clinic · an sla review that became sales

An SLA review that became a sales conversation

The SLA review was scheduled. Forty-five minutes on the calendar. Halfway through, the client asked about adding a new module. The PO answered. Sixty minutes later, nobody had reviewed the SLA. Three weeks after that, an SLA threshold was breached without warning — and nobody had read the trend, because nobody had run the review.

The artefact

Excerpt — Meeting notes, "2026-Q1 SLA Review × Acme University", April 2026

Attendees: Alex (PO), Dina (CS Lead), Acme rep.

Notes:

• Quick uptime review — all green for the quarter.
• Client asked about adding a second campus to the contract. We discussed the rough scope: 80 additional graders, possibly a new department.
• Tech Lead would need to check feasibility on the multi-tenant assumption — we'll follow up.
• Pricing discussion: client noted current contract was structured per-instance; multi-campus would need to revisit.
• Maintenance window — discussed shifting from Sunday 06:00 to Saturday 22:00 UTC. Tabled for review next quarter.
• Action items:
  • Alex: get TL feasibility on multi-campus by 2026-04-30.
  • Dina: send pricing options.
  • Acme: confirm timeline for second campus rollout.

The meeting ran 60 minutes. It produced three action items. Everyone left feeling productive. The relationship felt strong.

Three weeks later:

• Response time P2 quietly trended toward its 1h ceiling for two consecutive weeks. The 2026-Q1 data had already shown this approach — but nobody read it during the meeting.
• The P2 threshold was breached at 1h 04m on a Wednesday morning. The client noticed first.
• The maintenance window shift that was "tabled" remained ambiguous — graders showed up Sunday morning to a partially-deployed system.
• The multi-campus feasibility was still being chased six weeks later; meanwhile, the current contract wasn't being honoured.

The SLA review didn't happen. The conversation about the SLA happened next to a sales conversation, and the sales conversation won.

What's wrong?

Stop. Find three things wrong before reading the diagnosis.

Diagnosis (open when ready)

1. The numbers weren't read

Quick uptime review — all green for the quarter.

One line. All green. No dimensions. No approached-without-breaching list. No numbers. The corpus rule from Practice · SLA review · Step 1: the PO and CS Lead read the quarter's numbers before the meeting and walk each dimension at the table.

In the actual Q1 data, the team would have seen: Response time P2 trended toward 1h ceiling in March (week-over-week +12%). Cause: one CS team member departed in early March; cover not yet fully in place. That conversation would have led to adjust the threshold conditionally or invest in CS coverage before Q2. Instead, the team saw a green check mark and moved on.

The breach three weeks later was not a surprise to the data. It was a surprise to the people who didn't read it.

2. The sales conversation pushed the SLA conversation out

Client asked about adding a second campus. We discussed the rough scope: 80 additional graders, possibly a new department.Tech Lead would need to check feasibility on the multi-tenant assumption.Pricing discussion: client noted current contract was structured per-instance.

Three sales topics, none of them on the SLA review's agenda. Each one took meeting time that should have gone to the three questions: did we meet · where did we approach · are the categories still right.

The corpus's discipline from Practice · Step 3:

Client: "Since we're talking about SLA, we'd really like to discuss adding a third environment."PO: "That's a roadmap conversation; let's hold it for the bi-weekly sync. For the SLA review, we have ten more minutes — I want to walk through whether the P2 categories still match how you experience severity."

The PO refused the off-agenda topic. The review held. The sales conversation was rescheduled to a forum where the right people (account lead, leadership) could engage it.

In the actual meeting, the PO answered the multi-campus question on the fly. The PO is not the account owner; the answer was incomplete; the client now has a partial expectation that the team will not fully meet. Worse: the SLA review didn't happen, so the contract that currently exists wasn't reviewed.

3. The third question was tabled

Maintenance window — discussed shifting from Sunday 06:00 to Saturday 22:00 UTC. Tabled for review next quarter.

The third SLA-review question — are the categories still right? — was the only SLA-shaped topic that actually came up. And it was deferred.

Tabling the third question is a corpus-specific failure. The categories question is the deepest question in the SLA review and the one most likely to surface drift. Tabling it for the next quarter means the team will spend another quarter operating against a maintenance-window definition that the client has already signalled may not match their operations.

A corpus-compliant outcome would have been categories rewritten — schedule follow-up to draft new maintenance window by 2026-05-15. A real action with a date. Not tabled.

The fix

The same hour with the corpus's discipline applied:

# SLA review — 2026-Q1 × Acme University
Date:     2026-04-08
Attendees: Alex (PO), Dina (CS Lead), Acme rep
Duration: 45 minutes

## Numbers read before the meeting
Availability (target 99.9%):     99.97% met cleanly
Response time P1 (target <15m):  12m median  met
Resolution time P1 (target 4h):  3h 30m  met
Response time P2 (target 1h):    58m median  met (close)
                                  Approached: weeks 10-13, trended
                                  toward ceiling (week-over-week
                                  +12%). Cause: CS departure.
Resolution time P2 (target 1d):  20h median  met
Data integrity:                  0 incidents  met
Breaches:                        None.

## Q1 — Did we meet?
Yes, all dimensions. No breaches.

## Q2 — Where did we approach without breaching?
Response time P2 in weeks 10-13. Specific cause: CS team
departure in early March; cover plan in flight but not yet
back to full headcount. Mitigation: temporary contract
support starts 2026-04-15; expect P2 response back to
~30m median by week 4 of Q2.

## Q3 — Are the categories still right?
Client raised: Sunday 06:00 UTC maintenance window conflicts
with end-of-week grading at flagship campus.
Proposed: shift to Saturday 22:00 UTC.
Decision: draft new window definition; signed at next
bi-weekly sync (2026-04-22).

## Outcome
Status:     Contract holds — with one adjustment.
Adjustment: Maintenance window shifting from Sunday 06:00 UTC
            to Saturday 22:00 UTC. New definition signed at
            2026-04-22 sync; effective 2026-05-01.
Watch:      Response time P2 — Alex (PO) and Dina (CS Lead)
            report at 2026-04-29 sync.
Signed:     Alex (PO), Dina (CS Lead), Acme representative.

## Off-agenda topics (deferred, not lost)
Multi-campus expansion (sales conversation): scheduled with
account lead Sarah for 2026-04-15. Not an SLA conversation.

The review took 45 minutes. The sales conversation moved to where it belonged. The maintenance window was decided with an explicit follow-up. The P2 approach was named and watched.

Three weeks later, the team caught the P2 trend before it breached. The breach that happened in the original timeline didn't happen this time.

Where this comes from in the chain

This failure traces to Operation (Level 5) — the SLA review is the operational practice that keeps the contract honest. When the practice drifts into sales, every other level pays interest:

• L1 (Strategy) — the multi-campus question was answered without the strategic owner present
• L4 (Execution) — the maintenance window stayed ambiguous; an execution mistake on the next deploy was downstream
• L5 (Operation) — the P2 threshold was breached without warning because the trend wasn't read

The structural fix is at L5 — the discipline of holding the review's agenda. The SLA review checklist is the gate. The deepest line of the checklist: Sales topics deferred to the bi-weekly sync.

A senior practitioner catches this in 30 seconds: the moment a sales topic enters the SLA review, name it as such and move it. That's a roadmap conversation; let's hold it.

See also

• Practice — SLA review · Weekly client update
• Checklist — SLA review · agenda
• Canon — Did We Serve · The Ongoing Relationship — The SLA Review
• Principle — Predictions over plans — the SLA is the most operational form of prediction the team makes to a client