How to measure whether AI engines cite your brand

Quick answer

What AI visibility actually measures

AI visibility is how often, and how favorably, AI answer engines surface your brand when people ask questions in your category. Unlike a search ranking, there is no single position to read, so visibility is measured across a set of questions and engines rather than for one keyword.

The core signals are simple to describe. For any given prompt, an engine either cites you with a link, mentions you without a link, or leaves you out. Measuring visibility means turning many of those individual outcomes into numbers you can compare over time and against competitors.

Because engines answer the same question differently and update frequently, a single check tells you very little. The value comes from a consistent method run repeatedly, so you can distinguish real movement from noise.

Step 1: Define your category prompts

Start with the questions your buyers genuinely ask, not the keywords you wish you ranked for. Cover the journey from broad category framing to specific comparisons and how-to queries, and aim for a representative set of roughly 15 to 30 prompts.

Keep this prompt set stable. It becomes your benchmark, so changing it constantly makes results impossible to compare. Add prompts deliberately and note when you do, the same way you would version a test suite.

• Include category overview questions such as best tools for a job.
• Include head-to-head and alternative comparison questions.
• Include practical how-to and what-is questions buyers ask early.
• Write prompts in natural language, the way a real user would type them.

Step 2: Query each engine and record the outcome

Run every prompt across the engines your buyers use, then capture the full answer text. The answer matters as much as the citation, because being described inaccurately is a different problem from not appearing at all.

For each prompt and engine, classify the result so you can count it later. A consistent classification scheme is what lets you aggregate dozens of qualitative answers into a clear picture.

• Cited: your brand appears with a link or explicit source attribution.
• Mentioned: your brand is named but not linked or sourced.
• Absent: your brand does not appear in the answer.
• Note the sentiment and accuracy of any mention, not just its presence.

Step 3: Track citations and share of voice

Two metrics carry most of the weight. Citation frequency is how often you appear across the prompt set, and share of voice is your mentions as a fraction of all brand mentions for those same prompts. Together they tell you both your absolute presence and your relative position.

Share of voice is especially useful because it is comparable across categories and over time. A rising citation count means little if every competitor rose faster; share of voice captures whether you are actually gaining ground.

Slice these metrics by prompt type and by engine. You may be strong on how-to questions but invisible on comparisons, or well cited in one engine and absent in another, and those gaps are where the next round of work goes.

Step 4: Benchmark against competitors

Visibility is relative, so always record who else shows up. When you log every brand named in each answer, you can see which competitors own which questions and how concentrated the field is.

This competitive view turns measurement into a roadmap. If a rival consistently appears for a set of prompts where you are absent, those prompts and the sources behind them become priorities, since the engines clearly find that competitor's content the better answer there.

Step 5: Trend it over time and automate the loop

A single measurement is a snapshot; the signal is in the trend. Re-run the same prompt set on a fixed cadence, weekly or monthly depending on how fast your category moves, and compare runs so you can connect content changes to visibility changes.

Running dozens of prompts across several engines by hand, classifying each result, and computing share of voice every cycle is slow and easy to do inconsistently. That repetition is exactly what tooling is for.

AI-visibility tracking tools, including Dreamstate, automate this loop: they run your category prompts across engines on a schedule and report citations, share of voice, and competitor movement, so you spend your time acting on the data rather than collecting it.

Step by step

1. Define your category prompts Write the real questions buyers ask in your category, covering broad overviews, comparisons, and specific how-to queries. A representative set of 15 to 30 prompts gives you a stable benchmark to measure against.
2. Choose the engines that matter to you Pick the answer engines your buyers actually use, which today typically includes ChatGPT, Perplexity, Gemini, and Google AI Overviews. Measure consistently across the same set each time.
3. Run each prompt and record the outcome Ask every prompt in every engine and capture the full answer. For each one, log whether your brand was cited with a link, mentioned without a link, or absent entirely.
4. Track citations and share of voice Count how often you appear and compute your share of voice as your mentions divided by total brand mentions across the prompt set. This turns scattered answers into a comparable number.
5. Benchmark against competitors Record which competitors appear in the same answers and how often. Their presence tells you where you are being displaced and which prompts to prioritize next.
6. Trend it over time Re-run the same prompt set on a fixed schedule and compare results. Movement across runs, not a single snapshot, is what tells you whether your work is paying off.

Frequently asked questions

What is share of voice in AI visibility?

Share of voice is the proportion of brand mentions that are yours across a defined set of prompts and engines. If ten brand mentions appear across your prompt set and three are yours, your share of voice is thirty percent. It is useful because it measures your position relative to competitors, not just your raw count.

How often should I measure AI visibility?

Re-run your prompt set on a fixed cadence, commonly weekly or monthly depending on how quickly your category changes. Consistency matters more than frequency, because the goal is to compare like-for-like runs and detect real movement rather than noise.

Why do I get different answers each time I ask the same question?

Answer engines are probabilistic and update their sources and models frequently, so responses vary between runs. That is why a single check is unreliable and why you measure across a stable prompt set repeatedly, then look at trends rather than any one answer.

Can I measure AI visibility manually?

Yes. You can ask each prompt in each engine and log the results in a spreadsheet, classifying every answer as cited, mentioned, or absent. The tradeoff is time and consistency across many prompts and engines, which is why teams often automate the measurement loop.