I have watched this play out more times than I can count. A managing partner or digital director arrives at a board meeting with a proposal for a platform, website or transformation initiative. The proposal is well-intentioned and may be the right thing to do. Within twenty minutes, the conversation has drifted into vague questions, vague answers and a general sense that nobody is entirely sure what they are being asked to approve.

The board defers: "Come back with more detail." The proposer leaves frustrated. The board feels responsible. The project sits in limbo for another quarter.

I have been on both sides of this conversation. The problem is often less about the volume of material and more about the quality of the decision the paper makes possible.

Digital investment combines uncertain demand, changing technology, operational dependencies, customer behaviour and risks that may be difficult to quantify. Asking only "what is the ROI?" invites a single confident number where a range of assumptions and choices would be more useful.

These five questions give a board a practical route through the proposal. They are neither a substitute for sector-specific scrutiny nor an automatic approval score. They help expose whether the investment has a diagnosed problem, credible value, considered alternatives, accountable ownership and a proportionate first commitment.

Question 1: What specific problem are we choosing to solve?

Ask about the problem before the product.

"We will build a client portal", "migrate to a modern CMS" and "implement an AI document system" describe capabilities. None is a reason to spend money by itself.

A decision-quality problem statement identifies who is affected, what happens today, why it matters and the evidence behind it. For example: clients wait for documents that should be available during a time-sensitive transaction; routine publishing depends on scarce development capacity; colleagues cannot find approved knowledge and recreate it at cost and risk.

The proposal should separate observed facts from explanations. A falling enquiry rate is evidence of an outcome. "The website design caused it" is a hypothesis until other influences and behaviour are examined. The board does not need certainty. It does need to know where the diagnosis is strong and where the first phase must learn.

Then ask why this problem is strategic now. Does it constrain a priority client journey, service, risk reduction or growth choice? Which other problems will receive less attention if this one is funded?

I sat with a COO who had received a large quote for a combined website and CRM rebuild. Asking what problem it solved revealed that much of the brief was unrelated to the urgent issue: a broken handoff between enquiry and onboarding. A contained intervention became possible once the capability list gave way to the problem.

A proposer who cannot express the problem concisely may still have a valid concern. They are not ready for an investment decision until the board can see what success and scope relate to.

Question 2: What evidence will tell us whether it is working?

"Improved digital presence" is an intention. It cannot govern a programme.

Ask for a short chain from delivery to behaviour to outcome:

  • Capability: what will the firm or user be able to do?
  • Behaviour: who needs to use or change something for value to appear?
  • Outcome: which client, operational, risk or commercial result should follow?

A new publishing platform might allow authorised marketing colleagues to make changes without a development ticket. That should change publishing time and support demand. Faster publishing may then help the firm respond to priority market needs. The last link contains more uncertainty than the first, so the evidence and timescale should reflect that.

Request baselines, definitions, sources and owners for the measures. A target such as "increase enquiries by 25%" is decorative if nobody knows the current qualified-enquiry rate, how attribution works or which other campaigns will change during the period. A precise percentage does not make a weak assumption rigorous.

Include safeguards as well as benefits. A faster onboarding journey should not reduce necessary review quality. More personalised content should not depend on inappropriate data use. An AI-assisted process needs measures for exception, error and human oversight, alongside time saved.

Agree when the evidence will be reviewed and which decision it informs. Early measures may show whether the capability works and people can use it. Commercial outcomes may need longer. Do not delay all accountability until the end of a multi-year programme.

Early in Distinction's history, we delivered a project we were technically proud of. The client appeared satisfied, but no one had agreed what "working" meant. There was no defensible moment of success and the relationship faded. It captures the cost of delivering without an agreed outcome.

Question 3: What are the credible alternatives, including deferral?

Every proposal shows the cost of the recommended option. Far fewer compare it with meaningful alternatives.

"Do nothing" does not always cost zero, and it does not always produce catastrophe. The current state may involve maintenance, manual work, missed opportunities, security exposure or service failure. Some of those costs can be evidenced; others need ranges or qualitative risk assessment. Avoid copying a global breach statistic into a firm-specific decision as though it predicts the loss.

Ask the proposer to compare at least:

  • continuing the current arrangement with necessary maintenance;
  • a contained improvement or risk reduction;
  • buying or configuring a managed product;
  • building or substantially customising a service;
  • changing process or policy without major technology;
  • delaying until a dependency or uncertainty is resolved.

The comparison should cover total ownership, delivery time, internal capacity, supplier dependence, security and compliance, accessibility, data and integration, expected useful life and exit. It should state which requirements each option does not meet.

Deferral deserves its own scenario. What changes over three, six or twelve months? Which costs continue? Does a support deadline or contract create a real decision date? Could more evidence improve the choice? The cost of waiting can be material, while artificial urgency weakens trust.

This question protects the board from two symmetrical errors: comparing investment with an imaginary free status quo and approving a large replacement when a smaller change addresses the diagnosed problem.

Question 4: Who owns the outcome and can they act?

Ask for a name, authority and operating model.

A steering committee can provide expertise, challenge and decisions. It cannot replace a single accountable business owner. The owner should care about the outcome after the delivery team leaves, have authority over the necessary trade-offs and be able to secure participation from affected teams.

Clarify several roles:

  • the executive accountable for the intended outcome;
  • the product or service owner making regular scope and priority decisions;
  • technical, operational, security, data and compliance responsibilities;
  • supplier accountability and escalation;
  • people responsible for adoption, measurement and ongoing operation.

One name at the top does not mean one person does all the work. It prevents collective ambiguity about who can decide when evidence changes.

Test capability as well as governance. Does the firm have enough time and skill to act as an informed customer? Who will manage content, integrations, access, incidents and improvement after launch? A platform can fit the requirements and still fail because the proposed operating model exists only on an organisation chart.

One consulting firm's previous digital project had an eleven-person steering committee and no clear accountable owner; it reportedly ran fourteen months over schedule. Committee size does not prove why it was late. Ask whether decisions, escalation and outcome ownership are clear in practice.

Accountability is meant to improve decisions, rather than create a future target for blame. The owner needs agreed tolerances and a route back to the board when cost, benefit or risk moves beyond them.

Question 5: What is the smallest responsible commitment we can approve now?

Boards are often asked to approve a long programme as though scope, demand and technology will remain stable. A phased decision can reduce exposure and bring learning forward.

"Phase 1" should not be a smaller label attached to an arbitrary slice of the final build. It should produce something coherent: evidence that resolves a key uncertainty, a usable improvement for a defined group, a necessary foundation or a safe migration boundary.

Ask the proposer to state:

  • what the phase delivers and does not deliver;
  • its cost range, internal effort and main assumptions;
  • which risks it reduces and introduces;
  • the evidence available at the review gate;
  • what would justify continuing, changing direction or stopping;
  • how the work retains value if a later phase is declined.

Some obligations cannot be piloted away. A critical security remediation, legal deadline or end-of-support migration may require a larger committed scope. Phasing still helps expose dependencies and create control points, but the board should not demand a tiny experiment where only complete remediation makes the service safe.

At Distinction, our WHNN framework, What and How, for the Now and the Next, uses a quarterly rhythm to keep near-term work specific and the later horizon revisable. The principle matters more than the cadence. Certainty generally decreases further into a plan, so approval should distinguish committed delivery from directional intent.

The first phase should also include discovery, governance or research only when those activities answer a consequential question. A paid report that restates known problems and automatically recommends the full programme is not meaningful risk reduction.

Read the answers together

The questions are connected.

A precise problem gives measures a purpose. Measures make alternatives comparable. Alternatives reveal the choice the owner is accountable for. A contained first commitment tests the most important assumptions before the board increases exposure.

Weak answers also interact. A portal proposal with no user-side problem, an adoption target measured by logins, a committee owner and a Phase 1 consisting only of procurement is not rescued by a polished cost model.

Context adds further scrutiny. A regulated financial service may need detailed consumer, operational-resilience and data analysis. A client portal may require a threat model, accessibility evidence and service-support design. A major supplier contract may need concentration and exit review. These five questions organise the investment decision; they do not exhaust the board's duties.

If you are presenting next month, stress-test the proposal against all five. Show evidence and uncertainty plainly. Make the decision requested explicit, including what the board is declining or deferring.

If you are reviewing the proposal, use the questions to improve it rather than perform technological expertise. A strong answer may still lead to "no" or "not yet". That is governance working.

We have put together a formatted checklist for board members and proposers, with prompts for a full, partial or missing answer. You can download it here. Use it alongside the companion piece on presenting digital investment to a sceptical board, then adapt the additional scrutiny to your firm's sector and risk.