est. read: 11 minutes
By the end of this lesson you'll be able to open an AI-visibility report, read the two numbers that carry most of its weight, and point to the single biggest gap to fix first.
A report is only useful if it changes what you do on Monday. So this lesson ends on one action, not a dashboard tour. The metrics link to the Knowledgebase if you want the formal definition.
The two numbers that matter most
Most AI-visibility reports put a lot on the screen. Two metrics carry the story, and if you only ever read these two you'll be fine.
Citation share is the share of AI answers on your topic that name your brand, counted against every answer on that topic, yours and your rivals' alike. Run a category question across ChatGPT, Perplexity, and Google's AI Overview a hundred times. Each answer either mentions you or it doesn't. If you show up in eleven, your citation share is 11%. The other 89% went to someone else. This is the closest thing AI search has to a ranking position. See citation share for the full definition.
Prompt coverage is different. It measures how many of the distinct questions buyers ask about your category you have a citable answer for at all. If buyers ask forty questions and your content cleanly answers fifteen, your coverage is 37.5%. Coverage is about breadth. Citation share is about how often you win the questions you already compete on. See prompt coverage.
Hold the difference in your head, because it's the whole game. Citation share asks how often you win. Prompt coverage asks how many questions you even show up for. A brand can score well on one and badly on the other, and the fix is completely different depending on which.
Both of these roll up into what the market loosely calls LLM visibility, the umbrella term on the cover of most reports.
"Good" is relative to your category
Here's the first mistake people make. They see 14% citation share and their heart sinks. Fourteen out of a hundred. Sounds like failure.
It isn't, on its own. Citation share is a share, so it's split across everyone competing for the same answers. In a category with six real players, an even split would be about 17% each. Against that baseline, 14% is a close second or third, not a disaster. In a crowded category with twenty vendors, an even split is 5%, and 14% would make you the clear leader.
The number means nothing until you set it next to two things: how many competitors split the pie, and where the leader sits. A 14% share trailing a leader who holds 15% is a race you're winning. The same 14% trailing a leader on 48% is a category with a runaway winner and a lot of ground to make up. Read the gap to the top, not the raw percentage.
This is why the competitor columns in the report matter more than your own row. They're the benchmark. Ignore them and you can't tell a strong number from a weak one.
Prompt coverage shows you the questions you're missing
Citation share tells you how you're doing on the questions you compete for. It says nothing about the questions you never show up for at all. That blind spot is what prompt coverage exposes.
A coverage report lists the buyer questions the tool tested and marks, for each, whether any of your pages earned a citation. The rows where you're absent are the interesting ones. Each is a question your buyers are asking an engine right now, and getting an answer that names a competitor instead of you.
A coverage gap is a different kind of problem from a citation-share gap, and this distinction is worth slowing down for. If you have a page on a topic and still aren't cited, that's a quality problem: the page exists but isn't good enough, clear enough, or trusted enough to win. You fix it by reworking the page. If you have no page on the topic at all, that's a coverage gap. No amount of editing fixes it, because the fix is a page that doesn't exist yet.
That's often good news. Writing a missing page is usually cheaper and faster than clawing citation share off an entrenched competitor on a question you already contest. Coverage tends to be the cheapest lever a team has.
A worked example, start to finish
Let's read an actual report. The numbers below are invented, a small dummy dataset for a fictional company. Treat them as illustrative, not real TurboDemand data. But the way we reason through them is exactly how you'd read your own.
Meet Northstar Analytics, a mid-market B2B analytics vendor. Their AI-visibility tool tested 40 buyer prompts across ChatGPT, Perplexity, and AI Overviews. Here's the summary.
Citation share, by competitor
| Brand | Citation share |
|---|---|
| Datava (leader) | 34% |
| Northstar (you) | 12% |
| Quantly | 22% |
| Metriq | 18% |
| Everyone else | 14% |
Coverage, by funnel stage
| Funnel stage | Prompts tested | Northstar cited | Coverage |
|---|---|---|---|
| Top (what is / why) | 14 | 11 | 79% |
| Middle (how / compare) | 16 | 6 | 38% |
| Bottom (pricing / vs.) | 10 | 2 | 20% |
| Overall | 40 | 19 | 48% |
Read the citation table first. Northstar sits at 12%, fourth of four named players, behind a leader on 34%. Four vendors carry most of the answers, so an even split would be roughly 25%. Northstar is at about half its fair share. Not a rounding-error gap. A real one.
Now the coverage table, which tells you where the gap lives. Top-of-funnel coverage is healthy at 79%. Northstar shows up fine when someone asks what analytics tooling is or why it matters. The floor falls out lower down. Middle-of-funnel is 38%, and bottom-of-funnel, the pricing and head-to-head comparison questions, is 20%. Two citations out of ten prompts.
That's the diagnosis. Northstar's problem isn't broad invisibility. It's a brand that answers the easy early questions well and goes quiet exactly where buyers decide who to buy. The 12% citation share and the 20% bottom-funnel coverage are the same wound seen from two angles. Buyers researching a purchase ask "how much does Northstar cost" and "Northstar vs. Datava," find nothing citable, and the engine fills the silence with a competitor.
So what do you fix first? Not the top of funnel, where you're already strong. Not the middle, yet. The bottom-funnel coverage gap. It's the cheapest to close (write pages that don't exist), it sits on the highest-intent questions (buyers near a decision), and it's the most direct route to lifting that 12% share where it counts for revenue. One gap, chosen on evidence, not vibes.
From a report row to a content brief
A gap you can name but don't act on is just anxiety. The last step is turning one row into a brief someone can write from this week.
Take Northstar's worst-performing bottom-funnel prompt. Say the report flags "Northstar vs. Datava" as a question they're never cited for. That single row becomes:
Page to write: Northstar vs. Datava, an honest comparison Question it answers: "How does Northstar compare to Datava, and which is right for my team?" Must state plainly, high on the page: what each tool is best at, where each falls short, rough pricing for both, and who should pick which. Why it'll get cited: it's ungated, names the competitor directly, and states real tradeoffs, so an engine can lift a clean, trustworthy passage. That trust signal is source authority doing its job. Success metric: Northstar appears in the answer to that prompt within the next measurement cycle, and bottom-funnel coverage ticks up from 20%.
One row, one brief, one measurable outcome. Do that for the three or four highest-intent gaps and you have a quarter's content plan drawn straight from the data instead of guesswork.
The one thing to take away
Read two numbers and one comparison. Citation share against your competitors tells you how often you win the questions you compete on. Prompt coverage tells you how many questions you show up for at all. Set both against the category, find the stage where high buyer intent meets low coverage, and write the missing page there first.
That intersection, high intent and low coverage, is almost always your first move. It's the cheapest gap to close and the one nearest the sale.
What's next
You've got the flagship AI Visibility 101 for why any of this works, and the mechanics of writing a page that earns the citation live under answer engine optimization and generative engine optimization. Closing the highest-intent gaps is also where demand capture starts to pay off.
This lesson feeds the AI Visibility Analyst certification, which tests whether you can read a report like Northstar's and name the right first move. Take the quiz in quiz.md to check you can.