There is a moment in every agency engagement nobody writes into the proposal: the week after launch, when the Slack channel goes quiet. The code works, the invoices are paid, and the client is now the sole owner of a system they watched being built but never operated. Most agencies engineer for the demo. The gap between launch and competent ownership is where projects actually die.
Ship-and-vanish is a pricing strategy, not an accident
Agencies vanish because the market rewards it: handover weeks are hard to sell, look like padding in a proposal, and the costs of skipping them land six months later — safely after the case study is published. The client pays either way. They just pay later, to a different vendor, at incident rates.
Every system has an operations team. The only question is whether it was trained on purpose or recruited by the first outage.
The clause that started it
From our master agreement, §7
Delivery is complete when the client’s team has operated the system for 30 consecutive days — deploys, incidents, and rollbacks included — without vendor intervention. Until then, we are on call and the final invoice waits.
We added that clause after a painful retro in 2024, and it rearranged our incentives overnight. If our own money waits on the client’s team being genuinely able to run the thing, then runbooks, dashboards, and training stop being deliverables-in-name — they become the critical path.
What a real handover contains
- Runbooks written against real incidents from the staging period, not hypotheticals.
- A shadow rotation: their on-call shadows ours for two weeks, then we shadow theirs.
- A deliberately broken staging environment as the final exam — restore it, trace it, roll it back.
- The “bus file”: every credential, every vendor contact, every undocumented quirk, in one audited place.
Run-and-transfer costs us roughly eight percent of an engagement. It has also ended the 2 a.m. “the site is down and the agency is gone” story for every client who signed it — which is why they come back.