Every engineering organisation eventually has to answer a deceptively simple question: once a significant release goes live, who looks after it, and for how long? Some teams never answer it explicitly, which is itself an answer, usually a poor one. Others pick a model by habit rather than by fit. This article compares four common post-launch support models, sets out where each is strong and weak, and offers criteria for choosing between them. Whichever model you select, the underlying principle is the same: the period after go-live is a planned phase of delivery, not an afterthought.
Why the choice of post-launch support model matters
The weeks following a release concentrate risk. Real traffic exercises paths that tests did not reach, scheduled jobs run for the first time against new code, and integrations behave differently under production load. At the same moment, the people with the deepest knowledge of the change are being reassigned. The support model determines how those two forces meet: whether knowledge stays close to the risk or drifts away from it.
A model also shapes ownership, on-call load, documentation quality and how cleanly the release eventually transitions to business-as-usual operations. Choosing deliberately, before release, is the point.
The four models at a glance
- Model A: Builder-owned hypercare. The team that built the change supports it through a defined window, then continues to own it.
- Model B: Dedicated hypercare squad. A separate, temporary group supports new releases for a fixed period before handing them to operations.
- Model C: Structured handover to central operations or SRE. Builders brief a central team before go-live; that team supports the release from day one, with builders on escalation.
- Model D: Partner-managed support window. An external delivery partner provides support for an agreed period after launch, then transfers responsibility to the client.
Model A: Builder-owned hypercare
How it works
The development team stays responsible for the release throughout a defined hypercare window and, in most organisations using this model, afterwards as well. Team members join the on-call rota for their own services, and sprint capacity is reduced during the window to leave room for fixes.
Strengths
- Knowledge and responsibility sit in the same place; diagnosis starts with people who understand the code.
- Feedback loops are short: problems seen in production directly shape the next changes.
- No formal handover is needed at the end of hypercare, because ownership does not change.
Weaknesses
- On-call load can become heavy for small teams, particularly if they own many services.
- Without discipline, the hypercare window blurs into ordinary work and never formally closes.
- Teams may underinvest in documentation because “we all know how it works”.
Best fit
Product-aligned teams with stable membership and services they will own long term.
Model B: Dedicated hypercare squad
How it works
A small group, sometimes rotating, sometimes permanent, takes on support for each major release during its stabilization window. Builders brief the squad before go-live and remain available for escalation. When exit criteria are met, the squad hands the system to the long-term operating team.
Strengths
- Builders are protected from interruption and can move on to new work faster.
- The squad develops expertise in stabilization itself: common failure patterns, effective triage, good handover documentation.
- Exit criteria tend to be enforced, because the squad has a clear reason to close each window.
Weaknesses
- Two handovers instead of one: builders to squad, then squad to operations. Each transfer loses some context.
- The squad may lack deep knowledge of a specific change, slowing diagnosis of unusual issues.
- Requires headcount that some organisations cannot justify.
Best fit
Organisations releasing frequent large changes across many teams, where consistent stabilization practice is worth a dedicated function.
Model C: Structured handover to central operations or SRE
How it works
Before go-live, builders run a formal handover to a central operations or SRE team: architecture, known risks, runbook, rollback, contacts. The central team carries first-line responsibility from launch, with builders available on a defined escalation path during the stabilization window.
Strengths
- Clear, single point of operational accountability across many services.
- Strong incentive for builders to produce good documentation, since the central team will refuse a weak handover.
- Efficient for organisations with mature operations functions and standardised platforms.
Weaknesses
- Distance between the people who see problems and the people who can fix them.
- Risk of a “throw it over the wall” culture if escalation paths are weak.
- Central teams can become bottlenecks during periods of many simultaneous releases.
Best fit
Enterprises with an established central reliability function and relatively standardised infrastructure. Our article on organising reliability work when applications are scaling quickly addresses how such teams can avoid becoming a bottleneck.
Model D: Partner-managed support window
How it works
When an external consultancy or delivery partner builds or migrates a system, the contract includes a post-launch support period. The partner handles issues during that period, then transfers knowledge and responsibility to the client’s internal team or another provider.
Strengths
- The client gets builder-level knowledge during the riskiest weeks without needing that knowledge in-house yet.
- Support scope and duration are defined contractually, which forces clarity.
- The final transfer is a natural point to verify documentation and runbooks.
Weaknesses
- If the support window is too short, the client inherits an unstable system.
- Knowledge transfer can be rushed at contract end.
- Response expectations need to be explicit, or misunderstandings arise during incidents.
Best fit
Organisations bringing in outside expertise for major migrations or new platforms. The guidance in our primer on cloud migration for business and technical teams is a useful companion when scoping such engagements.
Side-by-side comparison
Summarising the four models against the dimensions that matter most:
- Knowledge proximity during the window: highest in A and D; moderate in C; lowest in B for unusual issues.
- Number of handovers: none in A; two in B; one in C; one in D.
- Load on builders: highest in A; lowest in B and C; external in D.
- Clarity of window end: weakest in A unless enforced; strongest in B and D.
- Documentation pressure: strongest in C and D; weakest in A.
No model wins on every dimension. The right choice depends on team structure, release frequency and how mature your operations function is.
An illustrative scenario
Consider a retailer replacing its checkout service ahead of a seasonal sales period. The service team is small and already owns several other components, so pure Model A would overload its on-call rota. The company has no dedicated squad, and its central operations group is stretched. A sensible hybrid emerges: the service team keeps first-line ownership for a short, intense window covering the first weekend of sales, with central operations shadowing the rota and attending every review. At the end of the window, a structured Model C handover moves first-line response to operations, with the service team on escalation. Each element is chosen for a reason, and each date is agreed before launch.
Elements every model needs
Regardless of the model, several planning elements recur in successful post-launch periods. They should be decided before go-live:
- A defined window with start and end dates.
- Named people for first response and for escalation.
- An on-call briefing held before the release, not during the first incident.
- One channel or label for all release-related issues.
- Follow-up reviews booked in advance at fixed points.
- Written exit criteria for ending the window.
- Recovery time for the people who carried the cutover.
Treating these as standard parts of delivery, as argued in our piece on building reliability into the delivery process, makes any model more effective.
A note from outside software
The principle of planning the after-period is not unique to engineering. In medical travel, for example, SurgeryPlanet, a planning site for patients seeking treatment abroad, includes lodging, transfers and aftercare within the plan rather than leaving them to the patient. The comparison is useful because it shows the after-period being treated as integral, which is precisely the stance a good post-launch support model takes.
Choosing a model: decision criteria
- If teams own services long term and are stable, start with Model A and enforce window closure.
- If many teams release large changes often, consider Model B for consistency.
- If a strong central reliability team exists, Model C is efficient, provided escalation is robust.
- If an external partner delivers the change, insist on Model D with an adequate window and a verified transfer.
Hybrids are common and often sensible: for example, Model A for routine releases and Model B for major migrations. What matters is that the choice is explicit, written down and made before the release, so the week after go-live is governed by a plan rather than by whoever happens to notice the first problem.

