The Discipline of Not Starting: A Five-Question Screen for New Initiatives

Few UK organisations suffer from a shortage of ideas. The more common condition is an excess of commenced ones. A new product line, a promising hire, a partnership that seemed obvious over lunch: each is individually defensible, and collectively they explain why so many capable teams end the financial year exhausted and slightly behind on the things that mattered most.

The ampersand in our name is a useful prompt here. An ampersand joins two things, and in doing so makes a claim: that A and B together are worth more than A alone. Every new initiative makes the same claim. It promises addition — more revenue, more reach, more capability — without always being honest about what it subtracts.

The screen below is designed to test that claim. It takes under an hour, requires no consultants, and works best when the person proposing the initiative is in the room and the finance lead is not allowed to leave early.

Question one: what will this displace?

Capacity is not elastic. A team that takes on a new priority slows down an existing one, whether or not anyone has acknowledged it. So ask for the trade in named units: whose hours, which release slot, which line in the account plan, which fortnight of the delivery calendar.

If the answer is that nothing will be displaced, the proposal is not funded. It is borrowed — against future delivery, future goodwill or future weekends. Both are real liabilities, and neither appears on the management accounts until they are expensive.

Question two: who owns it on a Wednesday in February?

Sponsorship is not ownership. A director who is enthusiastic in September may be consumed by year-end in February, which is precisely when a new initiative tends to need decisions rather than applause. Name the individual who will do the work, the time they have been given to do it, and the person who covers their existing responsibilities.

A useful test: can the proposed owner describe, unprompted, what they will stop doing in order to start this? If they cannot, the initiative has been approved in principle and resourced in hope.

Question three: which existing commitment does this make harder?

Second-order effects are where most initiatives quietly fail. A new customer segment strains the service desk. A new product stretches a supply chain that was already coping rather than comfortable. A new office adds a management layer nobody has designed.

Ask the operations lead and, separately, the front line. Their answers will differ, and the difference is the finding. The person closest to the work usually knows which commitment will bend first; they are rarely asked.

Question four: what is the cheapest version that would still teach us something?

Distinguish evidence from enthusiasm. Most initiatives can be reduced to a reversible, time-boxed test: one region rather than five, one client rather than a segment, a manual pilot rather than a system build. The question is not whether the full version is attractive, but whether a smaller version could prove the idea wrong cheaply.

Run a short pre-mortem. Imagine it is twelve months from now and the initiative has failed. What would have to be true for that to happen? If those conditions can be tested in eight weeks for the price of a fortnight's effort, test them.

Question five: what does it cost to run in year two?

Launch economics and run economics are different animals, and approval processes tend to examine the first. Year two brings licence renewals, training, governance, reporting, a bespoke process that needs a named maintainer, and the slow accumulation of expectations that were set in year one.

Ask for the steady-state cost, including internal time. Where an initiative creates a new recurring process, ask who will own it once the project team has disbanded — a question that has a habit of going unanswered in the room and being answered painfully afterwards.

Running the screen in a single meeting

Book forty minutes. Circulate the five questions in advance with a one-page template. The proposer answers; others add evidence rather than opinion. Close with one of four decisions: proceed, proceed small, park with a review date, or decline.

Record the answers. Six months later, that record is the most useful document in the review, because it shows what the team believed and what they chose not to examine.

Resist turning this into a scoring model. Weighted matrices invite gaming, and a score of 3.4 out of 5 tells you nothing about whether the thing should happen. The value lies in the conversation the questions force, not in the arithmetic.

The point is not to say no

An organisation that cannot decline anything eventually declines everything, usually by attrition and usually to its best people. The discipline of not starting is not conservatism. It is what makes a yes credible — to the team delivering it, to the customer waiting for it, and to the board funding it.

The ampersand promises addition. It is worth asking, every time, what exactly is being added, and what is being quietly removed to make room.