Most support SLAs we see in enterprise contracts are written by sales to reassure the prospect, with no operations review. The outcome is predictable: systemic breach by Q1, or unsustainable pressure on the team.
FRT, FFR, TTR: what's the difference?
- First Response Time (FRT): time to the first human response (or AI if it's qualitative).
- First Full Response (FFR): time to a response that delivers actual progress, not an acknowledgement.
- Time to Resolution (TTR): time to a closure confirmed by the customer.
Committing on FRT alone is a trap: teams optimise for acknowledgment and the customer feels the opposite of what the SLA promised.
Should you calibrate an SLA on the average or the p90?
The average is dominated by trivial conversations. The customer lives their tail. Measure your current p90 and commit to p90 + 10% at most, that's what holds up in a monthly review.
Which exclusions should a support SLA spell out?
- Tickets pending customer information (clock pause).
- Declared third-party incidents (documented provider status).
- Out-of-scope requests (training, custom integration).
How often should you report SLA attainment?
An SLA you discover in breach at the quarter is an SLA you're not steering. Monthly cadence, with per-account drill-down, is the minimum to react before the commercial relationship erodes. Between two reports, the SLA can be watched live: what Intercom's real-time dashboard shows, and what it misses.