Our program is behind plan. Should we continue, change scope, pause or stop?

Fund only against evidence: what is in production, what remains by function with its estimate, and what changed since the last request so the estimate is believable now. Attach conditions: milestone-based release, an independent check at each one, and a defined stopping point. It depends on whether the previous rounds delivered anything in production and on why the estimates were wrong.

Check each change against the regulatory approval: what was promised, what the change does to it, and whether the regulator must be told or asked. Then check whether the dropped parts carry obligations and whether the added parts have a business case. It depends on the terms of the approval and on how the regulator treats scope changes.

Get an independent read of the migration's state: what runs on the new platform, what is left, why it stalled and what the completion needs in money, people and vendor terms. Fund against milestones with that read as the baseline. It depends on whether the stall is a vendor problem, a team problem or a design problem, because the funding structure differs for each.

Approve nothing until the board knows what the fund needs from the program in business terms and why the previous attempts stopped. Then approve a first deliverable in use within a defined period, owned by the business, rather than a platform. It depends on whether anything from the previous attempts is usable and on what the fund's clinicians and managers would use first.

Get the integration status from the partners directly and compare with management's account, and ask for the launch's remaining dependencies with owners and dates. Fund against those dependencies clearing rather than against a date. It depends on whether the partners confirm the account and on how many dependencies remain.

Decide what the program is for before who runs it: which business outcomes it still serves, what it has delivered and what it costs to complete. Then appoint a sponsor with authority over those outcomes, with a mandate the board writes. It depends on whether the program's purpose survived the sponsor and on what has been delivered.

Fund only against a narrow first exchange with a named owner on each side, a written legal basis for that exchange, and a date. If no agency will own a first exchange, the initiative has no sponsor and funding it changes nothing. It depends on whether one exchange with clear value and a legal basis can be found.

Require a forward option comparison based on verified recoverability. The decision depends on essential operating needs, bounded alternatives, and what each option leaves the refinery able to do.

Judge forward options using current evidence rather than the public commitment. The decision depends on remaining value, viable alternatives, and an honest explanation of what changed.

Stop funding the program as a whole and fund only the pieces that change a measurable operation within months. Ask each workstream for the decision or process it changed and the number that proves it. It depends on whether any workstream can show a change already, and on whether the problem is the program's design or its leadership.

Ask both for the same two numbers: what it costs to finish from here and what the business gets at each remaining milestone. The money spent is gone either way; the decision is about the remaining spend against the remaining value. It depends on whether the overrun came from scope growth, estimate errors or vendor performance, because each has a different fix.

Stop the clock on the vendor until there is a written plan with dates you can check, and put the crews' records back into a system this month, even a simple one. Then decide whether the vendor's plan is credible. It depends on what the contract says about milestones and on how much of the system actually works today.

Get one list of what blocks each vehicle, with an owner and a date, because fleet rollouts stall on a few concrete things: installation slots, firmware, data, staff training. Then run a daily count until the authority's date. It depends on where most vehicles are stuck and on whether the authority will accept a plan with dates.

Find out exactly what the central body needs from you, because waiting often hides an incomplete submission or a question nobody answered. Then escalate with a written record of what was submitted, when, and what remains. It depends on whether the approval is blocked by your submission, by the central body's queue or by a policy question nobody owns.

Get the code, designs and access in your hands first, because without them you have no option at all. Then have someone independent assess what is built and what it takes to finish, and compare the agency's price with another firm's on that assessment. It depends on what the contract says you own and on how much of the new site actually works.

Decide on the remaining work. The money already spent shouldn't drive the decision. Ask for a function-by-function estimate of what is left, and check whether the delivered quarter actually runs in production. It depends on whether the slow start was setup cost that is now behind you or a rate that will continue.

Separate the portal's real needs from billing into a short list of reads and writes, because most portals need far less from billing than the integration plan assumed. Then decide on billing on its own merits, in its own project. It depends on which billing functions the portal must change in real time and on the billing system's remaining life.