The refresh cliff is a budgeting problem: planning device lifecycles beyond warranty expiry

Most device estates do not fail all at once. They fail in clusters. A batch of laptops bought in the same year reaches the end of its manufacturer warranty, battery life falls away, and the service desk starts to see the same faults repeated across users. The result is familiar to many UK mid-market organisations: a sudden capital request for several hundred replacements, presented to the finance director as unavoidable.

The cliff is rarely a hardware event. It is a budgeting event. Devices have a predictable life; the decision to replace them has been deferred year after year because no one owned the multi-year plan. A five-year refresh model does not remove the cost. It makes the cost visible, smooth and defensible.

Why the cliff forms

Refresh cliffs usually emerge from three ordinary behaviours. First, procurement is treated as a one-off purchase rather than a lifecycle. A fleet bought in a single year creates a single replacement year. Second, warranty expiry is not used as a planning trigger. By the time failure rates become visible, the organisation is already buying under pressure. Third, support costs sit in a different budget from hardware, so the true cost of old devices is never compared with the cost of replacing them.

The finance director then sees only the peak. They do not see the years of deferral that produced it. A lifecycle plan corrects that asymmetry by showing the estate as cohorts with known ages, warranty positions and replacement dates.

A five-year model in practice

The model need not be complex. Start with a device register that records purchase year, device type, warranty end date, criticality and assigned user group. Group the estate into annual cohorts. If the organisation intends to keep laptops for five years, each cohort should be replaced in its sixth year, with the annual replacement volume settling at roughly one fifth of the estate once the model is steady.

The first pass will expose lumpiness. A large cohort may be due in the same year as a lease expiry or a server renewal. That is useful information, not a failure. The plan can then phase replacements, extend warranties selectively, or adjust the standard life for different user groups. A call-centre device and a consultant laptop may not need identical treatment.

Include the less visible costs in the model: battery replacement, docking stations, operating system support, endpoint protection licences and disposal or buy-back. These are part of the lifecycle whether or not they appear in the original purchase order.

The hidden costs of running old hardware

Old devices are not free once the warranty ends. They consume support time, create security exposure and slow users down. A device that takes longer to boot, loses battery charge during a site visit or fails to run current collaboration tools imposes a cost on the business even when no invoice arrives.

The security dimension deserves board attention. Manufacturer warranties and operating system support are not the same thing, but both affect the organisation's ability to patch and recover. An estate with a long tail of unsupported devices is harder to assure and more expensive to defend. That is a governance issue, not simply an IT preference.

Support load is the most practical measure. Compare ticket volumes and average resolution time for devices by age band. If the oldest quartile generates a disproportionate share of incidents, the case for replacement becomes evidence-based rather than aesthetic. Staff productivity can be measured through survey responses, login and boot times, or the proportion of users reporting performance as a blocker.

A one-page board case

The board does not need a device catalogue. It needs a decision. A one-page case should state the size and age profile of the estate, the current replacement trajectory, the proposed annual plan and the cost range. It should show what happens if the organisation does nothing: a peak year, a support burden and a security exposure that grows with every deferred cohort.

The metrics should be few and consistent. Percentage of devices beyond warranty. Average age by user group. Support tickets per hundred devices by age band. Patch compliance. Replacement cost per year under the proposed plan. Where data is incomplete, say so and propose a short baseline exercise rather than inventing precision.

The decision requested should be explicit: approve a five-year lifecycle budget, delegate annual model selection to procurement within an agreed envelope, and require a yearly review of assumptions. That gives the finance director control without requiring them to approve every individual purchase.

Procurement and supplier implications

A lifecycle plan changes the conversation with suppliers. Instead of negotiating a single large order under time pressure, procurement can agree standard models, warranty terms, battery service and buy-back arrangements. Leasing or device-as-a-service may suit some organisations; outright purchase may suit others. The test is whether the arrangement smooths cost and clarifies end-of-life responsibilities.

Supplier management should also cover disposal and data sanitisation. A refresh plan that creates a storage room of old laptops has not solved the problem. Certified wipe, asset recovery and documented disposal should be part of the standard terms.

What makes the plan signable

A finance director will sign a plan that is predictable, evidenced and reversible. Predictable means the annual cost is broadly stable after transition. Evidenced means the age profile and support data are visible. Reversible means assumptions can be reviewed each year rather than locked into a five-year forecast.

The refresh cliff is best avoided before it appears. By treating devices as a managed estate rather than a series of emergency purchases, UK organisations can move from annual surprise to a procurement rhythm that the board recognises and the finance function can plan around.