McKinsey and ICE: Who Carries the Cost of a Saving?

Cost optimisation begins with the service obligation

An efficiency project needs a definition of what must remain true after costs fall. In a detention setting, the baseline includes legal duties, healthcare, food, safety and the welfare of people who cannot choose another provider. Those constraints are not secondary measures to be considered once the financial recommendation has been selected.

The consulting question is therefore broader than which bed, meal or service costs least. It asks which operating arrangement fulfils the institution’s obligations at a justified cost. A model that omits a mandatory standard has not found an efficient solution; it has changed the problem being solved.

This is an analytical principle, not a finding that every recommendation in the historical engagement breached a standard. The reporting, the firm’s denial and ICE’s response need to be considered together. The article preserves the distinction between the disputed characterisation of advice and evidence that a particular policy was adopted.

A recommendation is not an implementation record

Internal slides can show what a team explored or proposed. They do not necessarily show what the client accepted, how the idea was modified, whether it was implemented or what happened afterward. Those are separate stages in the decision chain. An investigative timeline should identify the evidence available for each one.

For a present-day programme, keep a decision register with the recommendation, the official who approved it, any conditions, the implementation owner and the observed result. If a proposal is rejected, record that too. Otherwise a later review may mistake a list of options for a record of policy.

This proposed framework is useful in any operating-review deck. It becomes particularly important where a decision affects people who have limited power to challenge its consequences.

The fee-versus-savings comparison needs aligned scope

The documentary places reported consulting fees beside reported detention-contract savings. That comparison raises a fair question about value for money, but it cannot by itself answer it. The fees cover a wider set of work than the specific savings measure. The periods, implementation costs and attribution may also differ.

A valid return-on-investment calculation would identify the work included, the period of realised benefit and the cost of implementing the change. It would exclude benefits that would have occurred anyway and explain uncertainty. It would also account for service outcomes that the financial measure does not capture.

The article therefore treats the figures as contextual measures with different scopes. It does not subtract one from the other and call the result the project’s net value. A business-case model should enforce those definitions before a waterfall chart is used to persuade a committee.

A cost baseline can be technically fragile

The GAO’s detention-cost review provides a separate government analysis of estimation issues. It examined how bed rates and population assumptions affected budget estimates. That review should not be presented as a verdict on every consulting recommendation, but it explains why baseline quality matters.

If the starting estimate is unreliable, a claimed saving may reflect a modelling correction rather than an operating improvement. If population assumptions change, lower total expenditure may occur without lower cost per person. Conversely, a higher service standard can increase a unit rate while improving the institution’s performance against its mandate.

For a useful exhibit, show volume, unit cost and service scope separately. State whether the figure is a budget request, an obligation, an invoice or an observed expense. The labels are not technical decoration. They tell the reader what inference the number can support.

An illustrative standards-constrained model

Imagine a fictional institution comparing two facilities. Facility A has a lower daily price but requires additional transport, provides less access to required services and has a different capacity profile. Facility B has a higher quoted rate but avoids some of those costs. Choosing between them requires a complete service and cost definition.

The model should include transport, staffing, healthcare access, capacity and every applicable standard. Some criteria may be pass-or-fail rather than tradable against price. The approving official should see which options meet the requirements before the team ranks their cost.

These facilities and assumptions are illustrative. They are not claims about a particular ICE site. The exercise demonstrates why a recommendation to use a lower-cost location is incomplete until service equivalence and the full operating cost are established. The Allstate claims case offers another example of why a favourable financial measure needs an independent service-quality test.

Scope needs review when the mission changes

The work began before the change in presidential administration. That history is relevant, but it does not settle the scope or risk of every subsequent activity. A continuing contract can encounter new priorities, new instructions and a different operational environment. The engagement review should examine those changes directly.

For consulting firms, an escalation process should let staff raise a concern about a changed assignment without having to prove misconduct. The approving authority should assess the actual work, the institution’s mandate and the firm’s own commitments. A broad statement of values becomes useful only when someone can use it to make a concrete decision.

For public buyers, contract modifications should have a documented rationale and independent approval. A supplier may contribute expertise to defining the next phase, but the institution must remain responsible for deciding what it needs and whether the procurement route is justified. The France case examines that ownership of public advice.

Present the contested account fairly

ProPublica defended its document-based reporting after McKinsey challenged it. The firm’s denial of recommending lower food or healthcare quality belongs close to the relevant allegation. ICE’s stated view of the work also belongs in the account. A reader should not have to reach the source list to discover that the characterisation is disputed.

At the same time, displaying a denial does not remove the need to examine the documents. The article can explain the contradiction and the evidence on each side without pretending to resolve every disagreement. It should also avoid unsupported universal statements about the absence of all lawsuits or later proceedings.

The final strategy recommendation for a comparable project should show the saving, its evidence, implementation status and the people affected. If the team cannot state who carries the consequences and how mandatory standards are protected, the work is not ready for approval.