End-User Refresh Business Case with an Unreliable Register
Corvus Insurance Services has 2,400 end-user laptops. Of these, 1,050 are 5.5 years old and run Windows 10, and 780 of those fail Windows 11 hardware checks: no TPM 2.0, or an unsupported processor. Repair incidents run at 0.64 per device-year on the old cohort against 0.19 on machines under two years old, at an average GBP 310 per repair and 3.5 lost hours per incident at a loaded GBP 48 per hour. A refresh quote for 1,050 devices is GBP 1.28M capital plus GBP 96 per device for deployment. Alternatively, extended security updates would cost GBP 61 per device per year for up to three years, and 300 of the oldest devices could move to VDI at GBP 420 per device per year. Finance uses a 12% hurdle rate. The register shows 210 of the 1,050 assigned to users who have not logged in for 60 days - contractors and leavers - and the last physical stocktake was 14 months ago. The field service vendor has also flagged 180 devices with swollen batteries.
1. What must be validated before any business case is presented, and why does it matter financially?
Read the worked answer
The register is not a safe basis for a seven-figure decision. Three checks come first.
1. Physical existence and assignment. 210 of the 1,050 are assigned to users silent for 60+ days. A stocktake and an identity reconciliation against HR and the contractor list will probably remove a meaningful share. At roughly GBP 1,411 per device (GBP 1,315 capital plus GBP 96 deployment), refreshing 200 devices nobody uses wastes about GBP 282,000 - and the same devices generate ongoing repair and licence cost. Scope reduction is the cheapest saving in the case.
2. The Windows 11 failure reason. Missing TPM 2.0 is sometimes a firmware setting or a module that can be fitted cheaply; an unsupported processor is a hard block. Splitting the 780 into "remediable" and "hard-blocked" can shrink the mandatory refresh substantially.
3. The failure-rate baseline. 0.64 incidents per device-year at GBP 310 plus 3.5 hours at GBP 48 gives GBP 305.92 per device-year on the old cohort, about GBP 321,000 a year across 1,050 units. The new cohort runs at GBP 90.82, roughly GBP 95,400. The GBP 225,000 annual delta is the number that must compete with the capital request - not the headline repair cost, and not a vague claim about productivity.
Also confirm warranty and lease positions and whether the 180 swollen batteries sit inside the 1,050.
2. Compare refresh, extended security updates, and partial VDI over a three-year horizon using the 12% hurdle rate. Which do you recommend?
Read the worked answer
Option A - Refresh 1,050: GBP 1,380,800 at t0 (GBP 1.28M capital plus GBP 100,800 deployment). Annual run-cost saving vs the old cohort is GBP 225,855. Three-year annuity factor at 12% is 2.402, so the present value of savings is about GBP 542,000. NPV ≈ GBP 838,000. The fleet is three years old at the end of the horizon, with residual value and several years of life remaining.
Option B - Extended security updates: GBP 61 x 1,050 = GBP 64,050 a year, plus the old cohort's run cost of GBP 321,216, so GBP 385,266 a year. PV = GBP 925,200. NPV ≈ GBP 925,000 - and that excludes the refresh still owed at year three, which at a 0.712 discount factor adds roughly GBP 983,000 in present value. B only looks cheaper if the horizon stops exactly where the deferred liability begins.
Option C - Partial VDI: GBP 420 per device per year, against roughly GBP 305 annual run cost plus GBP 470 a year of amortised capital on an owned device. VDI is competitive for task-based and contractor users with low compute needs, and pointless for power users.
Recommendation: refresh the hard-blocked devices, exclude everything the stocktake proves unused, and route the 210 low-utilisation and contractor seats to VDI rather than buying hardware. The combination beats either pure option.
3. How should the swollen batteries and the Windows 10 end-of-support timeline affect sequencing?
Read the worked answer
Both are schedule drivers that override the financial optimisation.
Swollen batteries are a safety and fire risk, not an asset-value question. They should be quarantined and removed from service immediately, independent of whether the device is refreshed, and before the business case is even approved. Identify them by serial from the vendor's field-service report and battery telemetry, collect through a battery-safe logistics route, and record disposal under WEEE obligations. Leaving 180 known-swollen units in users' bags while a business case is modelled is indefensible if one ignites.
The Windows 10 end-of-support date of 14 October 2025 is a hard external deadline. Extended security updates buy time but cover only security fixes - not feature updates, not third-party browser support - so check whether cyber-insurance terms, client contracts or regulatory obligations require a supported platform. If they do, ESU is not an option at any price and the refresh date is fixed.
Sequencing:
- Quarantine and replace the battery-risk devices.
- Stocktake and remove unassigned and leaver devices from scope.
- Split the 780 by remediable versus hard-blocked.
- Refresh hard-blocked devices before the deadline with adequate lead time for supply constraints.
- Move low-utilisation seats to VDI; retain remediable devices with a TPM fix.
Publish the plan against the deadline, not against the financial year.