Colay / Guides
Turn a complex ticket into a reference answer your team can learn from
A difficult ticket can become a useful teaching example once the response is reviewed and its reasoning is visible. Colay helps compare possible answers and expose missing questions. The final artifact should include the customer reply, annotations explaining its choices, and conditions for reuse. A fluent draft becomes a reference answer through review, not through a shortcut or a vote among models.
Reconstruct the case before drafting its answer
Write a short case record with the customer's goal, confirmed symptoms, checks already completed, previous commitments and remaining uncertainty. Distinguish what the customer reported from what the team verified. A complex ticket often becomes harder when those categories are collapsed into a confident diagnosis.
Use a de-identified version you are permitted to share. Include the relevant approved procedure and escalation rule rather than a complete private conversation. On macOS, ⌘⇧K opens Colay's overlay by default and can be changed in Settings. It brings up the drafting workspace; you still provide context and review the result.
A fictional unresolved ticket with a useful response
Imagine a fictional reporting service. A large export fails, a small export succeeds, and the customer has already repeated the large export twice. The cause has not been established. The assigned specialist has accepted the case, and the team has explicitly committed to an update at 16:00 UTC. These are invented inputs, including the time; they are not a service promise from Colay.
An appropriate example reply is: Thank you for confirming that the smaller export works. I understand that the larger report is still blocked after two retries. Our specialist is investigating the case. We will update you at 16:00 UTC with what we have found, even if the cause is not yet confirmed. You do not need to repeat the same export while we review the information already supplied.
| Sentence choice | Reason | Condition for reuse |
|---|---|---|
| Acknowledges the smaller export | Shows the supplied evidence was read | That observation is present in the actual ticket |
| Names the unresolved blocker | Avoids declaring the incident solved | The large export is still failing |
| Promises an update, not a fix | Preserves the approved commitment | An owner has actually agreed the update time |
| Avoids another identical retry | Respects checks already completed | The approved investigation does not require a new attempt |
Ask for contrasting drafts and an annotated specimen
Use Ask separately for initial answers from selected available agents. Compare how they handle uncertainty, repeated questions and the next action. A Consensus synthesis may help organize the alternatives, but keep the source facts available so the final text can be checked sentence by sentence.
Create a reviewed-answer candidate from this de-identified support case: [goal, confirmed facts, reported facts, completed checks, prior commitments, unknowns]. Approved procedure and escalation rule: [source]. First list missing facts that could change the response. Then draft a concise reply and a more explanatory reply without inventing a cause or resolution. For each meaningful sentence, explain which case fact supports it. Return a recommended customer reply, internal reviewer notes, and explicit conditions under which another agent may reuse it. Keep internal notes out of the customer text and mark any commitment that needs owner confirmation.
Approve the specimen on facts and service behavior
Zendesk’s instructions for creating a macro from an existing ticket recommend adjusting the comment for reuse. Apply that idea explicitly to a reference answer: show which details belong to the original case and which reasoning transfers. An unresolved case can teach a useful interim response without being presented as a solved issue.
- Have the responsible support reviewer verify every factual statement and promise against the case and policy.
- Check whether the answer addresses the customer's actual goal. A long technical explanation can still fail to say what happens next.
- Confirm that internal assumptions, personal data and unsupported diagnoses have not entered the customer-facing text.
- Store the approved response together with the anonymized case, annotations, owner, version and conditions for reuse in your existing team library.
Teach adaptation rather than copying a perfect paragraph
Ask a teammate to adapt the specimen to a nearby case: the specialist has not accepted it yet, the previous checks differ, or no update time is approved. They should identify which sentences must change. If they cannot, the reference needs clearer conditions rather than more polished prose.
Keep the example's status honest. It may be an approved interim response while the underlying ticket remains open. When the investigation changes the facts, update the specimen and its annotations. The reusable value lies in the reasoning that transfers to another case, not in making every customer receive the same confident answer.
Questions, answered
Can a reference answer be useful before the issue is solved?
Yes. It can show how to acknowledge evidence, communicate uncertainty and set an approved next step. Label it as an interim response rather than a resolution.
Does agreement between models make it an approved answer?
No. The responsible reviewer checks the actual case, policy and commitments. Model comparison helps draft and inspect the text; it does not grant authority.
How is this different from a normal support macro?
A macro is designed for repeated application. An annotated reference example teaches the reasoning and limits of a particular response; a separate review decides whether part of it should become a macro.
Sources and methodology
- Zendesk: creating macros from existing tickets
Primary guidance on adapting an existing ticket comment for reuse. The annotated teaching example here is original and does not imply a Colay integration.
Bring your next question to Colay
Choose a model, use Auto, or bring several perspectives together with Consensus.