TL;DR. A prompt group is the AI Search equivalent of a keyword cluster, built around intent and scenario, not search volume. For each segment, you build one prompt group with three dimensions (visibility hygiene, intent, scope). This gives you 12 trackable prompts (3x2x2) that surface where your brand wins and where it is invisible. The shorter (without filler words like 'what', 'how', …) and more scenario-specific the prompt, the more stable the output.
What we are building toward (Mint Mobile worked example)
Before we unpack each dimension, here is the full 3x2x2 = 12-prompt group you will learn to build in this module. This is the canonical pattern from the Rankscale Playbook.
| Intent | Scope | Non-branded prompt | Branded prompt |
|---|---|---|---|
| Information | Segment | mobile phone plans | mobile phone plans Mint Mobile |
| Consideration | Segment | best mobile phone plans | best mobile phone plans Mint Mobile |
| Transaction | Segment | mobile phone plans price | mobile phone plans price Mint Mobile |
| Information | Niche (prepaid) | prepaid mobile phone plans | prepaid mobile phone plans Mint Mobile |
| Consideration | Niche (prepaid) | prepaid best mobile phone plans | prepaid best mobile phone plans Mint Mobile |
| Transaction | Niche (prepaid) | prepaid mobile phone plans price | prepaid mobile phone plans price Mint Mobile |
Three intents x two scopes x two visibility-hygiene states (branded / non-branded) = 12 tracked variations. All prompts are short (2–5 words). That is deliberate, and the rest of this module explains why and how to build your own.
Keyword vs prompt
SEO keyword: "CRM small business."
Prompts in AI Search are different. AI engines deduce intent with high precision from very short queries. The shorter and more scenario-specific the prompt, the more stable the output. You do not write long natural-language questions for tracking. You track short, scenario-specific variations that the engine will expand internally. Don't use filler words like what, how, … It increases variability.
Why short prompts
AI engines are probabilistic. Each additional word increases the likelihood of a different result. Long natural-language questions like "Which CRM is tailored to a small business with 10 employees?" or "What is the best CRM for a small business with 10 employees?" are useful to illustrate that engines read intent, but they introduce volatility when tracked directly. Keep your tracked prompts to the core scenario. Let the intent and scope tags do the work.
The prompt group structure
| Field | Example |
|---|---|
| Region | United States |
| Segment | Mobile phone plans |
| Core Prompt (Group) | mobile phone plans |
| Tags (reflect scenario) | Non-branded, Consideration, Niche (prepaid) |
| Prompt Variation | prepaid best mobile phone plans |
| AI Engine | ChatGPT |
| Frequency | Daily |
Do this now:
Pick one segment. Write down the segment name as a short 2–4 word phrase. That phrase is the core of your first prompt group. You will build 12 variations in the scope dimension lesson and the lessons that follow.
Start improving your AI visibility today with Rankscale.
Get started