Reading and answering reviews yourself never costs anything. A rule action costs 1 credit. An AI reply costs more, because the AI writes it, and what it costs depends on the kind of review it answers. The exact cost shows under every draft before it goes out.
Rules
Each action a rule takes is 1 credit every time it happens, counted per action rather than per rule:
- Reply to the review, Flag for your team, Assign to a team member, Notify the team, Send to a webhook and Archive are 1 each;
- a rule that flags and then assigns spends 2;
- Send to a webhook is charged only when your address answers with success.
AI replies
An AI reply is priced by what it takes to write, so its cost depends on the reply. A short thank-you to a 4- or 5-star review costs less than a careful answer to a low rating, a reply written with extra instructions, or a reply to private feedback.
A low rating costs more because the reply has to do more: acknowledge the problem, keep the tone right and offer a next step. The cost shows under every draft, and the meter on your billing screen shows what was spent.
A reply costs the same whether the agent posts it, sends it for approval or leaves it as a draft. A draft you discard is not refunded, because it was written.
Translation
- Translating a review or a reply: 1 credit each time something is actually translated.
- Automatic translation of incoming reviews: 1 credit per review it translates.
- Rewriting an automated reply into the reviewer's language: 1 credit, on top of the reply.
The credit meter with the line underneath saying what counted this cycle.
What is free
- Reviews arriving, from every source, at any volume. A review is never charged for existing.
- Reading, searching, filtering and exporting.
- Everything a person does by hand: replying, copying a reply and marking it replied, assigning, archiving, approving and discarding.
- Connecting, reconnecting and disconnecting a source.
- A rule deciding to do nothing. Matching is free; only the action is counted.
- Trial mode, the backtest, and an agent leaving a review alone because of a skip keyword or a guardrail.
Where to see what was spent
The meter on your billing screen carries the total for this cycle, what is left and the reset date, and the line under it names what spent the most. Insights breaks it down by rule and by agent.
Keeping the number down
- Let rules handle the fixed answers. A rule thank-you is 1 credit; an agent's is more.
- Narrow the scope. An agent pointed at the sites and locations that need it spends less than one covering everything.
- Backtest a rule before you turn it on. A backtest tells you how often it would have fired, which is how often it will spend.
- Leave the webhook rule out if nobody reads the data it sends.
None of this touches the feed. Turn every automated thing off and the reviews still arrive, still land in one feed, and can still be answered by hand for nothing.