Skip to content

Plans & Usage

Usage has exactly one unit and one formula. Both are on this page.

The unit

1 run = 1 prompt × 1 model × 1 country × 1 execution Every engine counts the same. There is no cost weighting, no multiplier for expensive models, no engine sold as an add-on.

That means “one run” is literally one AI answer, and stays directly comparable to how other tools quote responses or credits per month. Keeping it comparable is a deliberate choice: weighting an expensive engine would protect margin at the cost of making the unit meaningless.

The formula

Region-aware engines run once per market. Region-agnostic engines run once, in the primary market, regardless of how many markets you monitor.

runs = prompts × Σengine (region-aware ? countries : 1) × rounds
EngineExpands by country?Runs per prompt, per round
ChatGPTYes1 × countries
Google AI OverviewYes1 × countries
Google AI ModeYes1 × countries
GeminiYes1 × countries
PerplexityYes1 × countries
GrokYes1 × countries
DeepSeekYes1 × countries
ClaudeNo1, whatever the country count
DoubaoNo1, whatever the country count

The first seven browse the live web from an exit point in the target country, so monitoring three markets means running them three times. Claude and Doubao are called through their official APIs, which take no country parameter — running them repeatedly would return the same answer and consume allowance for nothing.

Worked example

10 prompts, ChatGPT + Gemini + Claude, one round:

CountriesNaive 3 × NActualDifference
13010 × (1+1+1) = 30
26010 × (2+2+1) = 50−10
39010 × (3+3+1) = 70−20

Adding a region-agnostic engine gets relatively cheaper the more markets you monitor.

Rounds

A round is one full pass over your selection. For budgeting, a monitoring cadence converts to:

CadenceRounds per month
Daily30
Every other day15
Weekly4.33

Where the count is shown

The run dialog displays the exact number of probes before you start. That same number drives the progress bar, the probe matrix on the panel, and the deduction from your allowance — one figure, computed once, so the page can’t advertise one thing and charge another.

Retries don’t cost extra. A probe that fails and is retried is still one run; a probe whose analysis fails permanently was still executed, so it is still one run — it is simply excluded from the averages afterwards. See averages & denominators.

Two separate ledgers

Runs and articles are independent monthly allowances. They never convert into each other in either direction: unused runs don’t become articles, and an empty article allowance doesn’t consume runs.

Distribution orders are separate again — they are per-order purchases, not drawn from either allowance.

What plans control

LimitWhat it governs
Runs per monthTotal probes across all targets
Articles per monthGenerated articles
BrandsHow many targets the account may track
Regions per brandHow many markets a single brand may be probed across
HistoryHow far back score history and past runs stay available
Citation depthHow many citation rows are surfaced per probe

Paid plans don’t cap how many prompts you may store — runs are the only gate. How you spend them is up to you: a small prompt set sampled frequently, or a large one run less often.

Allowances reset at the start of each billing cycle, and the Account → Billing page shows what you’ve used this cycle and when it resets. There is no automatic overage charge: when an allowance is exhausted, runs stop until you upgrade or the cycle resets.

See these numbers for your own brand The free plan probes every engine we support — no card required.

Start free