Draft for Tamir's review. Not published.
Ask for evidence instead of status
Ask for the milestone list with original and current dates, and the reason for each move. Ask what was accepted at the last milestone, by whom and on what test. Ask for spend against plan and the forecast to complete, next to the last two forecasts.
Stop accepting revised plans without a reason for the revision. Ask what was wrong in the last plan, what changed, and why this one is right. One summary score can hide a failing dependency that blocks the whole benefit, so keep critical items reported separately.
Separate the reporting from the delivery
If the vendor presents alongside management, ask management to report on the vendor, with the vendor absent. Require acceptance evidence before each milestone payment. Have someone outside the vendor test a sample of what is marked done.
When delays are blamed on others, separate the external constraints from the decisions you could still make about them. Ask which work could go ahead, or be dropped, without the dependency.
Where the seats differ
If you're on the board or a steering committee, set an evidence routine at the gates from the start, and the program will report to it. Arrange an independent review before each major commitment or payment. Decide in advance what you will do if a gate fails: hold, re-plan or stop. When clinicians or engineers raise specific risks, ask management to answer each one with evidence.
If you run the company, ask for updates you can test: what customers or users can do today, what the next release delivers and what blocks it. If you must give a date, give one for a reduced launch the evidence supports, and date the rest separately.
What to check before you decide
- Ask for the milestone list with original and current dates, and the reason for each move.
- Ask what was accepted at the last milestone, by whom and on what test.
- Compare the current forecast to complete with the last two forecasts, and ask for its basis.
- Have someone independent of the vendor test a sample of functions marked done.
- Require acceptance evidence for each milestone before its payment.
- Ask which remaining items depend on your own decisions or data, and clear those first.
- Decide now what the board or committee will do if a gate fails.
Questions people ask
Our capital project reports green to the board every quarter but dates keep moving, how should directors verify progress?
Ask for evidence rather than status: what was tested and accepted, what was spent against what was planned, and what the next milestone needs to pass. Green should mean something passed. It shouldn't only mean that nothing has failed yet. It depends on whether the project has defined acceptance per milestone and on who verifies it.
I sit on a steering committee for a large public IT program with no experience, how do we oversee it properly?
Oversee by evidence at gates: before each major commitment, require what was delivered and accepted, what was spent, what the next phase needs and what the risks are, from someone who is not delivering it. Set that routine at the start and the program will report to it. It depends on the program's structure and on whether acceptance tests and gates exist.
Our oversight committee sees missed dates and rising spend but management says the program is on track, what do we do?
Stop accepting revised plans without a reason for the revision: ask what was wrong in the last plan, what changed, and why this plan is right. Then get an independent picture of the real state before approving further spend. It depends on whether the slips come from the program's own work, from decisions waiting elsewhere or from vendor delivery.
Our startup board gets technical updates from the CTO and can't tell if the product will be ready before the money runs out, what should we ask for?
Ask for updates in terms the board can test: what customers can use today, what the next release delivers and when, what blocks it, and how the team's capacity matches the plan. Then ask for an independent technical read once, at a point where the answer still changes decisions. It depends on how far the product is from what customers will pay for and on the months of runway left.
Our large vendor program always reports on track to the board, how do we get a view that isn't the vendor's?
Separate the reporting: management reports on the vendor rather than with it, with acceptance evidence, spend and risks, and the board gets an independent review before each major payment. A vendor that is on track will pass; one that is not will show. It depends on the contract's milestones and on whether acceptance is defined.
Clinicians warned the board about the medical record project and management says it's normal, what should the board do?
Take both seriously: ask for the specific risks the clinicians raised, each with management's answer and evidence, and have an independent reading of the contested ones before the next go-live decision. Normal does not mean safe. It depends on which risks touch patient safety directly and on what the go-live plan does about them.
Our open banking program has regulatory deadlines and management says on track, how does the board verify?
Ask for the deadline's requirements as a list with each item's state: built, tested, certified, and the evidence. Then ask what the regulator will test and whether the bank has tested the same. One independent check before the deadline is worth more than any report. It depends on what the regulator will verify and on how many items are still untested.
Should the board combine factory technology projects into one progress score?
Keep portfolio summaries connected to material dependencies and exceptions. Oversight depends on whether one delayed capability can prevent the promised business outcome despite progress elsewhere.
Should directors accept external dependencies as the explanation for every transport overrun?
Separate external constraints from the decisions management could make about them. Further funding depends on evidence, alternative scope, and a credible account of controllable work and retained uncertainty.
Vendor wants a six month extension on our flagship system and the minister wants a date, how do I get the truth about what's left?
Ask the vendor for the remaining work by function with the evidence of what is done, and have someone outside the vendor check a sample of it. Then give the minister a date for a reduced launch that the evidence supports, with the rest dated separately. It depends on whether the delay sits in the vendor's work, the agency's decisions or testing.
How I can help with this decision
- Ask or talk (Free)
- I give my view on what your reporting should contain and the questions that turn a green status into evidence.
- Review (Pay if it was worth it)
- I write an independent assessment of the program's real state, its reporting and the vendor's performance, for the board, committee or CEO. I recommend what to require and whether to change course.
- Retain (When it makes sense)
- I stay available at each major gate or payment to review the evidence before you approve it.