Draft for Tamir's review. Not published.
Put the request in a structure you can judge
Ask for every item in one of three groups: run, obligations and new. Each item needs a one-line justification, the service it supports, and what happens if it is cut for a year. That last answer should come from the business owner rather than from IT.
You can judge the new projects yourself by whether the services matter. For a large replacement, ask what forces the change and when: support ending, a product the system can't handle, or a regulation. Ask for the alternatives that were rejected, including keeping the core and replacing around it.
Find the costs the proposal leaves out
Estimates usually leave out integration, data migration, change and running costs. Supplier proposals often exclude work described as your responsibility. Lay the estimate and the quotes side by side by component, and give each excluded activity an owner.
Watch for savings that assume something happens by itself, such as jobs disappearing after automation. Separate added capacity from costs the business can actually avoid. A donated or cheap system can still carry operating, staffing and support obligations for years.
Depending on your seat
If you're on the board, put competing proposals on one basis: the risk each removes and the value each adds, with evidence. Safety-critical maintenance comes first. A digital program must then show which losses it reduces and how that is measured.
If you're the CEO, ask how the project would be staged so value arrives before the end, and fund the first phase with a decision at each step. Where a regulator sets the need, build the minimum the rules require by their date and decide on more later.
What to check before you decide
- Ask for the request split into run, obligations and new, with a one-line justification per item.
- For each new project, ask which business unit asked for it and who will own the result.
- Ask the business owner what breaks if each item is cut for a year.
- Ask what forces a large replacement and when, and which alternatives were rejected and why.
- List what the estimate or the supplier's scope excludes, and give each item an owner.
- Ask how much of the budget is the core and how much is surrounding systems and data.
- Ask for last year's budget against delivery, to see the track record.
Questions people ask
Board must choose between funding a plant digitalisation program and the maintenance backlog, how do we judge?
Judge both on the same basis: the risk each one removes and the value each one adds, with evidence. Maintenance items on safety-critical equipment come first by any measure; the digital program must then show which losses it reduces and how that is measured. It depends on what sits in the backlog and on whether the digital program targets measured losses.
Next generation wants to digitise our family company and the board is sceptical, how do we decide on evidence rather than family dynamics?
Ask the proposers for a first project with a measure the board understands, such as waste, delivery time or stock, and a budget the company can lose. Approve that rather than a program, and let the result decide the next step. It depends on which problem the first project targets and on whether the company has data to measure it.
Should directors approve robotics savings that assume jobs disappear automatically?
Require a credible path from changed tasks to business value. Approval depends on retained work, staffing arrangements, redeployment value, and responsibilities assessed with management and appropriate advisers.
Should we accept donated hospital technology with unfunded operating commitments?
Assess the full retained commitment separately from the donated asset. Acceptance depends on useful service value, ongoing responsibilities, and the hospital's ability to sustain or exit the capability.
What technical evidence supports treating software spending as a lasting business asset?
Provide a factual account of what the work creates and sustains. Accounting conclusions belong with finance and accounting advisers, informed by identifiable capability, dependencies, and expected usability evidence.
IT division sent me a large budget request I can't assess, how do I judge it as a director general?
Ask for the request split into keeping existing systems running, obligations such as security and regulation, and new projects, each with the service it supports and what happens without it. You can judge the third group yourself by whether the services matter. It depends on how much of the request is unavoidable maintenance and on which new projects have a business owner outside IT.
My CIO proposes a core system replacement that would be our biggest project ever, how do I know it's necessary and sized right?
Ask what forces the change and when: vendor support ending, a product the system cannot handle, or a regulation. Then ask for the alternatives the CIO rejected and why, including keeping the core and replacing around it. It depends on what the current system actually blocks and on how much of the proposal is the core versus everything attached to it.
Limited hospital budget, infrastructure refresh or patient facing digital service, how do I choose what to fund?
Fund the infrastructure items whose failure would stop clinical work, and a patient-facing service that runs on them safely, even if smaller. Ask IT for the failure risk per infrastructure item rather than a single refresh request. It depends on the age and failure history of what you run on and on which patient service the hospital can operate well.
Consultants estimated our technology roadmap and vendors quoted a multiple of it, what happened and what do I tell the board?
Roadmap estimates usually leave out integration, data migration, change and running costs, and vendors quote them in. Reconcile the two line by line so the board sees what was missing rather than a mystery. Then re-sequence the roadmap to what can be funded. It depends on what the consultancy's estimate covered and on whether the vendors' quotes include scope the roadmap never needed.
Regulator will require customer data APIs for third parties and our vendor proposes a big program, how much do we really need?
Build what the rules require at the date they require it, and no more; the first version is usually a small set of data products behind consent and security controls. Ask the vendor to separate the regulatory minimum from the platform it wants to sell. It depends on the regulator's published scope and timeline and on how your customer data is held today.
What costs are missing from this banking platform replacement proposal?
Evaluate the complete business commitment, including work outside the supplier contract. Signing depends on migration obligations, retained systems, customer transition, and accountable owners for exclusions.
Should we commercialise our internal software or keep it for our own business?
Treat commercialisation as a new business decision rather than a simple extension of internal use. It depends on outside demand, usable rights, product independence, support needs, and the effect on the core business.
How I can help with this decision
- Ask or talk (Free)
- I give my view on how to read the request in that structure and which items usually hide in each group. I tell you how to test necessity and sizing without becoming the project's specialist.
- Review (Pay if it was worth it)
- I write an independent assessment of the request or proposal by category and project: necessity, scope, missing costs, staging and risks. I recommend what to approve, reduce, stage, defer or question.
- Retain (When it makes sense)
- I stay close through the year or the program to review delivery against the approved items and each stage gate, in short reports in plain terms.