Standing vs doing nothing
The other 8 comparison pages on this site assume you have already decided to buy something. Most people have not, and the real alternative to us is usually a spreadsheet, an occasional manual check, or nothing at all. This page is about that.
We sell the thing this page is weighing, so read it accordingly. The test we set ourselves: the first section has to be a genuine list of reasons not to buy, written well enough that somebody acts on it and leaves. If it is not, the rest of the page is an advert with a fair-minded tone.
Stay where you are, if any of these is true
Not hedging. These are the four cases where we think buying anything in this category, from us or from anyone, is premature.
- Almost none of your buyers arrive through an assistant yet. Check your analytics for referrals from chatgpt.com, perplexity.ai and copilot.microsoft.com. If it is a handful a month, this is a thing to watch, not a thing to buy.
- You sell somewhere assistants are bad: heavily local, heavily relationship-led, or a category where the buyer already knows every vendor by name.
- Nothing about your site is fixed yet. If an engine cannot read your pages, measuring how often it names you tells you something you already know. Run the free page check and fix that first. It costs nothing and it is a real prerequisite.
- You would not act on the number. A measurement nobody is resourced to respond to is a subscription to a feeling.
The third one is the most common by a distance, and it is why the page check is free and uncapped rather than gated. It is the honest first step whether or not you ever pay us anything.
What doing nothing actually costs, when it costs something
Three specific things, rather than a fear.
You find out from a customer
The usual way this surfaces is somebody saying they asked ChatGPT for options and you were not one of them. By then it has been true for months, and you have no idea when it changed or what changed it.
You cannot tell a real move from noise
Ask an engine the same question twice and the answers differ. Checking manually once a month gives you one draw from a distribution, so a drop from 40% to 25% is indistinguishable from an ordinary Tuesday. This is the entire reason we sample 5 times per prompt per engine and publish a band.
The fixes are slow, so the delay compounds
Earning a mention on a source engines actually pull from, getting re-crawled, and showing up in retrieval takes weeks. Noticing late does not cost you the week you noticed, it costs you the quarter it takes to undo.
The middle option nobody sells you
Doing it by hand is a real choice, and for a small basket it is a reasonable one.
Open each engine, ask your five most important buying questions, note whether you were named, and put it in a spreadsheet. Repeat monthly. That is genuinely useful and it costs nothing but time, and anybody telling you it is worthless is selling something.
What it cannot do is tell you whether a change is real. One run per question is one draw, so month-to-month movement in a manual sheet is mostly noise, and the natural response to noise is either to chase it or to stop trusting the sheet. Both end the same way.
So the honest threshold is not a company size or a budget. It is the moment you would make a decision based on the number. Until then, keep the spreadsheet.
If you have read this far and none of the four applied
Then the question is not whether to measure, it is what you need from the measurement. If you have to defend the number to somebody else, the confidence band is the part that matters and it is the part almost nobody in this category ships. If you only need a rough direction, several cheaper tools will give you one, and our comparison pages say which and where they beat us.
The cheapest way to find out which you are
Run the free check on your own domain. If assistants already name you in your category, you have less to worry about than this page implies. If they name three competitors and not you, you now know that today rather than from a customer in March.