A customer opens ChatGPT and types "what is the best CRM for a small agency." In about three seconds they get a shortlist of four tools with a reason for each. If your product is on that list, you just earned a warm lead who did no comparison shopping. If it is not, you were cut before the race started, and you will never see it in your analytics.
The good news is that getting recommended is not luck. ChatGPT is not choosing at random. Once you understand what it is drawing on, you can influence the outcome with steady, unglamorous work.
ChatGPT builds its answer from two things: what it learned during training, and, when browsing is on, what it can read on the live web. In both cases it is looking for a confident, well supported picture of who the strong players are in a category. It leans toward brands that are described the same way in many places, mentioned next to the right problem, and covered by sources it treats as credible.
So the real question is not "how do I trick the model" but "does the web give the model a clear reason to name me." Usually the gap is that it does not, yet.
Before changing anything, ask ChatGPT the exact questions your buyers would ask. Not "is my company good" but the natural, category level questions like "best project management tool for construction teams." Read the answers. Are you there? In what position? Described accurately? Who shows up instead of you? Write it down. This is your baseline, and everything you do later is measured against it.
Models repeat the framing they see most often. If your homepage is full of slogans and light on plain facts, the model has little to work with. Make sure the basics are stated clearly somewhere it can read them: what you sell, who it is for, what makes it different, and the category words a customer would actually use. Marketing poetry does not travel well into an AI answer. Concrete, boring clarity does.
If a stranger read one page of your site, could they explain in one sentence what you do and who you are for? If not, the model cannot either.
Your own site is only part of the picture. Assistants lean heavily on third party sources: review platforms, comparison articles, industry roundups, forum threads, and reputable publications. Being present and accurately described in those places carries real weight. A single "best tools for X" article that includes you can do more than a month of homepage edits.
When you read AI answers about your category, you often learn what customers think the alternatives are, and what the model thinks you lack. If three competitors get praised for a feature you also have, that is a description problem, not a product problem. Fix the story.
AI answers drift as models update and the web changes. A brand that was mentioned in March can quietly vanish by June. Treat this like a health check you run every few weeks, not a one time project. Track the same questions over time so you can see whether your work is paying off.
None of these steps are dramatic on their own. Together, over a couple of months, they are usually enough to move a brand from "never mentioned" to "on the shortlist." That is the whole game.
Get a free report showing your position and the competitors that appear instead.
Check my AI visibility