In short
- Count only what your own organisation can observe, measured by the people who do the work.
- State plainly what will not be saved. A case with limits is easier to believe.
- Name who will confirm each number, and ask for a trial on real data before the full commitment.
The investment committee meets on Monday. Page four of the paper carries one bold line: the new programme platform will make the delivery office more efficient by a stated percentage. The finance director asks where the figure comes from. The author says it is an industry benchmark. Asked whose, he turns to a slide from the vendor's presentation.
The paper is sent back for more work. The more work consists of finding a second source for the same number.
I think the finance director is right, and I think the case can be made without the number. An organisation that runs a large programme already holds everything an honest case needs. It has never counted it.
What goes wrong
A borrowed percentage fails in four ways, and a careful reader of the paper will find each of them.
The number belongs to someone else. It describes other organisations, chosen by the party that gains from the sale and measured in a way nobody in the room can inspect. Nobody on the committee can confirm it before the purchase or check it afterwards.
A percentage needs a base, and the base was never measured. To save a share of the office's effort, someone has to know what the effort is today. The hours that go on collecting status are rarely counted. The case multiplies a borrowed number by a guessed one and prints the result to the nearest dollar.
Time is booked as money. Hours freed from collecting status stay on the payroll. They become cash only if a post is removed or a contract is ended, and the paper rarely says which. Finance knows the difference between capacity and cost, and stops believing the paper at that line.
Nobody is named to check. The case is approved, the software is bought and the benefit line is never opened again. Everyone involved knows this when the paper is written, which is why the number can afford to be large.
The result is a case argued on belief. If it is approved, it is approved because the sponsor wanted the software, and the number has added nothing except a risk to the author's credibility. A year later, when someone asks what the investment returned, there is nothing to measure against. The same gap opens whenever a benefit has no owner, as Delivered, and no benefit describes.
What good looks like
An honest case is smaller and harder to argue with. It needs no outside data, and a programme office can assemble it in a few weeks.
Count what your own people can observe
Four quantities are within reach of almost any programme office.
The hours spent collecting status and assembling packs. Ask the people who do this work to keep a simple log for two reporting cycles. Count the chasing, the copying and the checking of one sheet against another.
How long a decision waits. Take the last ten decisions in the committee's minutes. For each, note the date the decision was first needed and the date it was taken.
What one wrong green cost. Find the last date that was missed after being reported on plan. Ask finance what the slip cost: the contract extension, the team kept standing, the benefit that arrived late. One real case from your own ledger carries more weight than any average.
What the audit asked for and did not get. List the requests from the last audit or review that could not be answered from the record, and the days spent reconstructing the answers.
| What to count | How to measure it yourself | Who confirms it |
|---|---|---|
| Hours spent collecting status and building packs | A time log kept for two reporting cycles by the people who do the work. | The head of the programme office |
| How long a decision waits | For the last ten decisions in the minutes, the date each was first needed and the date it was taken. | The committee secretary |
| What one wrong green cost | The last missed date that had been reported on plan: the extension, the standing team, the delayed benefit. | The finance director |
| What the audit asked for and did not get | The open requests from the last audit or review, and the days spent answering them. | Internal audit |
Say what will not be saved
Give the limits a section of their own. Freed hours are capacity, and the paper should say whether any post or contract will end. A late project will stay late. The judgement of the people who run the office is still needed, and the paper should say what they will do with the time. Licences and the effort of the rollout are costs, and belong on the same page. A finance director who reads the limits first reads the rest with more trust.
Name who will confirm each number
Every line in the case gets a name and a date. After the trial, the named person repeats the same measurement by the same method: the same log, the same count from the minutes. That person should not be the author of the case. A claim confirmed by the person who made it has not been confirmed.
A finance director who reads the limits first reads the rest with more trust.
Ask for a trial on real data first
Ask the committee for a first stage in place of the full commitment: one portfolio, its real data, a fixed period, and the four measures taken before and after. The committee then makes a small decision it can check, and the larger one follows on evidence. If the measures do not move, the organisation has spent a little to avoid spending a lot. Seven tests for strategy execution software sets out what to try during such a trial.
This is the route we ask buyers to take with Lodestar. Send us a slice of a real portfolio and we load it, so the first session runs on your own programme data. Test on that data before any larger commitment, with your own people taking the four measures and the person you named confirming them. Tell us what you want to test.
Questions for your next review
- Which numbers in this business case were measured inside our own organisation, and by whom?
- What does the case say will not be saved?
- Who will measure each line again after the trial, by what method and on what date?
- What did the last missed date that had been reported on plan cost us, according to finance?
- What is the smallest trial on real data that would let us check the case before the full commitment?
A business case is a promise that someone will be asked to keep. Build it from numbers the organisation can count for itself, and there will be someone able to say whether it was kept.
