How AI decides which brands it recommends
How does ChatGPT decide which brands to mention and recommend? A plain-language explanation of language models, memory, live search and why testing once tells you little.

About
Co-founder
In search since 2008. Started with SEA and SEO, moved with the field through YouTube and social search, and spent the last two years reading how AI models decide what to recommend. He wrote down why we built Recommendable.
How does ChatGPT decide which brands to mention and recommend? A plain-language explanation of language models, memory, live search and why testing once tells you little.
A website was always built for people who look and click. Recently, a new and very important user has joined them: the AI agent that performs tasks on behalf of the customer.
Admit it, you've probably asked ChatGPT which solution is best in your market. Your competitor came up. You didn't. The reflex is predictable: create more content, start tracking prompts, get a GEO tool.
Since ChatGPT and other LLMs (Perplexity, Claude, Gemini) have been introduced, a fierce debate has raged between two camps. Both agree that these platforms are emerging as new search engines, but what really matters is how do you get your brand recommended?
Since ChatGPT and other AI tools emerged as search platforms, a whole new industry has emerged: Generative Engine Optimization, or GEO (we prefer to call it AI Search). The promise is tempting. People search differently, no longer via Google but on ChatGPT and similar. So you need to track your “AI visibility”. Optimize your content for LLMs. And pay a hefty monthly subscription for it.