Build a KPI Scorecard in a week: a five-day plan

Yes, if you accept a rough version. The work that takes months is not the building, it is the waiting for perfect definitions and clean data. Five short sessions get you a board with five to eight rows, an owner on each, and targets you already expect to change. That is a weekly Scorecard, and it improves faster in use than on paper.

The plan below asks for about four hours of leadership time, split into five sittings so that nobody has to clear a day. Short sittings also stop the exercise turning into a design workshop, which is the failure mode: a beautiful template with no numbers in it, waiting on a reporting project that never gets funded.

One warning before Monday. Do not start by listing every number your systems can produce. That list is long, comforting and useless, because it answers the question of what you can measure rather than what you would act on. The week starts from the second question and works backward to whatever data you happen to have.

Book thirty minutes and ask every leader the same question. If you could see one number on Monday morning and nothing else, what would it be? Write the answers down without editing them. You will get five to eight, one or two will be vague, and at least one will be something nobody else in the room was watching.

Ask for the number they would want, not the number they can get. The gap between those two is the most useful thing produced all week, and Wednesday exists to close it. If you let the question drift toward what is available in the reporting tool, you will end up with the reporting tool's opinion of your business.

Two rules keep Monday short. One number per person, so the sales lead cannot hedge with four, and no debate about whether an answer is a good measure. Bad answers get fixed on Tuesday or dropped in week three. Arguing about them now costs an hour and settles nothing, because none of you have seen the row in use yet.

A row nobody can define gets counted differently every week, so Tuesday goes on turning each answer into a sentence. Name, owner, and the exact rule for what counts. Half an hour per row is normal for the first three, then it speeds up because the same arguments repeat. Write the rule where the person collecting the number will read it.

The arguments are the work. Does a proposal count when it is sent or when the client opens it? Is a job on time against the date we promised in week one or the date we renegotiated in week four? Both answers are defensible and only one of them can be in the spreadsheet, so pick and record it.

Owner means the person who can change the number through their own actions this week, not the person who runs the report. If a row has two owners, a red week produces two explanations and no decision. If it has none, it is a fact about the market and belongs in the monthly review instead.

Every row needs a source and a person who pulls it, and Wednesday is where two or three rows fall over. If a number takes longer than five minutes to collect, either automate the collection or swap in something close enough that somebody can get it in one. A proxy you actually have beats an ideal measure that arrives late or never.

Walk each row from the definition to the click that produces the figure. Which system, which filter, which day of the week, and who is doing it. A row that lives in one person's head is a row that goes blank the week they are on vacation, and a board with blanks in it stops being read within a month.

Expect one favorite to die here. Somebody always wants a number that would need a change to the way work is logged. That is a real project with a real cost, and it can go on the issues list, but it does not go on this week's board. Take the closest thing you can pull on Friday and move on.

Targets are the part teams stall on, so give them an hour and no more. Where you have a few months of history, take a good week out of it and use that figure. Where you have none, ask the owner what fair looks like and mark it provisional. A wrong target beside a row still tells you more on Friday than an empty column does.

Prefer a band to a single figure on anything that bounces around week to week. Two numbers, a floor and a ceiling, keep a team from treating ordinary variation as a crisis, and they make the genuinely bad week obvious to everyone in the room without anyone having to say so.

Where you do have history, thirteen weeks is the usual window: long enough to show a pattern, short enough to still describe how you work now. Once a row hits its target three weeks running, raise it. That habit matters more than the accuracy of anything you set today.

Expect to change something in each of the first four weeks, and then to stop. Weeks two and three usually kill one row, rename another, and move an owner. By about week six the board settles down and the arguments move off the definitions and onto the business, which is the point of building the thing at all.

Resist adding rows. The pressure to grow the board is constant, because every leader can think of a second number they would like to see, and a board of twenty rows gets skimmed rather than read. If a new row goes on, something comes off, and the trade gets made out loud in the meeting.

Once the shape holds, the department lists are worth a look for rows you did not think of. Take one or two, not ten. The value of a weekly board comes from the same small set of numbers being read at the same time every week by people who have to answer for them.