Audiences
Tags answer what kind of question is this? Audiences answer who's asking it?
An audience is a buyer or customer segment - Enterprise IT Buyer, Solo Indie Hacker, Marketing Agency Owner. They're the kinds of people who'd realistically ask the questions you track, and Trakkr sorts every prompt into one of them, so you can read your visibility by who is asking instead of just by topic.
Audiences live on the Prompts page, in the Audiences tab next to Prompts and Tags.
Here's why that matters. Say your overall {visibility score|How prominently AI models mention your brand across your tracked prompts and models, scored 0-100.} is 47%. That's a single number, and it hides a lot. Maybe you're at 75% with enterprise IT buyers and 22% with indie hackers - two different problems with two different fixes. Maybe you're strong when people are first looking and invisible by the time they're choosing. Audiences split that one number into a handful of separate stories, each tied to a specific kind of customer.
How audiences are built
Once you have ten or more active prompts, Trakkr generates 3-8 audiences using Gemini. It reads your tracked prompts and brand profile and works out the distinct kinds of people behind the questions. It takes about 20 seconds and runs automatically the first time you cross the threshold - you don't have to ask for it.
Each audience comes with:
- A short name and one-sentence description
- A 2-3 sentence bio of who they are and what they care about
- A buyer journey stage - awareness, consideration, or decision
- 2-3 prompts they'd likely ask that you aren't tracking yet
Every prompt belongs to exactly one audience. If one looks misfiled, move it from the prompt's actions menu - your overrides stick across regenerations, so the fix is permanent.
How the stage is decided
The stage answers where in the buying process is this person? There are three of them, and Trakkr reads them off the intent of the questions, not from a guess about the person:
| Stage | The mindset | Questions that signal it |
|---|---|---|
| Awareness | Just learning the space | "what is...", "how do I...", broad discovery questions |
| Consideration | Weighing the options | comparisons, "alternatives to...", "best X for Y" |
| Decision | Close to choosing | pricing, "which should I pick", buying-intent questions |
Two things use these stages, and the difference is worth knowing:
- The stage badge on each audience row is where that audience mostly sits - the model's read of the mix of questions they ask.
- The Journey view scores an audience at all three stages by sorting its individual prompts into the funnel by intent, so one audience can be strong at awareness and weak at decision.
So when an audience reads as "Consideration," it's because the questions it asks are mostly comparison-and-evaluation ones - people weighing you against the field, not first discovering the category or ready to sign. (Set the wrong stage? Open the audience and change it by hand - see below.)
The Audiences tab
The tab gives you four views of the same audiences. The spread stat beside them is the headline number: the gap between your best and worst audience, the thing a single visibility score hides.
- Overview is the main table - every audience with its visibility score, {sparkline|A small inline chart of the audience's visibility over the last seven days.}, trend, stage badge, prompt count, and top prompt.
- Journey swaps those data columns for the funnel - awareness, consideration, decision - as a heatmap, so you can see exactly where an audience fades. Strong in awareness but weak at decision usually means AI understands your category role but doesn't trust you enough to recommend you when buyers are close to choosing.
- Ask Gap (the Simulate button) takes one buyer need, rewrites it in each audience's own words, and runs every phrasing across the engines. You can be #1 for how one audience asks and invisible for how another asks the same thing - and any phrasing you're dark on can be tracked as a prompt in one click.
- Discover sits at the bottom of the table and surfaces audiences your market clearly serves but you don't track yet, inferred from competitors winning in your space and the topics AI cites around you. Each comes with evidence and starter prompts; adding one starts tracking it.
The audience brief
Click any audience row to expand its brief inline. You get the bio, your visibility and trend with a chart, the journey split, the prompts ranked best-to-worst, competitor alerts for anyone winning more than 15% of this audience's prompts, the top sources AI cites for this audience, and a few prompts to track next.
Read it like a brief, not a status report. The competitor alerts and top sources are usually the most actionable things in it.
Custom audiences and regenerating
Auto-generated audiences are a starting point. Click New to add one by hand - useful for a key account, a vertical, or an ICP segment you're betting on. Assign prompts to it from the Prompts page.
Custom audiences are never overwritten. Auto-generated ones are - click Regenerate whenever you want a fresh take, or wait for the stale banner that appears once your prompt list has shifted significantly.
A few common questions
Why don't I see any audiences? You need ten active prompts. Below that, the Audiences tab shows your progress toward the threshold.
Can a prompt belong to two audiences? No - each prompt sits in exactly one. For overlapping organization, use tags.
Can I change an audience's stage? Yes. Open the audience's editor and pick a different stage. Regenerating reassigns auto-generated stages from scratch, so editing by hand is how you pin one in place.
My audiences feel generic. Fill out your brand profile (description, industry, domain) so Gemini has context, and spread your prompts across stages. If every prompt is a comparison, every audience will look like "buyers comparing options."