How to Measure Your Brand's Visibility in AI Search
To measure your brand's visibility in AI search, you test the real prompts your buyers use across ChatGPT, Gemini and Perplexity, then record whether your brand appears, how it is described, and who is cited instead of you. Tracked over time, those three data points become a clear picture of your AI share of voice and whether your GEO work is paying off.
Most brands cannot answer a simple question: when a customer asks an AI engine about our category, do we show up? The five steps below give you a repeatable way to answer it, the measurement half of the discipline in our citation playbook.
Why measuring AI visibility is different
Traditional SEO measurement assumes a ranking and a click. AI search has neither in the usual sense. There is no single position to track and often no click at all, just an answer the user reads and acts on. So the metrics have to change.
Instead of rank and clicks, AI search visibility is measured by presence and framing. Are you in the answer? Are you described accurately? Are you recommended, mentioned in passing, or absent while a competitor is named? These are the questions that map to revenue now, and they require a different method than a rank tracker.
Step 1: Build your prompt set
Start with the real prompts your buyers would type. Pull them from your keyword research, your sales calls and your own customers, then write them as natural questions rather than keywords.
A good starting set has three layers. Decision prompts name the need directly, such as "best AI search agency in Indonesia." Consideration prompts compare or ask how, such as "how do I get my brand into ChatGPT." Category prompts define the space, such as "what is generative engine optimization." Aim for fifteen to thirty prompts that together represent how a customer actually moves toward you.
Step 2: Test across the major engines
Run every prompt through each engine that matters to your market: ChatGPT, Gemini, Perplexity, and Google AI Overviews. Test in both Bahasa Indonesia and English if your buyers use both, because answers can differ by language.
Keep conditions consistent so results are comparable over time. Use the same prompts, the same engines and a fresh session each round. Engines vary their answers, so where it matters, run a prompt two or three times and note the range rather than a single snapshot.
Step 3: Record three things for every prompt
For each prompt and engine, capture presence, framing and competitors. Presence is whether your brand appears at all. Framing is how you are described and whether it is accurate. Competitors is who else is named, especially anyone cited ahead of you.
A simple scoring scale keeps this consistent: recommended as a top option, mentioned but not recommended, or absent. Logging the competitor cited in your place is the most actionable part, because that list tells you exactly who to study and outrank.
Step 4: Calculate your AI share of voice
Turn the log into a single trackable number. AI share of voice is the percentage of your prompt set where your brand appears, ideally weighted so decision prompts count more than category prompts.
This number is your headline metric. One reading tells you where you stand. A series of readings tells you whether your GEO program is working. Watch the trend, not the absolute, and pay special attention to movement on the high-intent decision prompts, since those are closest to revenue.
Step 5: Track it on a cadence and act
Re-run the full test every two to four weeks, log the results in the same place, and turn the gaps into content work. Each prompt where a competitor is cited and you are not is a brief waiting to be written.
The loop is simple: measure, find the gap, publish citation-ready content to close it, then measure again. Over time you should see absent turn into mentioned and mentioned turn into recommended. Want it run for you as an ongoing program? That is part of our AI search work in Indonesia. The content side of closing those gaps is covered in our guide on how to use AI for SEO.
