Imagine opening a support queue on Friday afternoon. The replies look cleaner. Drafting takes less time. A request from Tuesday is still sitting near the bottom, waiting for someone who can approve an exception.
That is the request we would follow. A faster draft can be useful, but a customer is waiting for the job to finish. We want the measurement to keep that unfinished job visible.
Take a fictional one-day rehearsal. There are twenty unresolved requests at opening. Thirty new requests arrive. Twenty-five are resolved before closing. Five of those resolutions clear requests from the opening queue; twenty clear new arrivals. Nothing is merged, reopened or transferred out.
The closing queue contains twenty-five requests: fifteen carried over from opening and ten new ones. The count checks both ways. Twenty plus thirty minus twenty-five equals twenty-five. Fifteen plus ten gives the same total.
Now add an encouraging result: average hands-on work per resolved request fell from twelve minutes in a separate baseline sample to eight minutes in this rehearsal. That measures work on resolved requests. It does not tell you how long the remaining twenty-five have waited, or whether those two samples contained equally difficult jobs.
The queue can grow while handling gets faster. In this invented day, arrivals exceed resolutions by five. That arithmetic does not explain why the Tuesday request remains open. Its next decision belongs in the record.
For the next rehearsal, keep a small closing note:
Unresolved at opening:
New arrivals:
Resolved:
Reopened, merged or transferred, if any:
Unresolved at closing:
Oldest unresolved request and original arrival time:
Next decision, owner and review time for that request:
Preserve the original arrival time when a request changes hands. Keep internal identifiers in the working record. A reassignment should not make an old request look newly arrived.
Then inspect the oldest item. Does it need a missing fact, permission or a decision that no draft can supply? Give that next step an owner and a review time. A new reply is useful only if it advances the job.
Our Shopify essay follows that handoff beyond classification: Who finishes the job?
At your next queue review, reconcile the closing count and inspect the oldest unresolved request beside it. Review what is actually preventing that one request from moving.
The Rebuild Editorial


