How to check what AI says about your business

The manual approach

Open ChatGPT, Claude, or Gemini and type the questions your customers would ask. If you run a hotel, try "Best boutique hotels in [your area]". If you sell software, try "Best [your category] tools for [use case]". Read the response and see if your brand appears.

This works for a quick check, but it has problems:

• You can only check one model at a time
• Responses vary between sessions — you might get different answers tomorrow
• You can't track changes over time without a spreadsheet
• It doesn't scale beyond a handful of prompts
• The free tiers of some models may give different results than the API

The systematic approach

To properly monitor what AI says about you, you need to:

1. Write the right prompts

Think about what your potential customers actually ask AI. Not "Tell me about [brand]" — that's too direct. Instead, think about the discovery questions:

• "What are the best [category] in [location]?"
• "Which [product type] should I use for [use case]?"
• "Compare [your type of product] options for [audience]"
• "What do people recommend for [problem you solve]?"

These are the prompts where being mentioned (or not) directly affects whether someone discovers your business.

2. Test across multiple models

Each AI model has different training data and different biases. ChatGPT might recommend you while Gemini doesn't. You need to test the same prompts across all major models to understand your full visibility.

3. Track changes over time

AI recommendations aren't static. Model updates, new training data, and changes to your online presence all affect what AI says. Running the same prompts weekly or daily lets you spot trends — are you gaining visibility or losing it?

4. Watch your competitors

The brands that appear alongside yours (or instead of yours) are your competitors in AI. They might not be the same competitors you track in traditional marketing. AI might recommend brands you've never heard of.

Automating it with Cited Monitor

Cited Monitor does all of this automatically. You write your prompts once, connect your API keys for each provider, and schedule runs. Every response is parsed for brand mentions, competitor names, and citations. You get share of voice data, trend tracking, and alerts when things change.

Because you use your own API keys, you pay the LLM providers directly at their published rates. A typical run costs about $0.003 per model. No markup.