Support team working at computers during a hypercare shift

Go-Live Is Not the End: Hypercare and the Handover Nobody Plans

There is a photograph taken at every go-live: the team around a laptop, someone pressing the button, applause. What the photograph never shows is week two, when the project manager has been reassigned, the vendor’s best people have rolled onto the next client, the first payroll run hits the new system, and the only person who knows why that interface behaves strangely is on a plane.

Projects are staffed, governed and budgeted to reach go-live. Almost nobody staffs, governs or budgets the part where the organisation actually starts depending on the system. That gap has a name, hypercare, and treating it as a phase with a plan, rather than an improvised afterglow, is one of the cheapest risk reductions in delivery.

What hypercare actually is

Hypercare is the defined period after go-live when the project team stays in place with heightened support: faster response, daily monitoring, direct access to the people who built the thing. Defined is the operative word. It has a start, an end, entry into it, and exit criteria out of it. Without those, one of two failures happens: the team dissolves on day three and incidents pile onto an unprepared helpdesk, or hypercare never ends and the project quietly becomes a permanent support department, which is one of the slow-motion failures we described in why IT projects fail.

Planning it before go-live, not after

The hypercare plan is written in the weeks before cutover, while there is still leverage over the vendor and attention from the sponsor. Ours contains five elements, none of them exotic:

Duration tied to business cycles, not calendars. Two to six weeks is typical, but the honest rule is: hypercare covers at least one occurrence of every critical business event. First month-end close, first payroll, first VAT submission, first peak-load day. A system that has survived its first month-end has proven something a quiet fortnight cannot.

A daily rhythm. Fifteen minutes each morning: incidents from the last day, severity, owner, trend. The same discipline as incident triage, run calmly. The trend line matters most; hypercare is going well when severity drifts down and the topics shift from “broken” to “how do I”.

Severity definitions and response times in writing. Agreed with the vendor before go-live, because negotiating response times during an outage is negotiating from the floor. If the delivery ran under a fixed-price contract, this is a clause we insist on early, as we argued in managing system integrators.

Named people, physically or virtually present. Hypercare is not a mailbox. It is specific builders on call, with their calendars protected. The vendor’s A-team rolling off at go-live plus one day is the oldest trick in the industry; the contract keeps names in place until exit criteria are met.

Exit criteria on one page. For example: no open severity-1 incidents, severity-2 below an agreed count and trending down, first month-end closed successfully, helpdesk resolving a defined share of tickets without escalation, operations sign-off. Exit is a decision at a gate, with evidence, exactly like every other gate in a right-sized governance setup.

The handover is a deliverable, not a meeting

What kills systems in month three is not bugs; it is the knowledge walking away. The handover to operations deserves the same seriousness as any deliverable: runbooks for routine operations and known failure modes, a triage guide for the helpdesk, access and credentials transferred and tested, monitoring alerts pointed at people who will still be there, and at least one shadowing cycle where operations runs the critical events while the project team watches, not the other way around.

A test we recommend to every client: before hypercare ends, have operations handle a full week, including one deliberately simulated incident, with the project team forbidden to touch keyboards. What that week exposes is exactly what would have been discovered in month three, at a far worse moment.

The uncomfortable budget line

Hypercare and handover cost real money, typically a noticeable slice of the delivery budget, and proposals that omit it look cheaper. That is precisely backwards: the omission is a cost moved to the future and multiplied, invoiced later as emergency support at premium rates. When we rescue projects, the ones described in the first 30 days of a project rescue, a missing or botched go-live transition is one of the most common root causes on the table.

We plan cutover, hypercare and handover as standard parts of our IT project management services. The applause photograph is nicer when nobody is faking the smile.

Frequently Asked Questions

How long should hypercare last?

Two to six weeks for most systems, but anchor it to business events rather than dates: it should cover at least one month-end close, payroll run or equivalent critical cycle, and end only when written exit criteria are met.

What are typical hypercare exit criteria?

No open critical incidents, declining severity trend, successful completion of the first critical business cycles, the helpdesk resolving routine tickets independently, and formal sign-off by operations.

What belongs in a go-live handover?

Runbooks, a helpdesk triage guide, transferred and tested access, monitoring routed to operations, and a shadowing period where operations runs the system while the project team observes.


Editorial note: This article was researched and drafted with the assistance of Claude (Anthropic), and reviewed and approved by Amazing Projects before publication.

Similar Posts