What you pay, and how we read it

List prices, what comes off them, and how they compare

What Search Watch costs

All figures exclude VAT. Two reductions never combine, and the larger one wins. Pay for a year and 2 months come off: the annual figure is the monthly one taken ten times, so twelve months cost ten. An audit is a one-time charge with nothing coming off it. Instead the amount you paid becomes a credit against your first program invoice when you start within 30 days (credit and net cost).

Work we run for you in the Korean market carries its own rate card on the Korean pricing page and bills on the invoice rail. We keep it off this page instead of restating it in a second currency we have not set.

Against published rates

We read the figure each company prints on its own public pricing page and noted the date we read it. We compared the number and nothing else: quality, outcome and contract terms sit outside this table. Both sides of every row price in dollars. Where we cost more, the row says what the money buys. The rows where we land level with the market face Korean vendors who publish in won, and those live on the Korean pricing page.

Why the expensive rows read that way. We measure one Search Watch unit like this: The same question goes to 8 AI answer surfaces, 3 times each. Asking once would hide how far the answer moves between runs, and that movement is part of what you pay for.

Forecast — how we will read the number

Forecast · how we judge a change. After the fixes we measure again the same way and tell you whether the change stands apart from luck. The method is one reading before the work, then the same questions at the same repeat count after it, and a Wilson 95% interval that has to stop overlapping before we call it a change. The first check lands four weeks after kickoff.

Forecast · the record so far. Measured across every account we have run, the average return came to 363% of the channel spend, read as net own-store revenue actually received divided by what went in. That is a record of accounts we have run, not a forecast we hand you. Where we run the work ourselves we set the target at 4.0 and the floor at 2.0, and the arithmetic behind those two numbers sits on the Korean pricing page.

Sources for every figure above are in Methodology and the public claim registry. Both paragraphs describe a way of judging, not a promise. The profit line stays blank until you hand us your cost of goods, because a profit figure built on a cost we guessed is a made-up number.