AI Visibility Monitoring Tools: US Pricing in October 2026, and Why the Unit Count Multiplies Before the Dollar Figure Does

In dollars, the headline plan price is not the comparable number. The comparable number is what you pay per tracked buyer question, per engine, per month, and in the United States that matters more than anywhere else because a national question set does not predict the city answer. Our city scan of 15 and 16 October 2026 proves the point in four metros: 43 of its 48 readings produced a named source list, no two cities had the same leading source, and in two of the four that leader was a national platform's own directory rather than a local firm at all.

If you sell in four metros, that is four question sets, not one. The unit count multiplies before you look at a single price.

Do the multiplication before the shopping

The arithmetic is simple and almost nobody does it before the demo. Take the questions your buyers actually type, times the engines you want read, times the refresh frequency, times the places you sell. Forty questions on five engines checked weekly is 800 recorded answers a month for one market. The same forty questions in four metros is 160 tracked questions, and 3,200 recorded answers a month.

Take that number into the pricing conversation, because vendors meter differently and both of the following can be honest. A tool charging per tracked question quotes one figure. A tool charging per recorded answer quotes a figure twenty times larger for identical work. They are selling different units, and until your own multiplication is written down you cannot line them up.

Line in a US quote	What to pin down
The meter	Tracked question, engine, recorded answer, or domain. Ask which, then recompute your total
Locations	Price the fourth metro, not the first, because per location is where these bills grow
Engines in the base tier	All five named individually, with the add on price for the ones that are not included
Refresh frequency	Weekly against monthly, which multiplies recorded answers and often the invoice
Seats	Included number and the cost of the next one, which matters more on a US marketing team than prompt volume
Raw answer export	Treat it as required, not optional, because without it you cannot audit the number you are reporting
Sales tax	Destination based and state specific for software, so the sticker is not the invoice

Sales tax is a state question, not a price question

Software subscriptions are taxed where the customer is, and whether a subscription is taxable at all differs from state to state. Two companies buying the identical plan on the same day can get different invoices for reasons that have nothing to do with the vendor. Ask the supplier how they treat your state and put the answer in the row next to the price. It is a short email and it removes the commonest surprise on a first invoice.

What the four measured cities showed, and why it changes the unit count

Six questions per city, both engines, 12 readings each, on 15 and 16 October 2026.

City	Readings that returned a named source list	Most cited source
New York	9 of 12	localseo.nyc, cited 3 times
Los Angeles	11 of 12	agencies.semrush.com, cited 4 times
Chicago	12 of 12	agencies.semrush.com, cited 3 times
Houston	11 of 12	houdigitalmarketing.com, cited 3 times

Two cities answered with a single city domain and two with a national software platform's agency directory. A tool configured to track one national question set reports one picture and misses both facts, which is the practical reason the per location line is the one to negotiate hardest.

The wider run behind those readings is larger. Across 1,260 readings on 16 October 2026 there were 20,143 citations across 4,947 domains, 82.0% of them on a vendor's or an agency's own website and 0.7% on review directories. The first run, on a different question set eleven days earlier in design, gave 85.6% and 0.8%. The question set and the full table are in our first benchmark, 597 readings across two engines.

The engines you are paying per unit for

Five surfaces are sold in this category: ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews. We have read two of them and published both runs, and the two did not agree with each other often enough to be treated as one line item on a quote. Paying for a second engine buys a second opinion rather than a bigger sample of the first, and that is the case to put to a finance team when the add on appears.

Our own American result, measured twice: zero citations of our domain in the United States on either engine on 15 October 2026, and 1 of 420 United States readings on 16 October 2026. Disclosure, because it changes how you should read the dollar arithmetic above: we sell one of the products this page is about, and America is the market where our own number is lowest.

Why there is no price table under this heading

We do not reprint another vendor's prices. On 17 September 2026 we watched an engine discount two comparison pages whose competitor prices were accurate, because they carried no source and no date, and a price copied today becomes a false statement about somebody else's product within weeks.

What is worth doing instead is building the row yourself, on the day you buy. Plan, amount, currency, term, included volume, engines, seats, and the date you read it. Mark anything that needs a sales call as quote only and leave the cell empty rather than filling it with an estimate, because an estimate in a price column is the one entry that will be quoted back at you. The full field list is in AI visibility tools compared, and per location pricing behaves here the way it does in what a local package costs per location and the line items it has to name.

The dollar table this page deliberately does not contain

We have not published a dated dollar price table across eight named vendors. That is a measurement we have not made, and the honest version of this page is the unit arithmetic plus the measured city evidence for why your unit count is larger than you think. The multiplication above is arithmetic on your own inputs, not a quote, and it is the number to walk into a demo holding.

What an American marketing team asks when it prices this work

What should we budget per month, in dollars?

Work out your own workload before you look at a single plan: tracked questions times engines times refresh frequency times locations. Then hand that workload to each vendor and ask them to price it. Two honest quotes for identical work can differ by a multiple, because the meters differ, and the published plan price tells you almost nothing on its own.

Which unit is the vendor actually billing?

Some meter the tracked question, some the engine, some the recorded answer and some the domain. Get the unit named, then recompute your whole workload in that unit. Until you do, two quotes are not in the same currency even when they are both in dollars.

Which tools cover all five engines?

Ask for ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews to be named individually at your exact tier, with the add on price for any that are extra. A blanket phrase about AI coverage is not an answer you can hold a vendor to.

Do we need a separate subscription for each city?

Price it as if you might. In the four American cities measured on 15 and 16 October 2026 the winning source differed in every one, and in Los Angeles and Chicago it was a national directory rather than a local firm, so a single national question set would not have surfaced either fact.

Is sales tax included in the quoted price?

Usually not, and whether software is taxable depends on your state. Ask the vendor how they treat your state and record it beside the price, because it is the most common difference between the quote and the first invoice.

Is a free scan enough to start?

For deciding whether you have a problem, a handful of questions read by hand will tell you. For reporting a number to somebody who will challenge it, you need the question set frozen, the engines named, the refresh fixed and the raw answers exportable, and that is the point at which a subscription starts earning its line in the budget.

Walk into the demo holding your own multiplication

Every dollar figure you are about to be shown is real, and almost none of them travels with the unit it was measured in. A clean list of plan prices, no units attached, nothing about your four metros, answers a question nobody in your company asked. So do the sum first: questions, then engines, then refresh, then locations. Then watch which quotes change shape when you hand that sum across the table. For the record, the tool we make, AI Knows Us, meters by tracked question and by engine and prints the denominator under every rate it publishes, which is a design choice a trial can verify rather than a result we are claiming. The sum is the one number in this purchase that nobody else can quote you.