Model Your Future State

Few decisions shape an organisation for longer than a change to its operating model. Moving work between teams, consolidating into a shared service, shifting activity offshore or integrating an acquisition can fix cost and service levels for years.

Yet these decisions are routinely approved on the strength of a target-state diagram and a spreadsheet of projected savings. The design is described. It is rarely tested.

The track record

The evidence on redesigns is sobering. McKinsey research found that fewer than a quarter of organisational redesigns succeed. Forty-four per cent run out of steam once under way, and a third fail to meet their objectives or improve performance after implementation [1]. An earlier Bain study of 57 reorganisations between 2000 and 2006 found that fewer than a third produced any meaningful improvement in performance [2].

The same McKinsey work points to a cause. Many companies, the authors note, choose a new structure on the basis of untested hypotheses or intuition. Almost four in five respondents who admitted relying on gut feel said their chosen blueprint had not succeeded. Sixty per cent said they had not spent enough time assessing the organisation before redesigning it [1].

Why assumptions go untested

The tendency is human, and well documented. Daniel Kahneman and Dan Lovallo described how executives consistently overestimate benefits and underestimate costs and time, because they plan from the inside view of their own project rather than the record of similar ones [3].

UK public spending rules build this in. HM Treasury’s Green Book guidance states that project appraisers tend to be over-optimistic, and requires explicit adjustments to estimates of cost, benefit and duration, based on data from past or similar projects [4].

The scale of the gap can be large. Bent Flyvbjerg’s database of more than 16,000 projects across many sectors shows that only about 8.5 per cent are delivered on budget and on time. Only 0.5 per cent achieve that and also deliver the benefits promised [5].

Operating model changes are not megaprojects, but they share the same exposure. A business case is a forecast, and a forecast built on untested assumptions is a hope with a number attached.

The assumptions that matter most are hidden in the flow of work

The riskiest assumptions in an operating model change are rarely about headcount. They are about how work moves.

• Availability. A task routed to a team in another region waits until that team is working. A spreadsheet that treats every hour as available will understate elapsed time.
• Utilisation. Queueing theory shows that as a resource approaches full utilisation, waiting time rises sharply rather than gradually [6]. A consolidation that runs a team at 95 per cent looks efficient on paper and slow in practice.
• Variation. Real processes branch. A share of cases needs rework, escalation or a second approval. Averages hide the tail that drives cost and complaints.
• Collective work. Many steps are done by groups, not individuals: a panel review, a batch approval, a shared queue. Modelling them as one person doing one task misstates both time and cost.

None of these is visible on an organisation chart. All of them can turn a projected saving into a service failure.

Testing before committing

The discipline to fix this is not new. Rita McGrath and Ian MacMillan’s discovery-driven planning, first published in Harvard Business Review in 1995, asks managers to write down the assumptions a plan depends on and to convert them into knowledge as cheaply as possible before committing major resources [7].

For operational change, the most direct way to test assumptions is simulation. Business process management research has long used simulation to run a process model many times with realistic durations, branching probabilities and resource constraints, and to compare the outcomes of alternative designs before any are implemented [8]. The idea now has an analyst category. In July 2026 Gartner published its first Magic Quadrant for Digital Twin of an Organization Platforms, describing a dynamic model that helps enterprises plan initiatives with many interdependencies [9].

A future state that can be tested has five properties.

1. A current-state baseline drawn as processes beneath the capabilities they serve.
2. Future-state variations that change specific things: who does the work, where, and in what order.
3. Realistic resources: people per team, working calendars by region and a cost rate per hour.
4. Comparable outputs: time, cost and efficiency for current and future states, run under the same conditions.
5. A link to funding, so the change portfolio is approved against the tested result rather than the original estimate.

With those in place, the board conversation changes. The question is no longer whether the target state looks right. It is which of three tested options performs best, and which assumptions still carry the most risk.

Where MIRAI fits

MIRAI is built to make future states testable. Current and future process variations sit side by side. Simulation runs a process many times with durations, branch probabilities, people per lane, working calendars by region and a rate card for each lane. It also models group work, so a task can be done by a team rather than one person. Results roll up from the process to the capability and the organisation map.

The same model holds the change portfolio: goals, funding rounds, approvers and time points. Funding is tied to the capabilities the change affects, so “what it is now” and “what it will be” come from one source. The simulation features have been built and tested on sample processes.

Questions to ask before your next operating model decision

• Which three assumptions would most change the business case if they were wrong?
• Has the future state been run against real working hours, real capacity and real variation?
• Can we compare at least two alternative designs on the same basis?
• Is the funding tied to a tested result, or to the original estimate?