Automation has 2 tabs. A rule always does the same thing to the same review, every time it matches. An agent, on the next tab, reads a review and decides what to write. This page is about rules.
What you need: at least 1 connected source. Rules act on reviews as they arrive, so there is nothing for them to run against until a source is connected.
The Rules tab of Automation with a list of rules, each showing its trigger, actions and a live switch.
The 5 steps of a rule
Create rule opens a drawer laid out in the order you think in.
- When a review matches. 1 trigger per rule: a rating at or below a number of stars, a rating at or above, a rating with no words, a review on certain sites, a review that looks suspicious, private feedback, or a keyword.
- Then do all of this. As many actions as you need: reply, flag, assign, notify, send to a webhook and archive. They run top to bottom.
- Wait before the actions run: no delay, a fixed number of minutes, or a random wait up to a limit you set.
- Where it runs: every source, or only the ones you pick.
- Check what it would have done, which is the backtest over the last 30 days.
The rule editor drawer showing the When, Then, Wait, Where and Check steps.
One rule, several actions
Flag the review, assign it to the store manager and notify the team are 3 actions on one rule, not 3 rules. They run top to bottom in the order you set, and the wait applies to the whole set.
Scoping it down
Leave every source unticked and the rule runs on all of them, including ones you connect later. Tick a few and it follows only those. A trigger on sites narrows it by review site instead, which is the better tool when the same rule should cover every location's Google reviews.
The order rules run in
Rules run top to bottom, and every rule that matches takes its actions. What is limited is the reply. 1 reply goes out per review, so when 2 rules both answer the same review, the higher one posts and the second reply is dropped. Everything else those rules do, flagging, assigning, notifying, archiving, still happens on both. The rank badge on each row, such as "Replies 1st", tells you which rule wins.
That is why duplicating a rule leaves the copy paused: edit the wording first, then switch it on.
Rules and agents
Rules run first. A rule reply settles the review, and an agent does not answer a review a rule already replied to. A rule that only flags or assigns leaves the reply to the agent.
Replies on sites without a reply API
A rule's reply posts on a site that takes replies through its API. On a site that does not, the rule writes a draft instead, for a person to copy to the site. The editor says so beside each site: Replies post to the site. or Drafts only. You copy the reply to the site. Private feedback is never posted.
Live, paused and trial mode
The switch on each row turns a rule on and off without deleting anything. Pausing keeps the whole setup and stops it acting on new reviews.
Trial mode is the middle setting: the rule runs and records its matches, and nothing is replied to, assigned or sent. See Backtest a rule before it runs.
Copying rules onto another source
Set one source up properly, then mirror it. Copy rules from another source in the header asks which source to copy from and which to copy onto, and brings every rule scoped to it, including its actions and waits.
Read the warning before you confirm. Rules already on the target sources are removed first. Sources that need reconnecting are grayed out until you reconnect them.
An agency has 3 more buttons for its clients: Copy to clients, Update in clients and Switch in clients. Each works on rules with the same name in the client locations you pick.
What this costs
Matching costs nothing. Each action a rule takes counts as 1 credit, so a rule that flags and then assigns spends 2 every time it fires.
At zero, automation pauses. Rules carry on matching and record what they would have done, and nothing reaches a review site until your credit pool is topped up. Collecting reviews, the feed and answering by hand never stop.