Branded, Organic, and Competitor Prompts: How to Segment AI Visibility
One blended share of voice number hides more than it shows. How to split tracked prompts by type and intent, why volume reads differently in each segment, and what to report for each.
The fastest way to look good in an AI visibility report is to track a lot of prompts with your brand name in them. Assistants mention you when asked about you. Your share of voice climbs, and it tells you nothing about whether buyers who don't know you yet will find you.
Segmenting prompts fixes this. It also changes how you read prompt volume, because demand behaves differently in each segment.
The three prompt types
Promptwatch tags each tracked prompt with a Prompt Type:
- Organic. Category questions without your brand in them. "What's the best invoicing tool for freelancers?"
- Brand Specific. Questions about you by name. "Does [brand] integrate with Xero?"
- Competitor Comparison. Questions that pit you against a named rival. "[Brand] vs [rival] for small teams."
It also tags an Intent Type: Branded, Informational, Navigational, Commercial, or Transactional. The two layers combine. An organic commercial prompt ("best invoicing tool for freelancers") is a different job from an organic informational one ("how do freelancers handle VAT on invoices?").
With both tags in place, Promptwatch shows share of voice broken down by how and why someone is asking, instead of one blended number.
How volume reads in each segment
Organic prompts are where volume matters most. These are the questions new buyers ask, and the volume band is your best estimate of how many of them there are. Prioritize by volume and difficulty here, and expect most of your growth work to land in this segment.
Brand-specific prompts often show low or no volume, and that's fine. Their job is accuracy, not reach. You track them to catch wrong pricing, missing features, or outdated claims. Promptwatch's fact-check feature compares what AI says about your brand against your knowledge base and shows how often a wrong claim appears. Don't drop a brand prompt for low volume.
Competitor comparison prompts sit in between. Volume can be modest, but intent is high, because someone asking for a head-to-head is close to a decision. Prioritize the comparisons your sales team hears most, even at one bar of volume.
What to report per segment
A single dashboard number hides the story. Report each segment against the metric that fits it.
| Segment | Lead metric | What a bad number means |
|---|---|---|
| Organic | Share of voice, visibility score | Coverage problem: you're not in the answers |
| Brand specific | Sentiment, accuracy of claims | Reputation or facts problem |
| Competitor comparison | Position in answer, sentiment | You're mentioned but framed as second choice |
This lines up with how Promptwatch's own metrics guide reads its numbers: low share of voice means a coverage problem, a late position means a prominence problem, and weak sentiment means a reputation problem.
Building a balanced list
Promptwatch recommends starting with 15 to 30 prompts per monitor or market. A mix that works for most brands:
- About two thirds organic, ordered by volume and difficulty.
- A handful of brand-specific accuracy checks, one per fact you can't afford to get wrong (pricing, core integrations, compliance).
- A few competitor comparisons against the rivals you lose deals to.
Use topics and tags on top of type and intent so you can filter by product line or market without rebuilding the list.
Common mistakes
Counting brand prompts toward organic share of voice. This inflates the headline number. Keep segments separate in every report.
Dropping comparison prompts for low volume. They're often the prompts closest to revenue. Judge them by sales relevance.
Tracking only your own comparisons. Add prompts comparing two competitors without you. If assistants recommend a third brand in those answers, you want to know.
The tooling question
You can approximate segments with a spreadsheet and manual tags, but the payoff comes from filtering every report by segment. Promptwatch supports type and intent tagging, topics, and tags on every prompt, with volume and difficulty attached, and its reports and PDF exports can be filtered by prompt type. Essential is $95/mo for 50 prompts with a 7-day trial, enough to build a segmented set for one market. Compare alternatives on our tool rankings.