Most marketing–sales SLAs have the same lifecycle: drafted in an alignment workshop, announced with mild optimism, violated within three weeks, and quietly abandoned by the end of the quarter — leaving behind slightly more cynicism than existed before. Then, a year later, a new leader arrives and the cycle restarts.
The problem is rarely the intent. It’s the design. SLAs fail for structural reasons, and the fixes are structural too.
Why SLAs die
They’re one-sided. The classic SLA is a list of obligations on sales — respond in X hours, make Y attempts — written by marketing. Sales experiences it as a performance review they didn’t ask for, drafted by a team whose leads they don’t trust. A one-sided SLA isn’t an agreement; it’s a memo.
They’re built on unmeasurable or disputed inputs. “Respond to qualified leads within 4 hours” presupposes that both sides agree what qualified means and that response time is actually instrumented. If the MQL definition has drifted or speed-to-lead isn’t measured honestly, the SLA is arguing about numbers nobody can produce.
They have consequences nobody will enforce. SLAs that end in escalation-to-the-CRO clauses die the first time enforcement would be socially expensive — which is immediately.
The two-sided structure
An SLA that survives is symmetrical. Both teams commit; both teams are measured; both teams get something.
Marketing commits to:
- Lead quality: MQLs meet the written, signed definition, with a floor on sales-acceptance rate (e.g. ≥70% — set from your actual baseline, then ratchet).
- Completeness: every handed-off lead carries the fields sales needs to act (the context, not just the contact).
- Volume honesty: a forecast range, not a single heroic number.
Sales commits to:
- First human touch within the agreed window on accepted leads (measure median and 90th percentile).
- A minimum follow-up sequence before a lead can be recycled (e.g. 5 touches across 10 working days).
- Disposition with reasons: every rejected lead gets a categorised rejection reason. This is the most valuable and most resisted clause — those reasons are the feedback loop that keeps the definition honest.
The mechanism that makes it self-balancing: rejected leads don’t vanish, they return to marketing with the reason, and acceptance-rate and response-time reports are reviewed together, monthly, in the same meeting. When acceptance drops, that’s marketing’s problem to fix. When response times slip, that’s sales’s. The symmetry is what keeps it from becoming a blame instrument.
Set thresholds from baselines, not aspirations
The fastest way to kill an SLA is to set targets from a benchmark blog post. If your current median response time is 26 hours, a 1-hour SLA is fiction, and everyone will treat the whole document accordingly. Measure the current state first (the speed-to-lead instrumentation covers how), set the first thresholds slightly better than baseline, and ratchet quarterly. An SLA that’s 100% met at a modest standard builds the trust needed to raise the standard. One that’s 40% met at an ambitious standard builds nothing.
Consequences that are mechanisms, not punishments
Skip the escalation theatre. The consequences that work are automatic and impersonal:
- A lead untouched past the SLA window is automatically re-routed to the next rep in rotation (nobody loses face; the lead doesn’t wait).
- Acceptance rate below the floor triggers the definition review, not a blame meeting — the system assumes the definition drifted before it assumes anyone failed.
- Both metrics live on one shared dashboard both leaders see, built on the numbers you’ve made bulletproof. Visibility is most of the enforcement.
The first honest baseline is usually uncomfortable for both sides, which is the point. Marketing learns that a chunk of “qualified” leads don’t survive contact; sales learns its real median response time is measured in hours or days, not the minutes everyone assumed. The clause that gets resisted is nearly always the one making response time visible per rep — because visibility is the actual enforcement, and everyone knows it. A quarter after both numbers live on one shared dashboard, the arguing tends to stop: not because anyone was punished, but because two teams can finally see the same figures and can’t dispute their way out of them.
The one-page rule
If the SLA doesn’t fit on one page, it won’t be read, and if it isn’t reviewed quarterly — same meeting as the MQL definition and the routing audit, they’re the same hygiene ritual — it will quietly detach from reality like every SLA before it. One page, two signatures, one monthly dashboard, one quarterly review. The paperwork is boring by design; the alignment it buys is not.
—
The handoff — definition, SLA, routing — is lens two of the Pipeline Leak Audit. If your SLA is on its second or third resurrection, a free health check can usually spot the structural reason in 30 minutes.