A method for finding out which brands assistants name in your place, why that list rarely matches the competitor set you track, and what to do about it.
Most companies maintain a competitor list built from sales calls and market knowledge. The brands an assistant names when asked for options in your category are drawn from something else entirely — whatever the web says about the category — and the two lists routinely disagree.
That disagreement is useful information. A brand you have never lost a deal to, appearing consistently in generated answers, is taking attention at the stage before anyone contacts you.
Never include your own brand name. Use category discovery (“best tools for X”), constrained recommendation (“what should a mid-sized company use for X”), and comparative phrasings (“alternatives to [known competitor]”). Ten to twenty prompts, held fixed.
This is where most teams lose the value. Logging only whether you appeared throws away the comparative picture, which is the part that tells you what to do.
A competitor appearing once is noise. A competitor appearing in most runs of the same prompt is a position. Run the set several times across different days before you treat any of it as real.
The useful number is what fraction of answers name you, measured against what fraction name each competitor. Appearing in forty percent of answers sounds poor until you see the leader at thirty.
Being named first is not the same as being named fourth. The first brand tends to anchor the comparison that follows, so record order rather than treating all mentions as equal.
Look for the brand that appears specifically in the answers where you do not. That is your real substitute in AI discovery, and it is often not the competitor your sales team names.
When a brand you do not recognise recurs, resist dismissing it. Assistants surface brands that are strong in the sources they read, and those sources are shaping your category whether or not the brand is commercially serious.