What should I look for when reading a case study?
Short answer
Look for six things: the starting state, the questions tested, the platforms, the metric definitions, the timeframe and the attribution method. A case study missing several of them is closer to a testimonial.
- Starting state: what AI said about the business before the work, including whether it was absent.
- Questions tested: the actual prompts, such as best Korean fine dining in New York. Each case study page lists them, so you can run them yourself.
- Platforms: which AI tools were checked, since ChatGPT, Gemini, Claude and Perplexity can answer the same question differently.
- Metric definitions: what exactly is counted, covered in the next section.
- Timeframe: the period between baseline and result. Atomix states 90 days after deployment, and the other five state a 3 to 6 month engagement.
- Attribution: how a business outcome such as revenue or sign-ups was linked to AI discovery.