When the repeatable half of a business is contracted out, what remains is a portfolio of temporary undertakings — and temporary undertakings have a failure profile the business case rarely assumes.
Ask a chief executive what their organisation does and you will usually get an answer in the present continuous: we make, we build, we serve, we treat, we ship. Ask the same executive how much of the organisation's work is now organised as discrete, funded, time-bounded initiatives with a start, a finish and a nominated leader, and the answer is frequently more than half. Those two answers describe different businesses.
The gap between them is not a communication problem. It is a strategic one, because the two businesses have different economics. A business that repeats gets better at what it repeats and can defend the improvement. A business that runs projects gets better only if it deliberately builds a mechanism to carry learning from one temporary undertaking to the next — and most do not, because nobody is funded to.
This is not a new observation. In June 2005, The Economist documented a set of firms that had made the shift consciously and described what they had become. Two decades on, the cases are historical, but the structural point they illustrate has not aged, and the decision they pose is one many boards have made without noticing that they were making it.
The Strategic Context
The 2005 report described several companies that had moved decisively toward co-ordination and away from production. A global footwear brand, it noted, no longer made shoes; it managed footwear projects. A global beverage company had handed most of the bottling and marketing of its drinks to others and had become, in the article's characterisation, largely a collection of projects run by people it called orchestrators. A German car maker treated each new vehicle platform — the basis of a whole range — as a separate project. A fast-growing American financial services group maintained a standing team to handle acquisitions. For all of them, the report concluded, project management had become a competitive instrument, and some described it as a core competence.
The scale claims in the same report are worth reading as period evidence rather than current fact. A survey by the Project Management Institute found three in four European companies employed project managers. One computer maker carried roughly 1,400 of them on its payroll when it was acquired in 2002. The institute itself reported 150,000 members across 150 countries. All of that is twenty-one years old and none of it should be quoted as a present figure. [FACT CHECK REQUIRED]
What has not dated is the structural claim underneath. Two of the cases are specific enough to name, because the specifics carry the argument. BP transformed the fortunes of its exploration division by converting it into a portfolio of projects, each largely free of head-office control, in a structure the company itself described as an "asset federation" — and the report is explicit about the consequence: asset and project managers could no longer rely on head office for support, and were required to build their own self-sufficient teams. Siemens, having worked out that half its turnover came from project-like work, launched a worldwide initiative to improve its project management and calculated that completing all of that work on time and to budget would add €3 billion to its bottom line over three years. That figure is a company's calculation of a hypothetical, not a result it reported achieving. [FACT CHECK REQUIRED]
What the Project Count Conceals
The common misreading is to treat a rising project count as evidence of discipline. It is usually evidence of something else: that work which used to be absorbed by a standing function is now being packaged, funded and staffed one instance at a time.
That repackaging is often the right call. It surfaces cost, forces a decision, and gives someone a name against the outcome. But it also does three things that rarely appear in the paper approving it.
It converts a capability into a series of engagements. A function that ran continuously accumulated judgement — about suppliers, about failure modes, about which estimates to distrust. A sequence of projects accumulates that only if something outside the projects is built to hold it.
It converts a fixed cost into a variable one, which looks like flexibility on the way in and like a lost floor on the way out. The floor was doing work: it was absorbing demand variability without a negotiation.
And it converts a known performance distribution into a less favourable one. Project work is, on the available evidence, less reliable than repeated work — the same 2005 report cites research finding that fewer than a third of information technology projects succeeded on their researcher's own definition, with substantial average overruns on both cost and time. [FACT CHECK REQUIRED] An enterprise that moves a line of activity into project form inherits that distribution. Very few business cases say so. [Related article: The Estimating Loop Nobody Closes]
Reframing the Issue
The question is not whether to run projects. Every enterprise runs projects. The question is which half of the business the enterprise is actually paid for, and whether that half is the one it kept.
Nick Lavingia, writing in Cost Engineering in 2003 from an engineering rather than a strategy tradition, put the stake in the bluntest available terms: a company that consistently selects the right projects and executes them well improves its return on capital employed and, ultimately, total shareholder return — and in his framing, the difference between doing that well and doing it badly is the difference between a profitable company and one that becomes a takeover target. He was arguing for better project management. The prior question is what proportion of enterprise value now depends on it.
For an organisation whose remaining work is co-ordination, that proportion approaches all of it. That is a defensible position — orchestration is a real capability, and it is scarce — but it is a different business from the one the annual report usually describes, and it should be governed as one.
Two Halves of a Business, and Only One of Them Compounds
Consider a trade book publisher, hypothetically, that has outsourced printing, warehousing and distribution over fifteen years. Each decision was individually correct: specialist printers ran at lower unit cost, and the capital released went into acquisitions and marketing. What remains is commissioning, editing, design and rights — and every one of those is a project. Each title is funded separately, has a nominated editor, runs to a date, and either works or does not.
The publisher's economics have changed shape entirely. There is no longer a volume effect: printing a hundred thousand more copies improves the printer's position, not the publisher's. The compounding asset is now judgement about which books to acquire and at what advance — a portfolio selection capability, exercised title by title. If the publisher does not treat that capability as the thing it is investing in, it will invest in marketing instead, which is the visible half.
The same structure appears in a very different setting. A touring theatre company, hypothetically, that contracts production, venue and technical crew for each season retains only programming, casting and the relationships that make the next contract available. It has no back catalogue in the operational sense. Its balance sheet is thin by design. What it owns is a judgement about which work will find an audience, and a reputation that makes good work available to it.
In both cases the durable asset is a selection capability, and in both cases the organisation is structurally tempted to under-invest in it, because selection is cheap to do badly and expensive to do well. [Related article: Is Your Portfolio Function Selecting, or Supervising?]
What the Model Imports
Converting a business to orchestration imports four things that are easy to miss at the point of decision.
A dependency you cannot inspect from the outside. The contracted party now holds process knowledge the enterprise used to hold. Whether that matters depends on how substitutable they are, which is a question about the market, not about the contract.
A handover at the end of every undertaking. Temporary structures always end, and something must receive what they produced. [Related article: The Hidden Cost of Putting Work Into Project Form]
A capability question the training budget does not answer. Investment in method, tools and certification splits into what becomes an organisational asset and what leaves in an individual's curriculum vitae, and the split is rarely examined. [Related article: What Does the Enterprise Own After a Capability Investment?]
An arithmetic problem in resourcing. A portfolio of concurrent projects staffed from a shared pool produces a capacity statement that is frequently fictional. [Related article: There Is No Such Thing as Half a Project Manager]
There is a fifth import that deserves separate attention, because it is the one most likely to be inherited unexamined. The instruments an orchestrating enterprise uses — the schedules, the stage gates, the contract forms — were built for large, contract-heavy, specification-stable work. They carry that inheritance into settings that share none of those properties. [Related article: What Kind of Work Were These Instruments Built For?]
Decision Framework
Three tests, applied line of activity by line of activity rather than to the enterprise as a whole. Most organisations are producers in some lines and orchestrators in others, and the mixed answer is the useful one.
The repetition test. Would doing this a hundred times make us materially better at it? If yes, it is repeated work, and the value is in the accumulation. Then ask the harder half: does anything in the current structure actually capture that accumulation, or does each instance start from the same place? An affirmative answer to the first question and a negative answer to the second is the most expensive combination available, because the enterprise is carrying the cost of repetition without collecting the return.
The margin test. For this line, what is the customer paying a premium for — the making or the orchestrating? Then: which half did we keep? Where the premium sits with the making and the making has been contracted out, the enterprise is holding the lower-margin half of its own value chain and calling it a strategy.
The failure-rate test. If this activity now runs in project form, its performance distribution is the project distribution: less predictable, with a longer tail. Has the business case been rewritten to assume that distribution, or does it still assume the reliability of a standing function? This test is usually failed silently, because the case was written before the conversion and never revisited.
A fourth question sits underneath all three and is not a test so much as a discipline: what, specifically, would still be possible next Monday if the largest contracted party stopped work? The answer is the residual capability, and it is the only honest measure of how much of the business the enterprise still holds.
From Strategy to Execution
Immediate. Classify the enterprise's major lines of activity as repeated or project, and record the classification where capital decisions are made. This is a half-day exercise and it is routinely revealing, because the classification is often contested inside the leadership team — which is itself the finding.
Medium-term. Build the mechanism that carries learning between undertakings, and fund it from a standing line rather than from the projects themselves. A project has no incentive to pay for the next project's estimate to be better, and it should not be asked to. Then decide, explicitly, which selection capability the enterprise is investing in — which acquisitions, which commissions, which bids — and whether the people exercising it have the standing to decline.
Long-term. Treat the producer-to-orchestrator boundary as a position to be held rather than a transition to be completed. The instinct at every review is to outsource one more thing, because each individual case looks marginal. The cumulative effect is not marginal, and there is no natural stopping point that the individual cases will supply. The stopping point has to be a deliberate statement about which capabilities the enterprise will not contract out at any price.
Signals to Monitor
- The proportion of revenue attributable to work organised as projects, tracked over time rather than measured once.
- Whether estimates for new undertakings are demonstrably informed by completed ones — a question of evidence, not of process description.
- Concentration: the share of project delivery dependent on a single contracted party, and whether that share is rising.
- Whether the people exercising selection judgement — commissioning, bidding, acquiring — are being promoted for volume or for outcome.
- Contracted parties beginning to offer the co-ordination service itself, which is the signal that the orchestrating position is being competed for from below.
Questions for the Leadership Team
- For each of our three largest lines of activity, are we the producer or the orchestrator — and would our people give the same answer?
- Which half of each line does the customer pay a premium for, and did we keep it?
- What is the largest capability we have contracted out in the past five years, and what would we be unable to do next Monday if that party stopped?
- Where in our structure does learning from a completed project reach the next estimate, and who pays for that to happen?
- Have the business cases for our project-form activities been rewritten to assume project-form reliability?
- What is the one capability we are prepared to state, on the record, that we will not outsource at any price — and why that one?
Closing Perspective
The shift from producing to orchestrating is rarely a decision. It is a sequence of individually sensible outsourcings, each approved on its own numbers, that eventually changes what the enterprise is. Some organisations have made the shift deliberately and manage the consequences well; BP's asset federation was a designed structure with a stated cost, not an accident.
The failure is not orchestration. It is arriving at orchestration while still governing as a producer — assuming a reliability the model no longer has, funding a learning mechanism nobody owns, and describing to the board a business that stopped existing several years ago. The useful discipline is the plainest one available: say out loud, line by line, what business this is now. [Related article: The Estimate Was Made by the People Who Needed to Win] [Related article: Value Management Is Not Cost Reduction]
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.