How AI decides the shortlist
When you ask an assistant for the best tool in a category, it doesn’t search. It remembers. The answer is a shortlist drawn from everything the model read while training: fast, opinionated, and revealing.
What “best” means to AI
A model answers from memory, not a live search. So “best” really means most-represented in its training: the tools the web describes most clearly and consistently. That correlates with quality, but even more with how much clear, structured content exists about a product. We ask with web search off on purpose, to isolate that trained judgment.
The snub pattern
On 81 shortlists, AI named 52 well-known tools that it then left off the ranking. These are products it clearly knows exist but didn’t reach for. The snubbed brands tend to have thinner, less-structured content about themselves than the tools above them. The product can be excellent; the representation isn’t. Being left off is a visibility problem, and visibility is fixable.
Frequently asked
Why does AI leave well-known tools off the shortlist?
A model answers from what it has read, so a tool is easy to recommend when many sources describe it clearly. The brands AI omits tend to share one trait: thin, unstructured, hard-to-read content. The omission is usually a visibility gap, not a verdict on quality.
Is the ranking sponsored?
No. No brand can pay to appear, move up, or be removed. Each shortlist is one ChatGPT answer published in full.
Does AI browse the web to build these?
No. We ask with web search off so the answer reflects the model’s trained judgment, the default shortlist it reaches for.
How current is a shortlist?
Each ranking reflects the model’s training as of the date shown on the page. Pricing and packaging move faster than training does, so treat any specific number as a strong first opinion to verify, not a live quote.
If a brand you work on is getting left off, it’s almost always a content and visibility gap, not a quality one.
Check your visibility →