Most candidates treat a digital transformation case like a strategy slogan: list benefits, wave at "agile," recommend a roadmap. That is how you lose the round. The interviewer is scoring whether you treat the program as a multi-year cash flow with adoption risk baked into every benefit line.
By the end of this piece you will price a transformation the way a partner prices a client memo: remaining cash out, dual-run cost, volume on the new stack, and a discounted net present value under more than one adoption path. Firm marketing slides are optional. Arithmetic is not.
What these cases actually are
The client is almost never a software vendor. It is a bank, insurer, hospital system, retailer, or manufacturer asking whether a core replacement, data platform, or "AI-enabled" process redesign is worth finishing. If the prompt is about SaaS unit economics or platform take rates, switch to a technology case. If it is telco ARPU or media ads, use TMT. Here the unit of analysis is percent of volume converted and dollars of process cost removed.
Public research is blunt about how often the board case arrives intact. BCG's 2020 study put only about 30 percent of digital transformations in the "win zone" of meeting target value with sustainable change; a 2021 update across 850-plus companies raised that only to 35 percent. McKinsey's 2022 survey found medians of 31 percent of expected revenue benefits and 25 percent of expected cost benefits captured. Survey figures, not deal memos — but they are what the firms publish, and they are your prior when a benefits register assumes full adoption on day one.
The decision the interviewer wants named
Clarify the fork in the first minute:
| Decision | What "yes" means | What "no" often means |
|---|---|---|
| Go / no-go on remaining spend | Finish the cash curve | Pause, rescope, or write off sunk |
| Build vs buy vs configure | Own the code path | Package + integrators + config |
| Sequence | Which process, country, or product first | "Agile everywhere" is not a sequence |
| Operating model | Who owns the KPI after consultants leave | IT owns a system; the P&L owns the benefit |
Sunk cost is already gone. The case is about remaining cash versus realized benefit. Realized benefit is not the benefits register. It is adoption times process delta, net of dual-run and double-counted FTEs.
What the firms' own research implies for your prior
You do not need to memorize every exhibit. You need one usable prior and the honesty to say it is a prior.
| Source (as of published date) | Figure | How to use it in a case |
|---|---|---|
| BCG 2020 / 2021 | ~30–35% meet objectives in the "win" sense | Base case is not the board deck |
| McKinsey 2022 survey | Median ~31% of revenue upside, ~25% of cost upside captured | Haircut optimistic opex savings |
| BCG 2024 large-scale tech programs | ~30% meet timeline, budget, and scope | Schedule and dual-run risk are first-class |
BCG's 2024 Build for the Future work found that only about 30 percent of companies with large-scale tech programs (more than 3 percent of annual tech budget) fully met timeline, budget, and scope expectations. That is a delivery statistic, not a value statistic, but dual-run cost lives in the delivery miss.
Incumbents disclose spend more clearly than outcomes. JPMorgan Chase told investors it expected about $17 billion of firmwide technology spend in 2024 ($4.5 billion products/platforms/UX; about $3 billion modernization). Its technology page now cites more than $18 billion annually. Nine-figure programs are routine; "we went live" still does not appear on a P&L. Microsoft's FY2026 Form 10-K warns that AI and cloud CapEx sits "in advance of fully developed revenue streams" and that revenue "may not be realized in the expected timeframes" if adoption lags. Even a vendor that sells transformation has to disclose adoption timing risk. Your prompt is the buy-side version of that sentence.
The scoring sheet underneath the chat
I stop listening when a candidate says "digital maturity" as if it were a KPI. The sheet rewards:
- Outcome metric in operations language — claims cycle time, cost per policy, first-time-right, on-time-in-full — not maturity scores.
- Cash profile — internal FTEs, SI, licenses, dual-run — not capitalized fairy dust.
- Adoption as a volume weight — 40 percent of claims on the new stack earns about 40 percent of the process benefit, unless you prove otherwise.
- No double count — procurement savings and "FTE efficiency" often describe the same people.
- A recommendation with a kill switch — what adoption or NPV threshold pauses remaining spend.
Build versus buy matters mainly for time-to-adoption and dual-run duration, not for which logo appears on the license invoice.
Worked example: Crestline P&C as a cash-flow decision
Invented company, invented numbers — the kind you will get on a whiteboard. Crestline P&C is midway through replacing its claims workflow stack. The partner asks: given remaining spend and the latest adoption exhibit, should Crestline finish, freeze scope, or pause?
Inputs (treat as cash, not accounting):
- Remaining program cash: $120 million over three years — $40m in Year 1, $50m in Year 2, $30m in Year 3.
- Claims-handling opex baseline: $280 million per year.
- At full adoption with real process redesign (straight-through processing, not a prettier screen for the same desktop procedure): 15 percent cost reduction → $42 million per year.
- Dual-run / legacy keep-alive while both stacks are live: $16 million per year, starting when the new system goes live in Year 2, ending only when legacy is switched off.
- Case discount rate: 10 percent.
- Current adoption path on the new stack (percent of claims volume): 0% Year 1 (build), 25% Year 2, 50% Year 3, then either 80% Year 4+ if they force cutover, or stuck near 50% if they leave dual-run open.
Gross benefit in a year = adoption × $42m. Net operating effect = gross benefit − dual-run (while dual-run runs). Program cash is separate.
| Year | Program cash | Adoption | Gross benefit | Dual-run | Net ops | Free cash |
|---|---|---|---|---|---|---|
| 1 | 40 | 0% | 0 | 0 | 0 | −40 |
| 2 | 50 | 25% | 10.5 | 16 | −5.5 | −55.5 |
| 3 | 30 | 50% | 21 | 16 | +5 | −25 |
| 4 | 0 | 80% | 33.6 | 0 | +33.6 | +33.6 |
| 5 | 0 | 80% | 33.6 | 0 | +33.6 | +33.6 |
Year 4 assumes Crestline kills legacy once 80 percent of volume is on the new stack. At 10 percent: PV ≈ −36.4 − 45.9 − 18.8 + 23.0 + 20.9 ≈ −57. Even with forced cutover, remaining cash does not clear the hurdle on a five-year window; later years would have to be durable to repair it.
Now the stalled path, which matches the spirit of the published medians: adoption stalls at 50 percent, dual-run never ends.
| Year | Program cash | Adoption | Gross benefit | Dual-run | Free cash |
|---|---|---|---|---|---|
| 1 | 40 | 0% | 0 | 0 | −40 |
| 2 | 50 | 25% | 10.5 | 16 | −55.5 |
| 3 | 30 | 50% | 21 | 16 | −25 |
| 4 | 0 | 50% | 21 | 16 | +5 |
| 5 | 0 | 50% | 21 | 16 | +5 |
Steady-state free cash is only +$5 million a year while you still burn $16 million on legacy. Five-year NPV is worse than −90 before you even argue about Year 6+. That is the interview point: go-live in Year 2 looked like progress; the cash flow never did.
Recommendation you should say aloud. Freeze remaining scope to the two claim types that are 60 percent of cost. Fund an adoption workstream with hard decommission dates and no dual keying. Set a kill switch: if volume on the new stack is not above 70 percent within 12 months of go-live, pause remaining spend. Do not defend the program with "we are live."
Ned's rule. Go-live is a project milestone. Benefits are a volume-weighted run rate net of dual-run. Score the second. If you cannot write the adoption path as numbers on the page, you do not have a business case.
How to open when the prompt is vague
Thirty seconds, then structure:
- Confirm the client is buying change inside an incumbent, not selling software.
- Name the outcome KPI in operations language and ask what is already spent versus remaining.
- Ask for the benefits register, the adoption or percent-volume exhibit, and the dual-run or parallel-run cost.
- Say you will treat benefits as adoption × process delta and compare PV of remaining cash to PV of realized net benefit under a base and a stalled path.
If they hand you a profitability tree, use it only for the opex baseline. The tree does not know about two systems running at once.
Practice this today
Run one transformation-shaped case and force yourself to put an adoption column next to every benefit line before you synthesize. On CoachNed, start with the five-minute first rep at /start (three typed turns, instant scores, no account required), then a full voice case at /interview if you want the seven-score debrief. The math muscle is the /drills/math set; everything is open for seven days with no card, then $120 for a recruiting season or $49 a month.
Frequently asked questions
Are digital transformation cases the same as technology cases?
No. Technology cases usually mean the client is a software, platform, or product company and the economics are ARR, attach, and unit cost. Transformation cases mean the client is an incumbent funding a program; the economics are adoption, dual-run, and benefits realization.
What adoption rate should I assume if the interviewer gives no exhibit?
Say you will use a conservative prior until you get data. Citing the published medians — roughly a quarter to a third of expected cost or revenue upside realized — is fair as a sensitivity, not as a fake "industry standard." Then ask for volume on the new journey.
Does build versus buy change the recommendation?
It changes time-to-adoption and dual-run duration more than it changes the license line. Buy-and-configure often wins when calendar risk dominates NPV; custom wins when the process is the product and packages force workarounds that kill adoption.
How is this different from an operations case?
Operations cases optimize a process that already exists. Transformation cases add a multi-year cash profile, a second system in parallel, and a political sequencing problem. The ops KPI still matters; the dual-run line is what ops trees omit.
Will every firm ask a digital transformation case?
Not every interview. Digital, tech, and operations practices ask them often because clients spend at bank-scale on technology. Treat the prompt as firm-agnostic: the cash-flow logic does not change with the logo on the building. CoachNed is independent and unaffiliated with McKinsey, BCG, Bain, or any other firm named for interview practice.
Sources
- Flipping the Odds of Digital Transformation Success (BCG, 2020) — ~30% win-zone success rate for digital transformations. Checked 2026-09-24.
- Performance and Innovation Are the Rewards (BCG, 2021) — ~35% achieve objectives across 850+ companies. Checked 2026-09-24.
- Three new mandates for capturing a digital transformation's full value (McKinsey, 2022) — median 31% of revenue benefits and 25% of cost benefits captured. Checked 2026-09-24.
- Most Large-Scale Tech Programs Fail: How to Succeed (BCG, 2024) — ~30% of large-scale tech programs meet timeline, budget, and scope. Checked 2026-09-24.
- JPMorgan Chase 2024 Investor Day transcript — firm overview — ~$17B expected 2024 firmwide tech spend and investment split. Checked 2026-09-24.
- JPMorganChase Technology — firm cites $18B+ annual technology investment. Checked 2026-09-24.
- Microsoft Corporation Form 10-K for fiscal year ended June 30, 2026 — CapEx ahead of revenue; adoption timing risk disclosed. Checked 2026-09-24.