after we build · part seven · metrics and review
Helpdesk metrics as chain signals
The helpdesk generates its own signals. These are not secondary to the bug tracker — they are complementary, and over months they tell stories the bug tracker cannot.
First response time by level — L1 acknowledges within 1 hour? If not, operational capacity is the bottleneck, not product quality.
Resolution time by level — L2 investigations taking 3 days instead of 1? The product is becoming harder to debug — complexity is outpacing the team's understanding.
Ticket categories mapped to the bug taxonomy — a spike in "how do I" after a release means the feature is usable but not learnable. The chain was intact through code but broke at content design. A spike in "this doesn't work" means the feature has gaps the scenarios didn't catch.
Escalation rate — what percentage of L1 tickets escalate to L2? A rising rate means the CS handoff is becoming stale, or the product is producing situations the handoff never anticipated.
Satisfaction score — the most direct signal of whether the person's situation improved. Not a substitute for the prediction check, but a complement that arrives faster.
A product with low bug count but high "how do I" ticket volume is technically correct and experientially confusing. A product with high satisfaction but rising resolution time is loved but becoming harder to support.
The helpdesk data tells the team which part of the chain needs investment — often a part no one is watching.
SLA review cadence
SLAs are not set once and forgotten. They are reviewed quarterly with the client. The review covers:
- Which SLAs were met consistently — can the target be tightened?
- Which were approached or breached — what structural change prevents recurrence?
- Whether the categories still match the client's priorities.
A client whose usage pattern has changed — more users, different peak hours, new critical flows — may need SLA categories the original contract didn't anticipate.
The SLA review is a chain conversation: is what we're promising still aligned with what the person actually needs?