Setting B2B support SLAs that survive production
Too ambitious an SLA pushes agents to close fast and badly. Too lax and the contract is empty. Here's how to calibrate between the two.
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.
Distinguish FRT, FFR and TTR
- 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.
Calibrate on p90, not the average
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.
Rule of thumb
For a standard B2B SaaS support setup, defensible p90 SLAs: FRT 15 min business, FFR 2 h business, TTR 1 business day on blocking incidents. Anything tighter costs you a dedicated agent per 30 accounts.Spell out exclusions
- Tickets pending customer information (clock pause).
- Declared third-party incidents (documented provider status).
- Out-of-scope requests (training, custom integration).
Report monthly, not quarterly
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.