There’s a conversation that happens in almost every large delivery relationship, and it goes wrong the same way every time.
The vendor says the feature is done. The enterprise says it isn’t. Neither is lying. They’re using the same word to mean different things.
I’ve sat on both sides of this table, Nearly more than a decade on the vendor side, delivering into Tier-1 telecom operators worldwide, and more recently on the enterprise side, receiving that same kind of delivery and deciding what’s actually ready to ship. That switch in seats taught me something that doesn’t show up in any delivery methodology: the two sides aren’t just negotiating scope. They’re operating from entirely different definitions of the same milestones.
“Done” means different things to different people, and nobody says so out loud.
To a vendor PM, done usually means the acceptance criteria in the SOW are met. Tested, documented, delivered per spec. To an enterprise sponsor, done means the business outcome it was funded for is actually happening, adoption is real, the metric moved, the board isn’t going to ask why nothing changed. Both are reasonable definitions. They’re just not the same one, and contracts rarely force anyone to reconcile them until a milestone review turns tense.
SLAs get gamed from both directions, not out of bad faith, but because they’re built to be gamed.
Vendors optimize for whatever the SLA actually measures, not what it was meant to capture. Ticket closure time gets protected even when the underlying issue reopens next week. Enterprises, in turn, sometimes write SLAs that are technically satisfiable while the real problem, a slow-moving root cause, a fragile integration sits untouched. I’ve watched both sides do this without anyone involved thinking of it as manipulation. It’s just what a metric does when nobody’s checking whether it still means what it was supposed to mean.
The real translation work isn’t contractual. It’s cultural, and it’s continuous.
The vendor side tends to escalate risk quietly, hoping it resolves before anyone senior notices. The enterprise side tends to assume silence means things are fine. Both instincts are understandable. Together, they produce exactly the kind of surprise that shows up in a board review three months too late.
What actually closes the gap isn’t a better contract. It’s someone in the room who’s genuinely sat on both sides who knows what the vendor isn’t saying and why, and knows what the enterprise sponsor actually needs to hear versus what they’re asking for. That’s not a process fix. It’s a specific kind of fluency, and most delivery organizations don’t have anyone who’s built it, because very few people move between these two seats in a single career.
The biggest lesson from doing both isn’t about SLAs or governance frameworks at all. It’s that most delivery failures aren’t failures of execution. They’re failures of translation — two competent teams, each doing their job well, quietly building toward different definitions of success.