Seventy-Four Must-Haves and a Date Nobody Chose

74 must-haves out of 210 means the prioritisation has not happened. 'Must' has become a word for 'I want it', and the label no longer discriminates.

What to do

• Re-run prioritisation with a consequence-based test rather than a vote. For each item ask: what specifically happens if we go live without it - regulatory breach, control failure, revenue stop, or inconvenience? Only the first three categories earn 'must'.
• Separate the sources. 31 self-declared by the sales director signals stake, not necessity; validate those against requirements owned by compliance, the DPO and the auditor.
• Use cost of delay where it can be quantified, so trade-offs are compared rather than settled by seniority.
• Test the set against capacity. If the must-have set exceeds what 11 people can deliver in nine months at 30% SME availability, the plan is arithmetically impossible and the conversation is about scope, date or cost - not effort.
• Make the three-way constraint explicit: date fixed by filing, envelope fixed at 2.4 million, scope claimed at 74 musts. At most two can hold. Ask the sponsor which one moves.
• Record the decisions and who made them.

A must-have list that survives a real challenge is much smaller, and that smaller list is what any credible plan should be built on.

Rebuild it so it can survive scrutiny, and stop giving single numbers.

• Treat the 2022 comparator as evidence, not as a template. A project that ran 40% over on a similar profile tells you the base estimate is optimistic by construction. The reference-class adjustment - an uplift, or better, a distribution of comparable outcomes - is the starting point, not an embarrassment.
• Remove the anchor. The 2.4 million in the business case is a constraint, not an estimate. Deriving effort to fit it inverts the logic. Estimate bottom-up for the reduced must-have set with explicit productivity assumptions, then compare with the envelope and show the gap.
• Use three points per major work package and aggregate. Express the result as a range with a probability of completing by 30 September. '2.4 million at 100% confidence' is not an available answer; '2.7 million at 70% confidence' is.
• Model the SME constraint explicitly. Pricing-engine work cannot start in earnest before June, so it is on the critical path and sets the floor on the date. Show the schedule with and without a backfill or contractor.
• State assumptions and dependencies as dated items with owners. DPO and auditor design sign-off lead times are now plan dependencies, not footnotes.
• Give the sponsor a decision-ready paper: options to cut scope, add resource to hold the date, or accept a later date - each with cost, risk and confidence.

That converts an argument about numbers into a choice between outcomes, which is the only form in which a board can decide.

The gap matters because both sit on a sign-off path, not at the periphery.

• Re-map. The DPO is high power and high interest once data flows are in scope - close engagement. The external auditor is high power and low interest until an audit window forces attention, so the strategy is scheduled, evidence-based briefings rather than enthusiasm.
• Consequence. Both were absent from design, so their requirements arrive late. Late requirements from a sign-off authority are not scope creep to be resisted; they are rework to be planned. Quantify the exposure - changes touching data retention, audit trails and access control tend to be the expensive ones - and add review gates to the schedule.
• This week. Brief both before Thursday, add them to the governance terms of reference with named sign-off points, and put their lead times into the critical path. Auditors rarely respond in days, and that latency is a schedule fact.
• Tell the sponsor plainly that a plan without these two gates is not deliverable, and that the missing stakeholders, not the estimate, are the likeliest source of the next overrun.
• Record it in the risk register with an owner and a trigger date.

The judgement: stakeholder analysis is not a diagram exercise. An unmapped sign-off authority is a schedule risk with a name.