Digital Transformation Business Case That Finance Approves
How to build a digital transformation business case that finance approves: honest cost models, risk framing, phased ROI and a clear stakeholder structure for any replatform or rebuild.
Your CFO has probably seen the research. McKinsey and Oxford studied large IT projects and found they run 45 percent over budget on average and deliver 56 percent less value than predicted. So when your replatform proposal arrives, the committee does not ask whether it is a good idea. They ask whether your numbers are any better than everyone else’s.
A digital transformation business case gets approved when it answers that question before anyone asks it. This article covers how to do that.
Get a costed, phased rebuild plan your finance team can test, built by the people who deliver it.
What is the approval committee actually afraid of?
Most people think the committee is afraid of the average overrun. It is not. It is afraid of the tail.
The researchers Bent Flyvbjerg and Alexander Budzier call these extreme outliers black swans. McKinsey reports that 17 percent of IT projects go so bad that they can threaten the very existence of the company. A finance director who has lived through one of those does not care about your central estimate. They want to know what happens in the bad case, and whether the business can survive it.
The money is there. A Gartner survey of 301 finance leaders found that 77% planned to increase technology spending. Your rebuild is not up against a budget freeze. It is up against every other technology proposal in the building, and the one with the most believable downside usually wins.
Cost modelling that survives the red pen
Weak cost models list one number: the build. Strong ones break total cost of ownership into three layers over a five-year horizon, because that is how finance will rebuild your model anyway.
- Build cost: discovery, design, engineering, QA, integration work and the internal staff time that usually gets left out. Product owners, subject matter experts and the security reviewer all cost money while they are on the programme.
- Transition cost: the layer most proposals forget. Dual-running licences while the old and new platforms overlap. Content migration and the editorial hours to clean it up. Redirect mapping, a temporary dip in organic traffic, retraining editors on a new CMS, and decommissioning the old stack.
- Run cost: hosting (for example Vercel, AWS or Azure consumption), SaaS licences for a headless CMS such as Contentful or Sanity, monitoring, and a realistic retained team for continuous improvement. A platform with no run budget turns back into legacy within three years.
Then model the option nobody wants to write up: doing nothing. The UK Treasury’s Green Book says the business as usual option must always be taken forward to shortlist appraisal, as the benchmark for every other option. Private companies can borrow that discipline. Staying put has costs too: rising support fees as a CMS version leaves vendor support, contractor premiums for skills that are getting scarce, security patching, and the conversion you lose on templates that cannot be tested or personalised.
Show every figure as a range, not a single point. A P50 and P80 estimate tells finance that you know uncertainty exists and have priced it. A single number tells them you have not.
Name the black swan before finance does
Good risk framing is not a long register of everything that might go wrong. It is the handful of risks that could move the cost by multiples, each with a mechanism for containing it.
Start by correcting for optimism. The Green Book tells practitioners to explicitly adjust for optimism bias by raising cost and duration estimates and lowering benefit estimates. If you make that adjustment yourself, and say so in the paper, the CFO does not have to apply a silent haircut of their own.
On replatforms, the risks that cause real damage are usually predictable. Integration with an ERP or order management system nobody has documented. A product information model that turns out to be three models. Personalisation rules buried in templates. Search engine equity that relies on URL structures no one has mapped. Each one deserves a named owner, an early spike to reduce it, and a cost for what happens if it goes wrong.
The risk section exists to show the committee that the worst case is limited by design: small phases, early technical spikes and pre-agreed points where the programme stops if the evidence turns bad.
Flyvbjerg and Budzier recommend breaking big projects into pieces of limited size, complexity and duration. That advice is the bridge to the next section, because phasing is your main tool for controlling risk, not just a schedule.
Phased ROI projections: fund the evidence, not the vision
Asking for the whole programme budget up front asks the committee to bet on the tail. Asking for the next phase, with clear criteria for releasing the one after, asks them to buy information. That is a much easier yes.
Here is the structure we use to turn one large request into a sequence of smaller, testable ones:
Separate leading indicators from lagging ones. Revenue and cost-to-serve are lagging measures that can take quarters to show up. Page performance, release frequency and editorial cycle time move within weeks, and they tell the committee early whether the mechanism is working.
The order of phases matters as much as their size. There is a fuller treatment of sequencing a replatform elsewhere on this blog. If your current estate is old enough that the architecture itself is in question, the legacy modernisation steps will shape what goes in phase one.
Fund discovery as its own decision:
Ask for a small, fixed budget to map integrations, audit content, benchmark current performance and validate the architecture. The output is a revised estimate with a narrower range, which is worth more to finance than any vision deck.
Pick a first slice that earns:
Choose one journey or market that can go live on the new platform while the old one keeps running, using a strangler pattern behind a routing layer. A single product category or one regional site works well. It should be big enough to prove the benefit and small enough to roll back.
Agree the measures before the build:
Set baselines now for the metrics the slice should move, such as conversion rate, Core Web Vitals pass rate, time to publish, or support tickets per thousand sessions. Agree who owns the analytics and how attribution will be handled.
Write the gate criteria into the paper:
State what result releases the next tranche, what result triggers a rescope, and what result stops the programme. Kill criteria agreed in advance are the strongest signal of credibility you can give.
Model benefits on a ramp:
Benefits arrive after launch, stabilisation and optimisation, not on go-live day. Show a benefits curve by quarter, lagged behind cost, so the payback period is honest.
Stakeholder alignment, structured
McKinsey describes a bank whose finance department joined its transformation only a few months before go-live, and the project stalled as a result. McKinsey’s surveys of IT executives found that an inability to master strategy and stakeholders, along with technology and project content, typically causes about half of all cost overruns.
So map stakeholders by the decision each one controls, not by job title. Finance owns the discount rate, the capitalisation treatment and the format of the benefits case. Technology owns architecture and security sign-off. Marketing and digital own the benefit assumptions and will be held to them. Procurement owns the commercial structure, which shapes how your phases turn into contracts. Legal and data protection can block a launch outright.
Meet each of them one-to-one before the paper goes to committee. Any objection that first comes up in the meeting can cost you a funding cycle. Ask the finance business partner to review the model’s structure early. Their fingerprints on the method are worth more than any slide.
Approval also depends on showing who will run the programme once it is funded. A light, explicit governance model with decision rights and a change-control route answers the question committees always ask: who stops this if it goes wrong?
What the final pack contains
Keep it short. A one-page summary covers the problem, the recommended option against business as usual, the phase-one request, and the gate criteria. Behind it sit the five-year cost model with ranges, the top five risks with their containment, the benefits ramp, and the stakeholder sign-off list.
If you are working with a delivery partner, their estimate should come with the assumptions written out. A supplier who gives you a single fixed number for an undiscovered estate is telling you something about how they price risk. Our guide to evaluating an enterprise agency covers what to look for.
Before you write another slide, build the business-as-usual cost line and the phase-one gate criteria. If those two survive a session with your finance business partner, everything else in the case is just presentation.
Frequently asked questions
What should a digital transformation business case include?
It should include a clear problem statement, a recommended option compared against a business as usual baseline, a five-year total cost of ownership with build, transition and run costs shown as ranges, the top risks with how each is contained, a phased benefits ramp, gate criteria for releasing further funding, and a list of stakeholder sign-offs.
How do you calculate ROI for a website replatform?
Compare the five-year cost of the new platform against the five-year cost of staying put, then model benefits such as conversion uplift, lower licence and support costs and faster publishing on a quarterly ramp that starts after launch and stabilisation. Use measured baselines from discovery rather than vendor benchmarks, and adjust benefits down and costs up to correct for optimism bias.
Why do replatform business cases get rejected?
The usual reasons are a single-point cost estimate with no range, missing transition costs such as dual running and content migration, no costed business as usual option, benefits that start on go-live day, and finance being involved too late to trust the model’s structure.
Should we ask for the full programme budget or fund it in phases?
Phased funding is usually easier to approve and safer to deliver. Ask for discovery and a first revenue-bearing slice, with pre-agreed criteria that release, rescope or stop the next tranche. This limits the downside the committee is exposed to and replaces forecasts with evidence at each gate.
Who needs to sign off a digital transformation business case?
Typically the CFO or finance business partner, the CIO or CTO for architecture and security, the CMO or digital lead who owns the benefits, procurement for the commercial model, and legal or data protection where customer data or new vendors are involved. Pre-wire each of them individually before the committee meeting.
Turn your rebuild proposal into a funded programme
You now know that approval depends on honest cost ranges, contained risk and phases that pay back evidence. Moburst Digital Experience runs the discovery that produces those numbers, then delivers the first slice your next funding gate depends on.