Most Businesses Need Fewer Metrics, Not More

Reports grow. Somebody asks a question in a meeting, a number is added to answer it, and it stays. Two years later the monthly pack is forty pages, three people read the first five, and nobody feels able to remove anything because each item was requested by someone.

This is not a discipline failure. It is what happens when adding is easy and removing requires a conversation.

The cost of a metric

Every number on a report has costs that are rarely counted:

Attention. Reading capacity is fixed. Adding an item does not add a reader; it divides the existing attention more finely. Past a certain density, the marginal metric reduces the probability that any given number is noticed.

Maintenance. Definitions need upkeep. Sources change. Someone has to check it is still computing what it claims. Forty metrics is forty small maintenance obligations, most of which are not being met.

False positives. Watch enough numbers without pre-set thresholds and some will move interestingly by chance. Each spurious movement consumes investigation time and, worse, trains people to discount movements generally.

Dilution of accountability. When a report has many numbers, no one is accountable for any of them. When it has seven, each has an obvious owner.

Argument surface. Every metric is a potential dispute about definitions. More metrics, more disputes, and the disputes crowd out the decisions.

Why reports grow anyway

Adding is cheap and removing is political. Whoever asked for a number may notice its absence. Nobody notices the cost of it being there.

Comprehensiveness feels responsible. It looks careful to include everything. Actually, choosing what matters is the responsible act; including everything defers the judgement to the reader.

Software makes it trivial. Modern tools can produce another chart in seconds, which removes the natural friction that used to limit growth.

Fear of the question you cannot answer. Numbers get included pre-emptively, in case someone asks. This is the biggest single driver, and the answer to it is that "I do not have that here, I will get it" is a perfectly good response.

Nobody owns the report as a whole. Individual sections have owners. The overall document does not, so its total length is nobody's problem.

How to cut

Start from decisions, not from the existing report. List the recurring decisions this audience actually makes — monthly, quarterly. For each, name the numbers required. Build the new report from that list. Anything not on it goes in an appendix.

Impose a hard limit and make trade-offs explicit. One page, or ten numbers, or whatever fits your context. A limit forces the ranking conversation that would otherwise never happen. Adding something means removing something.

Separate the standing report from the investigation. Most of what has accumulated is investigative — useful once, when a specific question was live. That material belongs in a place people can go and look, not on the page everyone reads every month.

Attach an expected range to everything that survives. A number with no expectation cannot be scanned quickly. With a range, a reader can process ten numbers in seconds and stop at the one outside its band.

Convert watching into alerting. Many metrics exist so somebody notices if they go wrong. That is a job for a signal, not a report. Moving them out is usually the largest single reduction available.

Set a review date. Once a year, remove anything that has not driven a decision. Making removal a scheduled routine rather than a proposal takes the politics out of it.

What a small set looks like

For most small and mid-sized businesses, one page:

• Cash position and aged debt
• Revenue and margin by service line
• Enquiries received by source
• Conversion at the one or two stages that matter most
• One operational quality measure customers would recognise
• One capacity or utilisation measure
• One or two items specific to your current priority, reviewed and replaced as priorities change

That last category matters. A small standing set plus two rotating items keeps the report both stable and relevant, and it creates a natural mechanism for retirement.

The objection, answered

The usual objection is that removing a metric means losing visibility. It is worth being precise here: removing a number from a report does not delete the data. It moves it from something everyone must read every month to something anyone can look up when a question arises.

The distinction is between what deserves standing attention and what should be available on demand. Almost everything belongs in the second category. Reports are for the handful of things where the business wants to be told without asking.

The instinct that more information leads to better decisions holds only up to the point where the information exceeds the attention available to read it. Most business reporting passed that point some time ago, and the fix is subtraction.