A migration proposal cuts total three-year cost from £4.2m to £3.1m. The finance director objects to it. What mechanism makes a cheaper design worse from their seat, and what would you change about the proposal?

The mechanism

Total cost is not the number the finance director is measured on. Where a cost lands in the accounts changes the reported profit even when the cash is identical.

Buying hardware is capital expenditure. £2.8m of servers is recorded as an asset and charged against profit as depreciation over its useful life — at five years, about £560k a year. A subscription is operating expenditure and hits profit in the year it is incurred. So a plan whose three-year cash total falls from £4.2m to £3.1m can still raise this year's reported operating cost: £560k of depreciation plus some run cost is replaced by roughly £1.03m a year of subscription, a swing of several hundred thousand pounds in the wrong direction on the line the director reports.

It gets sharper if the business is measured on EBITDA — earnings before interest, taxes, depreciation and amortisation. Depreciation sits below that line and subscriptions sit above it, so moving from owned kit to a service reduces reported EBITDA by the full subscription cost while the owned-hardware alternative reduced it by nothing. A business valued on an EBITDA multiple has a direct financial reason to prefer the more expensive plan.

One more wrinkle, often decisive: internal engineering labour on a new asset can sometimes be capitalised, while a vendor subscription cannot. That is why "build it ourselves" occasionally survives a cost comparison it should lose.

The consequence people miss

This is not finance being obstructive or short-termist. It is a stakeholder optimising a different, legitimate objective, and an architect who presents only total cost has not given them the information they need to say yes. The same mechanism explains a set of objections that look irrational from an engineering seat: resistance to managed services, preference for multi-year prepayments, and a sudden enthusiasm for reserved capacity near the year end.

What I would change about the proposal

• Show both views. Three-year total cash and the annual profit-and-loss effect by year, split into capital and operating. Two tables, same proposal.
• Ask which line the sponsor is measured on before choosing the commercial shape. It takes one question and it changes the recommendation.
• Offer a commercial structure, not only a technical one. A three-year prepaid commitment, a reserved-capacity purchase, or keeping one capitalisable component in-house can often deliver most of the saving in a shape the accounts accept.
• Name who else this affects. Procurement cares about contract term and exit; the tax function cares about the structure; the budget holder cares about which cost centre pays.

Where this stops mattering

In a business that plans on cash rather than accounting profit — many private companies, and most early-stage ones — the capital-versus-operating distinction barely moves a decision, and arguing it wastes the review. Ask how the organisation is measured before you build the argument. Public companies, private-equity-owned businesses and regulated entities with capital requirements care a great deal. A bootstrapped startup with 18 months of runway cares only about the cash.

Common weak answers

• "Explain the total cost of ownership more clearly." The director understood it. They are objecting to something the TCO number does not contain.
• "Escalate to the CTO." This wins once and converts a stakeholder into an opponent for every future proposal.
• "Cloud is operating expenditure and that is a benefit." It is a benefit to the organisation that wants to avoid up-front capital and a cost to the one measured on operating profit. Which it is depends entirely on the business, and that is the thing to find out rather than assume.