A general rule is an event-to-action link that runs outside any flow. When something about a contact changes, the rule fires and can start an automation.
A general rule is a link you build in the form "when this happens, do that", and it runs any flow. When a tag is added to a contact, a field changes, or a new contact is created, the rule fires by itself: it applies its actions and, if you like, starts an automation.
outside
💡 Rules live on the Automations › Rules tab and are kept separately for each account.
What a general rule is for
Flows start with something the customer does: writing a message, leaving a comment, clicking an ad. But not every change comes from the customer. An agent may add a tag on a contact card, you may tag hundreds of people in a bulk operation, an external request step may fill in a field. Rules are what catch exactly those changes.
Typical uses:
Enrolling a contact in a thank-you sequence when an agent adds the purchased tag.
Adding a new tag to every newly created contact and starting the welcome flow.
Closing the conversation and notifying the team on a Conversion recorded event.
Clearing a tag when someone leaves a sequence.
What a rule is made of
A rule has four parts:
Rule name — The name shown in the list (up to 80 characters).
Event — The change that starts the rule.
Filter — The value that narrows the event: which tag, which field, which conversion name, which sequence. Leave it blank and every event of that kind matches; on screen this option reads All.
What to do — The actions to apply (up to 10) and, optionally, an automation chosen under Also start.
A rule also has an on/off status, and the list shows how many times it has run.
The automation you pick under Also start must be live; a draft or switched-off automation is never started.
Event types
Tag added — When a tag is added to a contact. Type a tag in the filter to limit it to that tag.
Tag removed — When a tag is taken off a contact.
Field changed — When the value of a field on the contact card changes. You type the field name in the filter.
New contact — When a new contact is created in the account.
Conversion recorded — When a Conversion step in a flow records an outcome. You type the conversion's name in the filter.
Entered sequence — When a contact is added to a sequence.
Left sequence — When a contact leaves a sequence or finishes it.
An event does not care where the change came from: whether an agent added the tag by hand, an action in a flow added it, or a bulk operation added it, the rule fires the same way.
Which actions a rule can use
Rules can use almost all of the actions available in a flow's Action step: adding and removing tags, setting a field, increasing or decreasing a field, clearing a field, assigning the conversation, setting the conversation status, handing over to a human, notifying the team, and adding to or removing from a sequence.
Only two are missing: Hide comment and Delete comment. A rule does not run inside a flow, so it holds no comment context and those actions would have no comment to act on.
Loop protection
A rule's action produces new events: adding a tag raises a "tag added" event, which may trigger another rule. So that two rules cannot trigger each other forever, the chain has a depth limit and never goes beyond it.
In run logs this appears as Loop protection stopped this. If you see such an entry, two of your rules are feeding each other and you need to break the chain by narrowing the filters.
⚠️ Note: the most common loop is a pair of rules sitting side by side — "when tag A is added, add tag B" and "when tag B is added, add tag A". Forgetting to remove the old state tag produces similar chains.
Seeing that a rule ran
There are three places to confirm whether a rule fired:
In the rules list, each rule shows how many times it has run. If the number is not growing, either the event never happens or your filter does not match.
On the contact card you see what the actions did: whether the tag was added, the field changed, the conversation assigned.
In the automation's Logs tab the run the rule started appears, with its source shown as General rule. That is how you tell whether a run came from the person's message or from a rule.
If a rule is not working, check these in order: is the rule switched on, is the tag or field name in the filter exactly right, and is the automation chosen under Also start live?
Rule, or flow?
Both do automatic work, but they are started by different things:
A flow starts with something the customer does (a message, a comment, a story, an ad), or with the calendar (a scheduled trigger).
A rule starts with a change in the contact's data (a tag, a field, a conversion, a sequence).
If you need a messaging arrangement, write a flow. If you are quietly tidying data, or starting a flow from a different context, write a rule. Using both together is common: the rule applies the actions and hands the actual conversation to a flow with Also start.
A concrete example
A boutique organises its post-purchase experience like this:
A sales agent adds the purchased tag on the contact card of the customer whose order came in.
The rule bound to the Tag added event fires. Filter: purchased.
The rule's actions: Set field updates last_order_date, Notify the team tells the warehouse, and Add to sequence enrols the person in the "After delivery" sequence.
Also start launches the "Thanks and review" automation, which sends the customer a thank-you message.
When the customer finishes the sequence, the Left sequence event triggers a second rule that replaces the purchased tag with completed.
Common mistakes
Leaving the filter blank. A "Tag added" rule with no filter runs for whichever tag is added. That is usually not what you wanted.
Choosing a draft automation.Also start only launches a live automation.
Treating a rule like a flow. Rules have no conditions, buttons or waits; those belong in the flow they start.
Writing two rules that feed each other. Loop protection breaks the chain, but your rules will not do what you intended.
Looking for comment actions in a rule. They are not there, and cannot be.
Testing without switching the rule off. While a rule is on it runs on real people; keeping your tests behind a narrow tag is the safe way.