The Costs Below the Line: Why Outsourcing Business Cases Always Understate the Transition
The arbitrage was never the fiction. The journey was.
Executive Summary
Every outsourcing decision of the present wave rests on a single comparison: what a function costs to run here against what it will cost to run there. That comparison is almost always honest about the destination and almost always silent about the passage. The business case models the steady state — the lower run-rate that arrives once the work has moved, settled, and stabilised — and treats everything required to reach that state as a rounding error, a one-off, a line below the line.
The pattern that recurs across the offshoring and outsourcing programmes of these years is not that transition costs are occasionally missed. It is that they are systematically understated, and understated in the same direction, by the same margin, for the same reasons, again and again. This essay argues that the understatement is structural rather than accidental. It is produced by the incentives of the people who write business cases, by the accounting conventions that separate run-rate from exceptional cost, by a genuine asymmetry of knowledge between those who hold the work and those who will receive it, and by the simple political fact that a number has to clear a hurdle before anyone is allowed to act on it.
The consequence is not merely a forecasting error to be tightened with better spreadsheets. It is a window onto the wider gap between transformation intent and transformation reality. The model treats the move as an event; the organisation experiences it as a passage that takes far longer, costs far more, and hollows out capability it did not know it depended upon. Understanding why the transition cost is always missing tells us more about how our organisations actually change than any number of post-implementation reviews. This is the longer view: not a warning against outsourcing, but a plea to cost the journey with the same seriousness we bring to the destination.
The Number That Cleared the Board
The paper is signed off in the autumn. A transactional processing function of some three hundred people, running at a little over twenty million pounds a year, is to be moved to a lower-cost location. The unit-cost arbitrage is genuine and large — better than half — and the business case, prudently discounted, promises an annual saving approaching nine million pounds once the work has settled. Breakeven, the model says, falls a little after month nine. The board, which has seen three of these cases already this year, approves it in twenty minutes.
Fifteen months later the function has moved, the headcount here has gone, and the run-rate is, more or less, exactly what the case predicted. And yet the programme has not saved a penny. Breakeven, when it finally arrives, lands somewhere past month twenty. The gap between the promise and the passage — the year of dual-running, of travel and knowledge transfer, of a productivity trough nobody had named, of a retained team that turned out to be larger and more expensive than the slide had allowed — has quietly consumed the first year and a half of returns.
Anyone who has stood inside one of these programmes recognises the shape of it. The destination was described with precision. The journey was described with a phrase — transition costs, estimated — and a number that had clearly been reverse-engineered to leave the headline intact. This is not a story about one badly built model. It is a story about a mistake we make on purpose, and keep making, because the forces that produce it are stronger than the forces that would correct it.
What Actually Lives in the Transition
Before asking why the number is wrong, it is worth being concrete about what the number is meant to contain — because the abstraction “transition cost” is itself part of the problem. Collapsed into two words on a slide, it loses the texture that would make it frightening. Unpacked, it is a catalogue:
- Parallel running. The work does not stop while it moves. For months, the same process runs in two places at once — the outgoing team correcting the incoming team’s output, the incoming team learning by shadowing. Two payrolls, one output.
- Knowledge transfer. Someone has to teach the work, and the people who know it best are, by definition, the people being made redundant by the move. Travel, documentation, the slow and reluctant transfer of things that were never written down.
- The retained organisation. Contract management, service governance, quality assurance, a layer of retained expertise deep enough to challenge the provider. This does not appear when the work leaves; it grows, and it is permanent, and the first cut of the business case almost never sizes it honestly.
- Severance, retention and consultation. Redundancy is not free, and the people you most need to stay until the last day are precisely those with the best prospects elsewhere. Retention bonuses to hold the knowledge long enough to transfer it are a cost of the move, not of the destination.
- The productivity trough. For a period after go-live, the new operation is slower, less accurate, and more expensive per unit than either the old operation or the promised steady state. Error rates rise; rework rises; the exception queue lengthens. This is the least-modelled cost of all, because it does not arrive as an invoice.
Set out this way, the honest picture looks very different from the slide. Consider the same three-hundred-person function, with the transition costs restored to visibility:
| Cost line | As modelled | As incurred |
|---|---|---|
| Parallel running (dual payroll) | £0.4m (token) | £2.9m |
| Knowledge transfer & travel | £0.3m | £1.6m |
| Severance & retention | £1.8m | £3.1m |
| Retained organisation (year one) | £0.6m | £2.4m |
| Productivity trough / rework | not modelled | £2.2m |
| Total first-18-month transition | £3.1m | £12.2m |
The steady-state saving was real. The programme did, eventually, take nine million pounds a year out of the cost base. But a transition modelled at three million and incurred at twelve does not merely dent the return — it moves breakeven by more than a year and turns a case that looked decisive on a single page into one that, honestly costed, the board might have sequenced differently, scoped differently, or timed for a quieter year. The arbitrage was never the fiction. The journey was.
Why the Understatement Is Structural
If this were simply a matter of analytical carelessness, it would not recur so reliably. Careless errors scatter in both directions; this one points the same way every time. That regularity is the tell. The understatement is manufactured by the system that produces business cases, and four forces do most of the manufacturing.
The business case is an advocacy document
We speak of the business case as though it were a forecast — a neutral estimate of what will happen. It is nothing of the kind. It is the instrument by which a sponsor secures permission and funding for something they have already decided to do. Its author is rarely disinterested; more often they are the person whose credibility, budget, and standing now ride on the decision proceeding. A forecast wants to be right. An advocacy document wants to be approved. The two produce very different treatments of an inconvenient cost.
Transition costs are the most inconvenient costs of all, because they fall early, when the case is most fragile, and because they push breakeven out past the horizon over which sponsors are typically judged. Every incentive pulls toward compressing them into a small, confident, below-the-line figure. No one falsifies anything. They simply resolve every uncertainty in the favourable direction, and there are a great many uncertainties to resolve.
The accounting frame hides the passage
Our financial conventions actively assist the understatement. The saving is a run-rate — a permanent reduction in ongoing cost, the kind of number that flows into next year’s budget and the year after’s. The transition is a set of one-off, exceptional, non-recurring items, quarantined in a separate section of the model and often in a separate part of the organisation’s accounts. The comparison that decides the case sets an annual run-rate saving against a one-off cost, and the human mind — and the discounting arithmetic — treats a permanent gain as obviously worth a temporary pain.
The most dangerous phrase in any outsourcing case is one-off cost of transition. It invites the reader to discount the very number most likely to sink the return, precisely because it will not recur — as though a cost that strikes only once strikes only lightly.
The framing is not wrong, exactly. Transition costs genuinely are non-recurring. But labelling them as exceptional licenses a carelessness we would never permit in the run-rate. A ten per cent error in the ongoing saving would be litigated line by line. A three-hundred per cent error in a one-off is waved through, because it is “only” a one-off.
Nobody yet knows what the work really is
There is a deeper and more honest reason the number is wrong, and it deserves respect rather than cynicism. At the moment the business case is written, no one — not the sponsor, not the provider, not the team doing the work — fully knows what the work actually consists of. Years of accumulated exceptions, workarounds, undocumented judgement calls, and quiet local knowledge live in the heads of the people who do the job. The process map describes the process as it was designed. The transition has to cope with the process as it is practised, and the distance between the two is exactly the distance the transition budget fails to cover.
This is why knowledge transfer always overruns, why the productivity trough is always deeper than expected, and why the retained organisation always ends up larger: each is a cost of discovering, in flight, how much tacit knowledge the function actually held. You cannot budget accurately for the retrieval of knowledge whose extent is, by its nature, unknown until it is lost. The understatement here is not dishonesty. It is the unavoidable optimism of not yet knowing what you are about to find out.
The number has to clear the hurdle
Finally, and most simply, the case has to work. Somewhere in the organisation there is a hurdle rate, a required payback period, a threshold below which capital will not be committed. The transition cost is the most elastic variable in the model and the one over which the author has the most discretion. If the honest number breaks the hurdle and the optimistic number clears it, there is an almost gravitational pull toward the optimistic number — not through fraud, but through the endless small negotiations by which an uncomfortable estimate is talked down to a fundable one. The hurdle does not test the case. The case is bent to fit the hurdle.
The Case for the Defence
It would be too easy to leave it there, as an indictment. The strongest version of the opposing view deserves a hearing, because it is not foolish and is held by serious people.
The defence runs like this. Perfect foresight is not available, and demanding it is a recipe for paralysis. Every worthwhile transformation requires acting on incomplete information; if we insisted on a fully-loaded, pessimistically-costed transition before committing to any move, we would never move, and our competitors — less fastidious, more decisive — would take the arbitrage we hesitated over. A degree of optimism in the business case is not a bug but the necessary lubricant of action. Transition costs, moreover, genuinely are one-off; over a five- or seven-year horizon even a badly understated transition is dwarfed by the cumulative run-rate saving, so the understatement, while real, is second-order. Better to be roughly right and moving than precisely right and stationary.
There is truth in this, and any honest treatment has to concede it. Over a long enough horizon, a large arbitrage does swamp a mis-costed transition; many of these programmes, judged at year five, did deliver. And it is true that a culture which punishes every optimistic estimate will simply teach its people to stop proposing anything ambitious.
But the defence answers a charge that has not been made. The argument here is not for perfect foresight, nor for pessimism as a virtue, nor against outsourcing. It is against a systematic, one-directional error that the organisation has learned not to see. Three things the defence cannot dissolve:
- The asymmetry corrupts the choice, not just the estimate. If transition costs were understated randomly, the defence would hold — we would be roughly right on average. But they are understated always and only downward, which means every option that front-loads cost is penalised relative to every option that hides it. The bias does not just make cases optimistic; it makes the organisation prefer the wrong ones — the big-bang lift over the phased move, the aggressive timeline over the survivable one — because the honest alternative always looks worse on the page.
- The horizon is a convenient fiction. The five-year vindication assumes the arrangement survives five years unchanged. Many do not. Providers are switched, contracts renegotiated, functions repatriated or moved again — each event triggering a fresh transition whose cost was never in the original case. If the steady state is itself transient, the “one-off” cost is not one-off at all; it is a recurring toll on a strategy of perpetual motion, and the long-horizon defence quietly collapses.
- A mis-set baseline compounds. The understated transition does not merely delay the return. It sets the expectation against which the programme is then judged, funded, and staffed. Teams are disbanded on the assumption that the trough will be shallow; contingency is refused on the assumption the number was sound; the next case in the pipeline is built on the last one’s fictional transition. The error propagates through the portfolio.
The defence, in the end, is an argument for acting under uncertainty, which no one disputes. It is not an argument for acting under a bias we have chosen not to correct.
What the Gap Tells Us About Ourselves
Step back from the arithmetic and the transition cost becomes something more interesting than a line item. It is the precise point at which transformation intent meets transformation reality, and the size of the gap between the modelled and the incurred is a fairly exact measure of how much our organisations misunderstand their own change.
We model change as an event: a decision, a transition, a new steady state — a clean step from one plateau to another. We experience change as a passage: a long, turbulent, expensive crossing in which capability is lost before it is regained, in which the organisation is for a time worse at the thing it is trying to improve, and in which the value leaks out through a hundred unbudgeted seams. The transition cost is what the passage costs. Its chronic understatement is the accounting signature of an organisation that believes, against all its own experience, that change is an event.
“The steady state is a place we describe with great confidence and reach with great difficulty; the whole cost of transformation lives in the distance between the two.”
There is a particular institutional forgetting at work. Each programme is run by a different sponsor, approved by a board whose composition shifts, staffed by people who move on. The lesson of the last transition — that it cost four times the estimate — is not carried into the next case, because the next case is written by someone new, under the same hurdle, with the same incentives, in an organisation that keeps no honest memory of what its transitions actually cost. We do not learn, because learning would require us to hold, in one place, the modelled transition and the incurred transition side by side, and almost no organisation of this era troubles to do so. The post-implementation review, where it happens at all, celebrates the run-rate and passes over the year that was lost reaching it.
This is the deeper reading of the hidden cost of outsourcing. The offshoring wave is, in one light, a rational response to a genuine and durable cost differential. In another, it is a vast natural experiment in how badly we cost our own change — and the transition line, buried below the line, is where the experiment records its result.
The Longer View
None of this is an argument against moving work to where it can be done better or more cheaply. The arbitrage is real, the differential is durable, and standing still is its own expensive decision. The argument is for costing the journey with the same seriousness we bring to the destination — for treating the transition not as an inconvenient one-off to be minimised on the page, but as the thing that actually determines whether the strategy pays.
What would a more honest practice look like? Not a new methodology — the failure here is not the absence of a technique but the absence of the will to use the ones we have. A few principles are enough, and they are principles of temperament more than of analysis:
- Give the transition its own profit-and-loss. Model the passage as a phase in its own right, with its own costs, its own duration, and its own governance — not as a haircut on the destination. A transition that has to justify itself is a transition that gets estimated honestly.
- Baseline the tacit. Assume, as a matter of discipline, that the work is more complex than its documentation, that knowledge transfer will overrun, and that the retained organisation will be larger than the first cut suggests. Cost the process as practised, not as designed.
- Budget the trough. Name the productivity dip, put a number on it, and fund the organisation to survive being temporarily worse at its own job. The trough is not a risk to be hedged; it is a certainty to be financed.
- Protect the contingency from the hurdle. The most valuable number in the case is the one most likely to be negotiated away. If the honest transition breaks the hurdle, that is information about the decision, not a defect in the estimate — and the right response is to change the plan, not the number.
- Keep an honest memory. Hold the modelled and the incurred transition together, programme after programme, so that the organisation finally learns its own coefficient — the reliable multiple by which it under-costs its own change — and applies it before, not after.
The maturity we are reaching for is not better forecasting. It is the willingness to look directly at the part of transformation we would rather not price. In my experience the organisations that eventually master outsourcing are not the ones with the sharpest arbitrage models; they are the ones that stopped lying to themselves about the year in between. They learned to read the transition line not as an exception to the business case but as its most truthful part — the one place where the number tells you what the change will actually take.
The savings, in the end, usually arrive. They simply arrive later, and cost more to reach, than we were ever willing to write down. The longer view begins with admitting that — and with granting the journey the honesty we have always reserved for the destination.