Across twenty five programmes in nine countries and three sectors, spanning twenty four years and budgets that differ by a factor of a thousand, the same five pieces of change work were missing every time, and this page sets out what those five are, which programmes did and did not do them, and where the risk actually sat in each case.
You do not need to know the framework to read this page. Each part opens by explaining what you are looking at.
The Pattern
The same five things were missing every time.
We took twenty five large programmes, from the NHS to Ford to Aadhaar, and asked one question of each. Before the money was committed, did anybody do the five pieces of work that decide whether people can actually carry a change?
The programmes have almost nothing in common. Different countries, different sectors, different decades, and budgets that differ by a factor of a thousand. What they have in common is what was skipped.
Part one
What should have happened, and what did.
Read each row across. On the left is one of the five checks MOSAIC makes before a commitment becomes irreversible, and which part of the framework it belongs to. On the right is what happened instead, with a real example, and how many of the 25 programmes skipped it.
What should have happenedThe five things MOSAIC checks before money is committed
What actually happenedTaken from the published record of each programme
01Count what the organisation is already carrying, before adding anything to it.Belongs to Anchor PointDid anyone establish the change the affected population was already absorbing?
The new programme was planned as though the organisation were otherwise idle.HMRC moved 170 offices into 13 while fourteen other major programmes ran on the same workforce. Nobody added them together.20 of the 25 programmes did not do this, and for 1 the public record does not say
02Evidence leadership alignment. Do not accept agreement in a room as a position.Belongs to Anchor Point and FuseWas alignment tested against something, or inferred from the absence of objection?
A nod was recorded as agreement and the programme proceeded on it.At NS&I the word deliverable carried three incompatible meanings across three parties, and that surfaced years after commitment.16 of the 25 programmes did not do this, and for 1 the public record does not say
03Locate the culture gap. Name where it sits and what it will cost to close.Belongs to FuseWas the gap named as a specific, located practice, or listed as a generic risk?
Culture appeared as a line on a risk register and never as a decision.Lidl found the business valued stock at purchase price and the software at retail price. That is a question about changing a core practice. It was treated as a configuration setting.18 of the 25 programmes did not do this, and for 6 the public record does not say
04Activate the managers who carry it. Equipping is not the same as informing.Belongs to PitchWere managers given something to do differently, and was it checked that they did it?
Managers were briefed once and then left to carry it unaided.Klarna removed the escalation route to a person, which disabled the exact layer that would have surfaced the quality problem.12 of the 25 programmes did not do this, and for 12 the public record does not say
05Define adoption before go live. State what using it properly will look like.Belongs to Shape and LoopWas a measure of real adoption agreed in advance of the commitment?
Adoption was defined after go live, usually as whatever the available data could show.The NHS business case assumed time saved. Clinicians reported time lost. The Gates evaluation measured whether the system was implemented, not whether outcomes moved.21 of the 25 programmes did not do this
Part two
Which programmes did what.
Every programme, against every check. One row per programme, one column per check. We scored these from published audits, parliamentary committees, regulators and independent evaluations only. Where the record does not tell us, we say so and leave the square open rather than guessing.
Gold block. They did not do itGreen circle. They did do itHatched. The record does not say, so we left it open
Programme
Counted the load
Evidenced alignment
Located the culture gap
Activated managers
Defined adoption
NHS National ProgrammePublic
London 2012Public
Universal CreditPublic
NS&IPublic
Birmingham City CouncilPublic
US Veterans AffairsPublic
HMRC LocationsPublic
Healthcare.govPublic
Operation Warp SpeedPublic
India GST 2017Public
AadhaarPublic
NEOM, The LinePublic
UAE agentic AIPublic
MalaffiPublic
Qatar, Education for a New EraPublic
TSB migrationPrivate
Zillow OffersPrivate
Ford, One FordPrivate
Lidl and SAPPrivate
Revlon SAPPrivate
Klarna AIPrivate
Mizuho BankPrivate
Kids CompanyNonprofit
Intensive PartnershipsNonprofit
Etihad equity alliancePrivate
Did not do it
20of 25
16of 25
18of 25
12of 25
21of 25
Part three
How much could go wrong, and did anyone check?
Two questions decide how dangerous a change really is. First, if it goes wrong, how much goes wrong at the same time? Rolling out to one office is a very different risk from switching the whole country on one date. Second, before committing, did anybody actually check whether the people affected could cope with it? Every read sits somewhere on those two questions. There are twenty six dots for twenty five programmes, because London 2012 is read twice, once for the build and once for the security workforce.
How to read it. Each dot is one programme. The further right, the more that would break at once if it went wrong. The further up, the more somebody had actually checked whether people could cope before committing. The number at the top of each square is how many programmes sit there. Hover any dot to see which programme it is.
Green Somebody checked whether people could cope, before the money was committedGrey They checked for one group of people and assumed it for anotherGold Nobody checked at all
London 2012 appears twice. The construction of the venues sits top left. The security staffing sits bottom right. Same programme, same leadership, two completely different positions.
The first test costs nothing
If you cannot answer this from data you already hold, you have your answer.
Who does this change land on, and what are those people already carrying?
An organisation that cannot answer that today, from data it already has, has not failed the first check during delivery. It failed it before the programme was approved. The absence of the data is not an administrative gap to be filled later. It is the finding.
Six of the seven programmes we ran in full could not answer it. Neither could the public record, for exactly the same reason. Nobody had collected it.
This is not only for programmes that have not started
Every programme has a next irreversible commitment.
The five checks can be made at any point in a programme's life, including one you have inherited from somebody else, because they are checked against evidence that already exists. Nobody has to have been there at the start to establish that the load was never counted or that adoption was never defined.
The next go live. The next region. The next tranche of funding. The next contract extension. The read is always taken before the next money moves.
The date gets fixed first. The business case is written to fit the date. So the assumption about whether anyone can absorb the change becomes whatever the date requires.
That assumption is the single largest line in the case, and it is the only one nobody tests.