01 — A simple test before building any automation
Automation feels powerful, and that is exactly the trap. The temptation is to automate everything from day one, and end up a few months later with a tangle of rules that contradict each other.
A useful question to ask before building any rule: has this exact thing been done by hand at least ten times in the last two weeks? If yes, automating it is worth the effort. If not, keep doing it manually a while longer until the pattern is clear.
Bad automation is worse than no automation at all — it routes work to the wrong place, sends the wrong reply, and erodes trust in the system itself. The rule that holds up over time: build slowly, delete decisively, and review every rule each quarter.
02 — Business hours: telling customers when the team is actually available
Business hours define when a given inbox is available, and three other things depend on it being set correctly.
- The out-of-hours message — customers writing in outside business hours see a message set in advance, instead of nothing at all
- The SLA clock — when an SLA is tied to business hours, the clock stops overnight and on weekends, so a conversation that comes in Friday evening is not already "late" by Monday morning
- Report accuracy — response-time numbers are measured against actual working hours, not the full 24-hour clock
03 — Auto-assignment: who gets the next conversation
By default, a new conversation in Chatwoot sits unassigned until someone picks it up. That works fine for a small team, and becomes painful once real volume arrives.
Auto-assignment offers two strategies. Round robin distributes new conversations evenly across everyone online and available, and is the right starting point for most teams. Balanced assignment looks at current workload first, and gives the next conversation to whoever has the fewest open conversations, which suits teams where some conversations take far longer than others.
The system skips anyone marked busy or offline, only considers agents actually added to that inbox, and skips anyone already at a configured capacity limit. Importantly, auto-assignment only handles the first handoff — a manual reassignment always overrides it whenever context calls for it.
04 — Automation rules: an if-X-then-Y engine
Where business hours and auto-assignment treat every conversation the same way, automation rules target conversations that match a specific condition. Every rule has three parts: an event that triggers it, conditions that must match, and actions that run once they do — for example, a rule that routes any conversation mentioning invoice or refund to the finance team and tags it billing, or one that flags a top-tier customer as high priority and routes it straight to a senior team.
The most common mistake is a rule that triggers itself — a rule that sends a message while listening for new messages. Chatwoot already prevents automation from re-running on an action triggered by another automation, but writing narrow, well-scoped conditions still matters. The other common mistake is a quiet rule nobody remembers exists, which a quarterly review of the automation list catches.
The difference from a macro is simple: a macro needs someone to click it, an automation rule runs on its own whenever the event fires. Whenever the same macro gets run manually on every conversation matching the same pattern, that is the signal to turn it into an automation rule instead.
05 — SLAs: a clock that times the promise made to a customer
An SLA is a policy stating that this kind of conversation must receive a reply within X and be resolved within Y. Chatwoot times three things automatically: time to first reply, time to each subsequent reply while the customer is waiting, and time to resolution.
An SLA is never applied to every conversation automatically, since that would just be noise. It gets attached through an automation rule, so it only applies to the group it is meant for.
The most important caution is not to promise a target that cannot be met. A five-minute first-response target across every inbox looks great on paper, right up until it is missed eight times out of ten. Pick a realistic target first, then tighten it as the team gets faster.
