How prompts work

A prompt in SimplyRank is a question you send to AI models on a schedule. SimplyRank runs every active prompt against ChatGPT, Claude, Perplexity, and Gemini, then records who they mention, where, and how.

Prompts are the heart of the product. The data is only as useful as the questions are realistic.

The core idea

Prompts should look like things your real customers would type into ChatGPT or ask Perplexity. Not internal jargon. Not your brand name. Not vague placeholders.

✅ "What's the best private hospital in London for orthopaedic surgery?"

✅ "Should I use Notion or ClickUp for a remote design team?"

❌ "Tell me about [Your Brand]" — too direct; doesn't simulate discovery.

❌ "CRM software" — too short; AI gives generic output.

What makes a good prompt

1. Customer-intent. Imagine someone deciding between options. What would they type?

2. Specific. Add a qualifier — location, audience, budget, use case. "Best CRM for a 5-person agency under $50/user" beats "best CRM".

3. Comparison or recommendation. Most useful prompts ask for either "What's the best...", "Should I use X or Y...", "What's an alternative to...", or "Recommend a...".

4. No brand bias. Don't include your own brand name unless you specifically want to track how AI describes you when asked directly.

5. Realistic length. A sentence or two. AI gives different answers to one-word queries vs. fully-formed questions, and your customers ask the latter.

Prompt patterns that work

Pattern	Example
Best in category	"What's the best AI rank tracker for a small marketing team?"
Comparison	"Should I use Ahrefs or Semrush for content audits?"
Alternatives	"What's a cheaper alternative to Mailchimp for a 10k-list newsletter?"
Use case	"What CMS should I use for a B2B SaaS landing page?"
Geography-specific	"Top private dentists in Manchester for cosmetic work"
Audience-specific	"Best invoicing software for freelance designers"
Problem-led	"How do I track AI brand mentions across ChatGPT and Perplexity?"

How many prompts do you need?

It depends on how broad your category is.

Category breadth	Suggested prompt count
Narrow (e.g. one product, one geography)	15–30
Medium (multiple use cases, one country)	30–80
Broad (multiple regions, audience segments)	80–200
Agency / multi-brand	25–50 per client

Too few prompts and your visibility score is noisy — one missed mention swings the average. Too many and the data becomes hard to act on.

Your plan determines the maximum:

Plan	Max prompts
Trial	10
Starter Lite	25
Starter	50
Pro	150
Advanced	300

Categories — organising your prompts

You can group prompts into categories to filter results by topic. Examples:

A private hospital might use: General queries, Cosmetic surgery, Orthopaedics, Maternity.
A SaaS company might use: Direct comparisons, Use case prompts, Pricing prompts, Alternative-to prompts.
An agency tracking a client might use: Brand awareness, Service-specific, Local SEO.

Categories make it easier to spot patterns ("we're strong in Use case prompts but weak in Direct comparisons").

See Prompts and categories.

How often prompts run

Prompts run on a schedule defined by your plan:

Plan	Available frequencies
Trial	One instant scan only
Starter Lite	Weekly
Starter	Daily, weekly
Pro	Daily, weekly, monthly
Advanced	Daily, weekly, monthly

You can also trigger a manual scan from the dashboard at any time (within plan limits).