Why transformation stalls: recommendations are not outcomes

A board agrees a transformation. The logic is sound, the market read is right, and the leadership team is aligned on the destination. Twelve months later, the numbers that board actually reviews have not moved.

Very little of that is a strategy failure. The strategy was usually fine. What failed came after it, in the part nobody stayed for.

There is public evidence for this, and it is unusually good evidence, because the UK government publishes what most companies keep private: an independent assessment of whether its largest programmes are likely to deliver. At the end of March 2025, 213 projects in the Government Major Projects Portfolio carried a delivery confidence rating. Thirty of them were rated green. That is 14%. Every one of the other 183 had an approved business case and a plan somebody had signed.

The plan is almost never the thing that fails

A delivery confidence rating is not a judgement on the idea. It is an assessment of whether the thing will land, on time and on cost, if nothing changes. So it separates the two questions most organisations ask as one: is this the right thing to do, and are we going to do it.

The answers come apart. In the portfolio for 2024 to 2025, 15% of rated projects were red and 63% were amber. These are programmes with dedicated teams, external assurance and parliamentary scrutiny. If a plan alone were enough, this is where it would be enough.

The cost of that gap is not theoretical. The whole life cost attached to red rated projects rose from £97 billion to £198 billion in a single year. Nothing about those plans got worse. The assessment of whether they would be delivered did.

The gap is between the plan and the operation

A plan describes a business that does not exist yet. The operation is the business that exists today, complete with every compromise it has accumulated on the way to its current size. Change lands in the second business, not the first.

Ask how a company runs and you are given the structure. Ask a team to walk you through what they actually did last Thursday and you are given the operation. They are rarely the same thing.

This is why a diagnostic can be completely accurate and still change nothing. It describes the first business. The work of transformation is in the second, and it is mostly unglamorous: deciding who owns a number, agreeing what happens when that number moves the wrong way, and staying in the room long enough for the answer to hold.

What people find when they measure management instead of strategy

The Office for National Statistics runs a survey that is unusual in this field, because it measures management practice directly rather than asking executives how they feel about it. The Management and Expectations Survey went to 53,433 businesses between November 2023 and February 2024. It scores four things: continuous improvement, or how a business responds to problems; the use of key performance indicators; the use of targets; and employment practices covering promotion, training and underperformance.

Read that list again. It is a definition of governance, arrived at independently by people with no interest in selling any. None of the four is a strategy question. All four are questions about what happens on an ordinary Tuesday.

The mean score across the UK was 0.55 out of 1, and it varies with size in a way that should interest anybody scaling a business.

The ONS also reports that firms in the bottom decile of management score were the most likely to say they used little or no analysis to support important business decisions. Not the wrong analysis. None.

The international picture points the same way. The World Management Survey, which has scored management practice across firms and countries for close to two decades, finds that management accounts for about a third of the gap in total factor productivity between the United States and other countries in its sample, and as much as half for some. Strategy is not the variable being measured in any of this work. Execution is.

The handover is where the value leaks

Most advisory work is built around a delivery date. A diagnostic is scoped, a recommendation is written, a presentation is given, and the engagement closes. The date the work is handed over is the date the incentive ends.

That is the exact moment the difficulty starts. The recommendation now has to survive contact with an operation that has its own priorities, its own quarter, and nobody whose job depends on the recommendation landing. The people who understood the reasoning have left. The people who have to act on it were not in the room when it was argued.

A recommendation ends at the recommendation. An outcome requires somebody still there when the number is due.

This is not a criticism of advice. Good advice is worth paying for and a company that cannot get a clear read on its own position should buy one. It is a statement about what advice is and is not. A recommendation is a description of a decision. An outcome is a number that changed and stayed changed.

What an outcomes partner does differently

The difference is not method, and anybody selling it as method is selling a diagnostic with a longer contract. The difference is exposure. Somebody stays until the number is due, and they are accountable for it when it arrives.

Three things follow from that, and each of them is visible from outside.

• The operating model changes, not just the strategy. Who decides what, who owns which number, and what happens when it moves. This is the part that outlives the engagement, and it is the part a slide cannot carry.
• Governance is designed in, not bolted on. The four things the ONS measures are the four things that have to exist after everybody leaves. If they are not built during the work, they are not going to appear afterwards.
• The measure is agreed before the work starts. Not activity, not milestones delivered, but the figure the board already reviews. Agreeing it afterwards is how a programme comes to be judged on the things it happened to achieve.

GoFusion was founded by executives with first-hand experience of leading and transforming businesses. As CEO of DDC OS, John Callachan grew the business from 180 people to more than 3,000 across seven countries. EBITDA grew 31% a year, sustained over seven years. He built the leadership, operating model and governance to sustain it. That last sentence is the one that matters here: the scale is the outcome, and the operating model and governance are why it held. You can read the whole record and how we work, or the four practices we deliver against.

The same decision, described two ways

None of this is abstract. Every pair below is one decision, written first as a recommendation states it and then as the outcome it actually implies. The second version is not longer. It is decided, and somebody is named in it.

• Recommendation: consolidate the three regional profit and loss accounts into one. Outcome: one profit and loss account, owned by a named director, reporting on the same cycle the board already meets on, from the third quarter. The two regional directors who lose a line have their new remit agreed before the change is announced, not after.
• Recommendation: improve data quality in the customer system. Outcome: the four fields the revenue forecast depends on are mandatory at entry, one team owns them, and the forecast is rebuilt from that system rather than from a spreadsheet somebody maintains privately.
• Recommendation: introduce a monthly operating review. Outcome: a monthly review with a standing agenda, a named owner for every number on it, and a written rule for what happens when a number misses twice. Without the last part it becomes a meeting where numbers are read aloud.
• Recommendation: move to a shared services model for finance operations. Outcome: the eleven processes that move, the date each one moves, the service level the receiving team commits to, and the person in the business who can stop a migration that is going wrong. The last of those is the one usually left out.
• Recommendation: build automation capability. Outcome: two processes automated end to end this year, chosen because somebody measured how long they take today, with the hours released named and reallocated rather than assumed.

The difference in every pair is the same. The first sentence describes a decision. The second commits a person to a date. Anybody can write the first. The second requires knowing the operation well enough to be wrong in public, which is a different relationship with the work. It is the one described in how we work.

The objections, and what they are worth

We already have a programme office

Most organisations that stall do. A programme office tracks whether the activity in the plan is happening. It is generally not empowered to change the operating model when the plan turns out to be wrong about the business, and that is the decision that governs whether the number moves. Reporting is not the same authority as ownership.

Our own people should own this, not an outside partner

They should, and eventually they must, or nothing is sustainable. The question is what happens in the period before they can. Ownership transfers when the governance exists to hold it, not on the day a contract ends. A partner who leaves before that point has moved the risk rather than reduced it.

This sounds like a longer and more expensive engagement

It is usually longer. Whether it is more expensive depends on what the alternative cost. The portfolio figures above are the answer to that question at national scale: £198 billion of whole life cost now sits behind programmes rated red. A diagnostic that produced no change was not cheap. It was unbilled after the invoice.

How to tell whether it is actually working

Activity is the wrong thing to watch and watching it is how these programmes get defended long after they have stopped working. Agree the measures before the work starts, not after the first review.

• Whether one named person owns each number, and whether they can be asked about it without a meeting being convened.
• What happens the week a number moves the wrong way, and whether that is written down anywhere.
• How long it takes the business to notice a problem, measured from when it started rather than from when it was reported.
• Whether the figure the board reviews is the same figure the programme is run on. If they differ, one of them is decoration.
• Whether the change survives the first quarter in which nobody from outside is in the room. This is the only one that settles the argument.

Questions readers ask

Is this an argument against using consultants?

No. It is an argument about what a recommendation can be expected to do on its own. A firm that needs an accurate, independent read on its position should buy one, and the best of that work is very good. The failure described here happens after the read is delivered, which is a different stage of the work and needs a different arrangement.

The portfolio figures are public sector. Does that transfer to a private business?

The mechanism does, and it is the mechanism that is being quoted rather than the sector. Government publishes independent delivery confidence assessments for programmes that already have plans, funding and scrutiny. Very few private companies produce that assessment at all, which means the gap is not smaller there, it is unmeasured. The ONS survey, which is drawn from private firms, points the same way.

How long before a transformation should show something?

Something should be visible in the operation within a quarter, even when the financial result is a year out. If nothing has changed in how decisions get made after three months, the programme is describing the business rather than changing it.

What is the first thing to look at?

Take the number your board reviewed last month and find the person who owns it. If that takes more than one conversation, you have found the problem, and it is not a strategy problem. Tell us what you are navigating right now and we will tell you what we would look at first.

Sources

• NISTA Annual Report 2024-25, Government Major Projects Portfolio, delivery confidence at 31 March 2025. Contains public sector information licensed under the Open Government Licence v3.0.
• ONS, Management practices in the UK: 2016 to 2023, Management and Expectations Survey of 53,433 businesses
• Scur, Lemos, Sadun, Bloom and Van Reenen, The World Management Survey at 18, NBER Working Paper 28524