The Commissioning Trap: How to Buy Transformation Without Buying Dependency
You can outsource the doing and still keep the knowing.
The lunch that measured the wrong thing
The programme was, by every measure on the steering-committee pack, a success. It had gone live on the revised date, inside the revised budget, and there had already been a celebration lunch with the sponsor thanking the integrator’s partner by name. Three months later the delivery team — some forty consultants who had lived in the building for the best part of two years — rolled off to their next engagement. Within a fortnight the organisation discovered something it had not thought to check: it no longer understood how its own core process worked.
The people who knew why the new system had been designed the way it was, which compromises had been made and why, where the awkward workarounds lived — all of them had left, wearing a lanyard that was not the organisation’s own. What remained was a working system and a hollow where the understanding used to be. To change a field, fix a defect, or answer a regulator’s question about how a particular figure was calculated now meant going back to the firm that had built it, at the firm’s day rate, on the firm’s timetable.
This is not a story about a bad supplier. The supplier did precisely what it had been asked to do, and did it well. It is a story about how the organisation bought — and about a truth we are strangely reluctant to state plainly: dependency of this kind is rarely an accident. It is manufactured, and it is mostly manufactured by the buyer, in the way the work is commissioned.
Capacity is not the same as judgement
The prevailing wisdom of the moment is sound as far as it goes. Concentrate on what you do distinctively well; buy in the rest from those who do it at scale. The outsourcing wave now cresting across the sector — the managed-service deals, the offshore development centres, the multi-year application-support contracts — rests on that logic, and much of it has delivered real, bankable savings. I have no quarrel with the principle.
The trouble is that the principle quietly conflates two very different things: outsourcing capacity, and outsourcing judgement. When you buy capacity, you are hiring hands to do work you understand and could, at a pinch, do yourself. When you outsource judgement, you hand over the deciding — what the requirement really is, how the thing should be shaped, what “good” looks like — to people whose commercial interest does not perfectly align with yours. You can outsource the doing and still keep the knowing. Outsource the deciding and you lose both, usually without noticing until it is too late to matter.
The test of any commissioning decision is not “will this be delivered?” It is “when it has been delivered, will we be more able to run and change it ourselves — or less?”
Transformation is the worst possible thing to buy as if it were a commodity, because the very act of transforming rewires the knowledge the organisation depends on. The integrator arrives with method, with people, and with a plan at exactly the moment the client is least sure of itself — mid-restructure, under an efficiency mandate, with its own best people already stretched thin. The temptation, then, is to hand over not just the hands but the head. That is the commissioning error from which most manufactured dependency flows.
How dependency is actually built
Once you look for it, the mechanism is depressingly consistent. Dependency is not one decision; it is an accumulation of small commissioning choices, each defensible on its own, that together leave the buyer captive.
- Knowledge asymmetry by design. The design rationale ends up living in the supplier’s staff and the supplier’s documents. The client signs off deliverables it cannot itself reconstruct. When the team rolls off, the reasoning leaves with them.
- The missing intelligent customer. No one is left on the client side who can specify precisely, challenge an estimate, or read the technical design well enough to know when it is being sold a poor one. The buyer loses the ability to be a discerning buyer.
- The renewal ratchet. At the first support renewal there is no credible alternative, because no one else could take the system on without a costly discovery exercise the client would have to fund. The supplier knows this. I have watched a maintenance rate move from roughly £650 to over £900 a day at a first renewal — a rise of around forty per cent — for identical work, simply because the client had no exit and both parties knew it.
- Specification by the supplier. The client, short of capacity, lets the integrator write the requirements it will then be paid to meet. The requirement is optimised, subtly and rationally, for the person holding the pen.
None of these requires bad faith. Each is the natural result of buying delivery while neglecting to buy — or retain — capability.
The strongest case for the other side
It would be too easy to turn this into an argument against outsourcing, and I want to resist that, because the opposing case is a serious one and deserves its strongest form.
It runs like this. Trying to retain everything in-house is how organisations end up with bloated, inward-looking technology functions that are expensive, slow, and mediocre precisely because they never face competition. The large integrators and the offshore providers have scale, method, and a depth of specialist talent no single organisation can hope to keep on its own payroll, idle between programmes. Specialisation is the whole basis of the modern economy; insisting on doing your own transformation is as quaint as insisting on generating your own electricity. And the savings are not theoretical — the last few years have shown they are real. On this view, a little dependency is simply the price of access to capability you could not otherwise afford, and fretting about it is sentiment dressed as strategy.
That case is largely right, and it is why the answer is not to outsource less. The answer is that the case is an argument about how much to do in-house, when the real determinant of dependency is how you commission — and those are not the same question at all. You can outsource almost all of the doing and remain a sovereign, intelligent customer. You can also outsource very little and still be captive, if you have surrendered judgement over the part that matters. The scale of outsourcing and the degree of dependency are far more loosely coupled than the debate assumes. What couples them, or decouples them, is the commissioning.
Commissioning as if capability mattered
So what does it look like to buy transformation without buying dependency? Not a procurement template — the disciplines are few, and they are matters of intent more than of process.
The first is to retain an intelligent-customer core, however small. You do not need to keep the whole capability in-house; you need to keep the handful of roles that let you remain a discerning buyer — someone who owns the target design, a process owner who can say what the business actually needs, a technical lead who can read a design document and tell you when it is being gamed. This is the cheapest insurance an organisation can buy and the first thing an efficiency drive tends to cut, because it looks like overhead until the day you need it and it is not there.
The second is to treat knowledge transfer as a deliverable with teeth. In most contracts it is a closing slide and a folder of documents nobody can act on. Make the acceptance test operational: the work is not accepted until your people, not the supplier’s, can run and change it under observation. Knowledge transfer that is not tested is knowledge transfer that did not happen.
The third is to design the exit at the entrance. Ownership of the code, the data, and the intellectual property; documentation to a standard you have specified; source code in escrow; a realistic second source. Settle all of it before signing, when you still have leverage, not at renewal when you have none.
“If you cannot leave a supplier, you are not a customer — you are a captive.”
The fourth is to commission in stages you can actually stop. A transformation bought as one indivisible, multi-year lump is a transformation you cannot course-correct without detonating the whole contract. Break it into commissioned stages with genuine decision points between them — points at which you can change direction, change supplier, or stop — and price them so that stopping is a real option rather than a ruinous one.
The contrast is not abstract. Consider two organisations doing much the same programme with much the same integrator. The first kept a design authority of three people and insisted on operational acceptance before sign-off; a year after go-live it was making its own changes at a fraction of the integrator’s rate and negotiating its renewal from a position of choice. The second kept no one, accepted delivery on the strength of a demonstration, and a year on was paying three to four times as much for every change and waiting months for each one. Same supplier, same technology, same era. The only material difference was in how each had commissioned.
What the steering committee should have asked
The celebration lunch was not wrong to mark the go-live. It was wrong to treat the go-live as the answer to the only question that had been asked. “Did we deliver?” is the question of a project. “Are we now more capable of standing on our own than we were before?” is the question of an organisation that intends to keep transforming, and to do so as an owner rather than a tenant of its own operations.
We are, as a profession, fluent in the mechanics of procurement — the evaluation matrices, the service levels, the penalty clauses — and far less fluent in commissioning for our own future capability. The mechanics protect us against a supplier that fails to deliver. They do nothing to protect us against a supplier that delivers beautifully and leaves us unable to live without it. That second risk is the larger one, and it is entirely within the buyer’s gift to manage. It is decided not on the day the work is delivered, but on the day, long before, when we decide how to buy it.