Check schedules, costs, and allowances
Two schedules per brand, the price of an answer on each kind of engine, and the monthly allowance each plan includes.
On this page
Two kinds of answer, two prices#
| Engine | What it is | Cost per answer | Usual cadence |
|---|---|---|---|
| ChatGPT, Gemini, Google AI Mode, and AI Overviews (the apps) | What a person sees in the app or on Google, asked as a person would | about 0.1¢ | daily |
| GPT, Claude, Perplexity, and others through the labs' APIs | A live web search through the lab's API, what products built on the model get | 1¢ to 5¢, by model | monthly |
That difference is why a brand has two schedules under Tracking setup. One sets how often to ask the apps, and the other sets how often to ask the labs. Apps are cheap enough to ask daily on every paid plan. Labs run as often as your plan allows, and you can run a full check yourself at any time. The apps can be asked every day, every 3 days, every week, or only when you run a check.
Priority questions#
On plans that include them, a few questions can be asked of the labs every day while the rest wait for the scheduled check. Mark them in the question list.
When you start a paid plan#
Every brand you track is switched to what the plan includes at once. ChatGPT and the other apps your plan allows are asked every day, the labs on the plan's schedule, and a first check of the newly added apps starts straight away. If a brand's alerts go nowhere yet, they are sent to your account email. A brand that already has a schedule keeps it.
Before a check runs#
Press Run a check at the top of the dashboard. The confirmation shows how many answers will be collected, roughly what they will cost, and how much of your allowance is left. The estimate starts from the engines' list prices. Once you have enough history, it uses what your answers have actually cost. It also shows what your last check cost.
After a check runs#
The finished notice says what it cost. Answers → Checks run lists every check with its date, engines, answers, failures, cost per answer, total, and the basis. For app answers that is a flat rate per answer. For lab answers it is the token counts the labs reported, priced at their list price plus the web search fee. Where a provider reports the dollar amount itself, that figure is used. The provider's own bill is always the final word.
Allowances#
Each plan includes a monthly amount for visibility checks, separate from the AI included for assistants. A month of brand tracking never eats your chat budget, and chat never eats your tracking allowance. When it runs out, checks pause until it starts again on the 1st of the next month (UTC), unless you add credit or your own API key. You get an email when 80% of it is used, and again when it is all used. The free plan includes a one-time amount to start.
| Plan | Brands | Questions | Apps | Labs | Allowance |
|---|---|---|---|---|---|
| Free | 1 | 10 | 2 of 4 | 2 engines, when you ask | $1 once |
| Starter | 1 | 50 | 3 of 4, daily | 3 engines, monthly | $11 a month |
| Pro | 3 | 150 | 3 of 4, daily | 3 engines, monthly | $33 a month |
| Business | 5 | 350 | 3 of 4, daily | 3 engines monthly, and 5 priority questions daily | $110 a month |
| Agency | 30 | 1,000 | 3 of 4, daily | 3 engines monthly, and 5 priority questions daily | $255 a month |
A check that runs out of allowance stops where it is. The answers already collected stand, and the dashboard says what it would take to finish.
When something fails#
An engine that is overloaded or rate-limited is tried once more after a few seconds. An answer that still fails is kept as a failed row, so the count is honest, and it costs nothing. Twenty minutes after the check, every answer that failed because an engine was busy, slow, or could not be reached is asked once more, and an answer that comes back takes the failed one's place.
If an app stops answering altogether, the rest of its questions are not left to wait. US questions to ChatGPT and Gemini are asked another way, and the rest are asked again on the next check. A search that Google shows no AI Overview for is not a failure. It is marked "No AI Overview" and left out of the rates. If the server restarts during a check, the check picks up where it left off.
Updated Nov 10, 2026