the assistant has never heard of you. find out what it says instead.
Being unknown is not the same problem as being ranked badly, and it does not have the same fix. The first useful thing is knowing exactly who the assistants name in your category today.
what your buyer actually types
These are the shape of the questions we ask on your behalf. You never write them: a set is drawn from a shared library for your category, and on the higher plans a set is written specifically for you.
What are the best tools for a team that just raised a seed round?
Who are the up-and-coming companies in this space?
What should I use instead of the expensive enterprise option?
Which new startups are worth watching in developer tooling?
Is there a cheaper alternative to the market leader?
Each question is asked repeatedly rather than once. Two identical runs minutes apart agree on only 36% of the brands they name, so a single answer is one draw and not a measurement.
what is actually at stake
Absence is the normal state, not the finding
An assistant not naming a company nobody has written about is expected, and treating that as a failure wastes the measurement. What is worth knowing is the shape of the answer without you in it: who is named, in what order, and on the strength of which pages. That is the map you are trying to get onto.
The incumbent list is your positioning brief
Ask the buying question and the answer names three or four companies with a reason attached to each. Those reasons are what a buyer currently believes the choices are. It is unusually direct competitive research, and it is the same text your own positioning has to argue with.
A baseline now is worth more than a report later
The value of measuring early is not the first number, it is having a first number at all. Movement only becomes readable against a starting point, and you cannot go back and measure what an assistant said six months ago.
the things people ask us before they start
- Is this useful if nobody has heard of my startup yet?
- Yes, but for a different reason than for an established company. You are not measuring your own visibility, which will be close to zero and unsurprising. You are measuring what the default answer in your category is: who gets named, in what order, and which pages the assistant read to decide. That is competitive research you can act on, and it doubles as the baseline you will measure progress against later.
- Should I pay for this before we have customers?
- Probably not more than the cheapest plan, and possibly nothing at all. The free scan covers five questions on ChatGPT with no card and no account, and for many early companies that is the whole answer for now. Local at 19 USD a month is the first step worth taking, and it exists so that measuring is not a decision you have to justify.
- How will I know if anything we do is working?
- By comparing runs, not by reading one. Two identical runs of the same question minutes apart agree on only 36% of the brands they name, so a single answer moving means nothing on its own. Every figure carries a confidence interval, and when two runs are too close to separate the report says so instead of drawing you a line.
- What do we do with the answer if we are not in it?
- Read what was cited. Every answer we store lists the exact pages the assistant used, and for an early company those are almost always somebody else’s: a review roundup, a comparison post, a forum thread. That list is a concrete brief for where a mention would have to exist. It is not a guarantee, and the report does not pretend otherwise.
The scan is free and needs no account. Most companies like yours start on Local, at 19 USD a month, and the per-call provider costs behind every price are published.
This will very likely tell you that you are not named, and it cannot tell you how to become named. It maps each finding to a probable cause, but for an early company the honest cause is usually that not enough has been written about you yet, and no tool fixes that.