Digital transformation cases are programs inside an incumbent: value is adoption times process change, not a go-live date. They are not “value a software company” and not “price a streaming bundle.”
Last full pass: 24 August 2026.
Firms staff these cases because clients spend nine-figure sums on cores, data platforms, and “AI.” The interview is a benefits case, a build-vs-buy, and a change-management test in one. McKinsey, BCG, Bain, Accenture Strategy, and the Big Four all run them. If the client is the software vendor, use the technology case interview. If the client is a telco or streamer, use TMT. Here the client is a bank, insurer, hospital system, or manufacturer asking whether a program is worth running — and why last year’s program did not move the P&L.
What you are deciding
- Go / no-go / rescope a program given cost, time, and realized benefits.
- Build vs buy vs configure (package vs custom vs BPO).
- Sequence: which process first, which country, which product line — not “agile everywhere.”
- Operating model: who owns the KPI after the consultants leave.
Clarify the outcome metric in operations language: claims cycle time, first-time-right, cost per policy, on-time-in-full — not “digital maturity.” Clarify what is already spent (sunk) vs remaining.
Exhibits you should expect
| Exhibit | How to use it |
|---|---|
| Program cost (internal + SI + licenses) by year | Cash, not capitalized fairy dust. Include internal FTEs. |
| Benefits register | Must be owned, timed, and tied to a process KPI. “10% efficiency” is not a number. |
| Adoption / % volume on the new journey | Go-live ≠ benefits. 35% of volume on the new stack earns 35% of the benefit. |
| Current-state cost stack | Baseline. If claims ops is $220m, an 18% cut is $40m at full adoption. |
| Risk / regulatory calendar | A core replacement that misses a reporting year is a non-option. |
Traps that are specific to this case type
IT spend vs value. A $160m program is not “innovation.” It is an investment with an NPV, like the finance guide, plus an adoption discount.
Licenses vs outcomes. Seat counts of a new CRM can rise while win-rates fall.
Double-counting benefits. Procurement savings and “FTE efficiency” often describe the same people.
Shadow IT and parallel run. You pay for two systems until cutover. Parallel-run cost belongs in the case.
Practice a transformation-program case
Run a scored case where adoption, not go-live, drives the benefit math.
Worked mini-case: go-live is not a business case
Prompt. MutualSure will replace its claims core. Remaining spend $160m over three years (case input; treat as cash). At full adoption, claims-handling cost of $220m would fall 18% → $39.6m per year. Year-3 exhibit: only 35% of claims volume is on the new system (the rest is a parallel run). Run-rate IT to keep the old core is $14m/year until it is switched off. Is the program on track to pay back?
Math. Realized opex benefit at year 3 = 0.35 × 39.6 = $13.9m. You still pay $14m to run the legacy core. Net operating benefit ≈ −$0.1m in year 3 while the program is still burning tens of millions. Even if you ignore legacy dual-run: $13.9m vs ~$53m/year amortized ($160m/3) is not a payback story.
To earn the full $39.6m you need volume and process change (straight-through processing, not a new screen for the same desktop procedure). If process change only happens for the 35%, you never get 18% of $220m.
Recommendation. Stop talking go-live. Freeze remaining scope to the two claim types that are 60% of cost, fund an adoption workstream (incentives, decommission dates, no dual keying), and set a hard date to switch off legacy for those types. If adoption cannot clear 80% in 12 months, pause remaining spend — the NPV is negative at 35%. Risk: regulators still need the old extracts; that is a conversion project, not a reason to keep 100% dual-run. Next exhibit: claims cost by type and % STP on the new stack.
What a generic profitability tree misses here
- Benefits are adoption × process delta, not system existence.
- Cost is a multi-year cash profile with internal people, not a one-line “IT opex.”
- Two systems at once is the default, and it can cancel the benefit.
- Political sequencing (which business line goes first) determines whether the 18% is real.
- Build vs buy changes time-to-adoption more than it changes license line items.
If you treat this as a technology-industry case, you will talk ARR. If you treat it as TMT, you will talk ARPU. MutualSure is a program, and the unit is percent of volume converted.
See where you stand on a transformation case
Practice benefits-realization math before a Big Four or MBB digital round.
Related guides
CoachNed is independent and unaffiliated with McKinsey, BCG, Bain, or other firms named for interview-style practice. Cases on CoachNed are AI-simulated.