The dirty secret of most support macros: they were written once, get used 500 times, and sound like autoresponders because they are autoresponders. Customers can tell. This workflow builds a small library of 8-12 macros that solve 80% of your recurring tickets and — critically — tunes them weekly based on which ones are landing badly, so they stay human as your product changes.
Identify the top 10 recurring ticket types
Goal: Data-driven; not intuition. The macros should cover the actual top-of-mind support workload, not the exciting edge cases.
Here's a sample of the last 200 support tickets [paste subject lines + first-line]. Cluster them into recurring types. Return the top 10 clusters with a name for each, an estimated share of total ticket volume, and one representative subject line per cluster.
Draft each macro in the team's voice
Goal: 10 macros that read like a specific person wrote them for this specific customer, not like a form letter.
Draft a macro for each of the 10 recurring types above. Voice guide: warm but specific, no 'we understand your frustration', no 'we appreciate your patience', name the specific action we're taking rather than describing it abstractly. Each macro includes 2-3 [BRACKETED] fields the agent fills in per-ticket.
Score last week's macro usage
Goal: Which macros are landing (customer replies positively or issue resolves) vs failing (customer replies frustrated or opens a follow-up)?
Here are last week's tickets where a macro was used [paste them, tagged with the macro used and the outcome]. For each macro, calculate: (1) usage count, (2) landing rate (positive follow-up or resolution), (3) failure signals (frustrated reply, second ticket opened within 48h). Flag any macro with a landing rate below 70% for tuning.
Tune the flagged macros
Goal: Rewrite the failing ones with the actual failure pattern in mind.
The Cancel macro is landing at 58%. From the failure examples above, the pattern is: customers cancelling for a specific product reason don't want a smooth exit — they want the reason acknowledged. Rewrite the macro so the agent surfaces the reason (asks in the first paragraph, not the last). Keep the length short.
Handoff: New v2 macro replaces v1 in the tool. Track landing rate for the next two weeks; if v2 doesn't beat 75%, iterate again. If it does, that's the pattern.
Repeatability, not cleverness.
What lives where when the workflow finishes.
Macros live in your support tool (Front, Intercom, Zendesk, HelpScout, whatever). Weekly tuning happens at a fixed slot — Friday 4pm works well. Retire macros with usage below 5/week and no clear replacement need. Add new macros only when a genuinely recurring pattern emerges over 3+ weeks — no premature standardization.
The same shape, different jobs.
- Chat vs email variation — chat macros need to be 40% shorter; same steps, different draft target.
- Multilingual — Step 2 generates macros in a base language, then LADLE translates preserving voice guide.
- Escalation variation — a subset of macros are 'this needs engineering'; the workflow scores whether the escalation summary was actionable.
- Auto-tag variation — Step 1 becomes part of the ticket-processing flow, not a monthly cluster; each new ticket is auto-tagged with the closest recurring type.
Try this workflow once. Then decide if it earns a recurring slot.
The value compounds after the third run, not the first.