None of these need a particular tool. All of them are easier with one that was built with them in mind.
1. Every conversation has exactly one owner
Shared responsibility is no responsibility. Assign every ticket to one person, even if a team works on it. Watchers and collaborators can follow along; one name owns the next reply.
2. Statuses say whose turn it is - honestly
Keep the status list short and meaningful: waiting on us, waiting on the customer, done. Do not create statuses that exist to stop the clock. If waiting on another team pauses your SLA, every difficult ticket will end up waiting on another team.
3. Measure on business hours
A promise of "4 hours" means business hours to your team and wall-clock hours to your customer unless you say otherwise. Define the calendar, include holidays, and publish the target in the terms customers will judge you by.
4. Never move a promise after it is made
Changing an SLA policy should affect new tickets, not deadlines already given. If a deadline must be reset, make it a deliberate, recorded act with a reason - not a side effect of editing a setting.
5. Route before anyone reads
Triage meetings do not scale. Route by mailbox, category, customer and priority automatically, and use round-robin inside each team. Keep a short unassigned view for what the rules could not place.
6. Write the twenty common answers once
Saved replies with placeholders keep wording consistent and save minutes per ticket. Review them quarterly: the refund policy changed, the screenshots are old, the product was renamed.
7. Put a second pair of eyes on the replies that can hurt
Refunds, compensation, legal wording, anything to a VIP, anything a new agent writes in their first weeks. Do not ask people to remember to forward these to a manager - make the rule hold the reply until it is approved, and make the approval apply to the exact text that is sent.
8. Keep internal notes truly internal
The most damaging support mistake is a customer reading what agents wrote about them. Use a tool that stores notes separately from the conversation, so they cannot be sent or shown by accident.
9. Treat a silent mailbox as an emergency
When a mailbox connection breaks, nothing errors - mail just stops. Monitor when each mailbox last received mail and alert when it goes quiet during business hours.
10. Know when a reply bounced
"We replied on Tuesday" means nothing if the reply bounced. Track delivery, surface bounces to the author, and stop mailing addresses that complained.
11. Read four reports every week
Backlog by age, first-response attainment, workload by agent, and unassigned count. Add approval turnaround if you use approvals, and CSAT once the basics are stable.
- Backlog by age - the six-week-old tickets are the ones that matter.
- First-response attainment on business hours.
- Open and resolved per agent, with unassigned counted separately.
- Median approval time, not mean.
12. Ask customers, but not too often
A one-question survey after resolution, at most once a week per person, tells you more than any internal metric. Attribute the rating to whoever resolved the ticket at the time, not whoever owns it later.