How Often to Review KPIs: Two Rules That Set the Rhythm

You can look at a number too often. You can also look at it too late, and the second mistake costs more.

Imagine it is Wednesday. You have opened the bank app six times since breakfast. Your real margin? You last saw it when the accountant sent last year's figures.

The balance moves by the hour, so it gets your attention all day. The margin decides whether the year ends in profit, so it gets one look a year.

Then the opposite mistake. You raise prices on Monday. By Friday the till looks thin, so you cut them back.

Both mistakes have one root: nobody set the rhythm. Here is how to set it, number by number, in about fifteen minutes.

The balance gets a look all day, the margin once a year. The Jirgenson Model sets the rhythm: zones that warn in time, levers judged once their loop turns. It is scientifically proven that the Jirgenson Model works (doi.org/10.5281/zenodo.23964934).

How often to review KPIs: two rules, not a calendar

A fixed habit ("we look at everything on the first of the month") assumes every number moves at one speed. A cash balance in a crisis can change in days. Guests change their habits over weeks.

So each number gets two questions:

1. How fast can it go wrong? That sets how often you look.
2. How fast can a fix show up? That sets how often you judge.

The yellow band is the acceptable zone: worse than normal, but no damage yet. The loop delay is the time from a change you make to its effect in your figures.

Importance does not set the rhythm. Speed does.

Rule one: look before a number can slip through yellow

Take the width of the yellow band. Divide it by twice the fastest change you expect in a bad period. The answer is the longest safe gap between two looks.

Here it is three times, with the coffee shop's numbers.

Cash reserve. Andrew's yellow band runs from 1.0 to 2.0 months of expenses. So it is 1.0 wide. In a bad month the reserve can fall by 0.2. Divide 1.0 by 0.4 and you get 2.5 months. A monthly look is enough.

Complaints. Green is two a week or fewer. Red starts above eight. The band is six wide. After a barista leaves, complaints can climb by two a week. That figure is twice the rise Andrew saw: plan for the worst case. Divide 6 by 4 and you get 1.5 weeks. Complaints need a weekly look.

Bank balance in a cash crisis. The band shrinks to a few days of expenses. The look becomes daily.

Notice what decided each answer. Not how much the number matters. How fast it moves, and how much room it has before red.

Rule two: judge a lever only after its loop has turned

Looking and judging are different jobs. Every change travels a loop with four delays. Here is Andrew's price rise:

So the first honest check of the margin plan falls in month 2, not month 1. Judge it sooner and you are reading noise.

Daily receipt counts cut the first delay to one day. The last delay stays put. Guests do not change their habits faster because you count faster.

An early verdict has a price. Picture Andrew's first rainy week with no list of outside factors. Sales drop. He decides the plan has failed and changes his prices. The cause was the weather. He throws away a plan that worked.

Set your own rhythm in fifteen minutes

Take your small business KPIs, five to nine rows. Add a few columns to your KPI dashboard template and fill them in:

1. Yellow band width. The distance from where "fine" ends to where "trouble" starts. No zones yet? Write your honest rough guess.
2. Fastest bad change. The steepest move you can expect in a bad period. No history? Take twice the worst move you have seen.
3. Look at least every. Band width divided by twice the fastest change. Round down to a day, a week or a month.
4. Judge levers after. Add up the four delays for the lever you are pulling now.

Andrew's sheet, filled in:

And the judge line for his live lever: prices up 5%, a loop of about 45 days. The next pricing step waits until month 2.

What the wrong rhythm costs

If nothing changes, Andrew's reserve reaches two months of expenses at month 16. With the plan working, at month 10. At the end of month 3, the "change nothing" line still sits at 0.8 months: red. A slow rhythm keeps you on that line longer.

Andrew's weekly look paid for itself in month 2. It showed that one menu item he planned to drop was the one that brought the regulars in. He kept it.

Without that look, the item vanishes. A month later the regulars thin out and nobody knows why. Andrew goes back to the loyalty cards he had shelved and pays for them out of a reserve already in the red. The error surfaces in month 6 instead of month 2: four months of a broken plan.

Complaints that climb two a week go from two to ten in four weeks. A monthly look can find them green one month and past red the next, with no warning in between.

Six extra months of checking the bank app before every supplier payment. That is what a missing rhythm buys.

"I have no time for weekly reviews." Andrew had none either. Once the numbers had a rhythm and a shared sheet, his Monday meeting shrank from an hour to fifteen minutes. There was nothing left to argue about.

Key business metrics for startups: rhythm without history

A new business has no record of how fast things go wrong. The two rules still work. Guess the fastest change on the high side and start with the quicker rhythm.

Leading indicators matter most here. Elena, a freelance designer, launched her first book with no track record. Every week she checks visits to the book's page and the share who buy. Every week she compares her forecast with the real figures.

Where the rhythm does not reach

A rhythm treats one symptom: late discovery. You stop learning about problems at the end of the month, after the damage. That part works today.

The cause sits one level deeper. Rule one runs on the width of your yellow band, and a guessed band gives you a guessed rhythm. And a look, however frequent, still shows today. It says nothing about month 6.

Two steps close that gap.

• Zone thresholds and the red protocol. Where green ends and red begins for each number, and what you drop first when two rows go red in the same week. Chapter 4.
• Two forecasts and control by forecast. "If nothing changes" against "if the plan works", in numbers and dates, so you act before the needle moves. Chapters 8 and 9.

The rhythm needs a panel to look at. The five-row panel and how to build it are in small business KPIs.

Weekly, monthly, or too late

You have two choices. Keep checking the bank app every hour and the margin every year, and keep reversing prices after one bad week. Or add a "look every" column to your numbers tonight.

The rhythm tells you when to look. The thresholds, the red protocol and the forecasts tell you what each look means. They also tell you what to do when two rows turn red.

Picture next Monday: a short list of what to check this week, and a shorter list of what to leave alone.

Test the logic for free: Chapter 1, with the refusal test. Or get the full control system on Kindle, on Amazon. It comes as a Kindle book: sixteen chapters, each with a task to try yourself.

Published research

It is scientifically proven that the Jirgenson Model works. The published experiment, its record and its data:

• Zenodo (DOI): 10.5281/zenodo.23964934
• Hugging Face: E5 Jirgenson Model blind test
• Kaggle: E5 blind test dataset
• GitHub: experiment folder
• Wikidata: dataset Q142022866, author Q141981336
• ORCID: 0009-0001-3034-3552