Most growing businesses eventually reach a point where the owner cannot personally know everything that is happening.
When the company is small, the owner might be involved in every sale, every project, every customer problem and every important decision. They do not need a dashboard to tell them how the business is doing because they are standing in the middle of it.
That stops working as the company grows.
People are hired. Managers take responsibility. Processes are created. The owner becomes further removed from the day-to-day work.
And that creates a perfectly reasonable question:
How do I know whether the business is actually working?
The answer is, at least partially, measurement.
Revenue, gross margin, close rate, utilization, leads, production output, defects, response time, schedule performance, customer satisfaction and dozens of other measurements can help leadership understand what is happening without personally observing everything.
Measurement is an essential part of delegation and accountability.
But it creates another problem.
Eventually, the organization can stop using the measurement to understand the goal and start treating the measurement as the goal.
Goodhart’s Law
There is already a name for this problem: Goodhart’s Law.
It is commonly summarized as:
“When a measure becomes a target, it ceases to be a good measure.”
The underlying idea is associated with British economist Charles Goodhart, who described the problem in the context of monetary policy in 1975. His original observation was that an observed statistical relationship could break down once pressure was placed on it for control purposes.
The familiar wording above came later. Anthropologist Marilyn Strathern used that generalized formulation in a 1997 paper about auditing and accountability in British universities.
So this is not a new idea, and it is certainly not an argument against measurement.
Goodhart’s Law is a warning about what happens when we confuse a measurement with the thing we actually care about.
The Measurement Is Not the Goal
Imagine a business that wants more sales.
Management believes consistent prospecting contributes to sales, so it starts measuring the number of sales calls each salesperson makes.
The original logic looks something like this:
Goal → Activity → Measurement
Profitable sales → Consistent prospecting → Calls made
There is nothing inherently wrong with that.
Then management creates a target:
Every salesperson should make 50 calls per day.
Eventually the system can start working in the opposite direction:
Measurement → Target → Behaviour designed to improve the measurement
The salesperson’s job quietly changes from generating good customers to making 50 calls.
Those are not necessarily the same thing.
People may shorten valuable conversations so they can get to the next call. They may contact poor prospects because those calls are easier to make. They may become more concerned about logging activity than advancing opportunities.
The dashboard improves.
The business might not.
Metrics Do Not Just Measure Behaviour. They Create It.
This is the part of measurement that leaders need to take seriously.
The moment you tell people something is being measured, you have changed the system.
The effect becomes even stronger when the measurement is connected to bonuses, performance reviews, promotions, rankings, reprimands or constant management attention.
People naturally pay attention to what the organization signals is important.
If management repeatedly says that 50 calls represents good performance, employees will try to produce 50 calls.
That does not necessarily mean anyone is dishonest or intentionally gaming the system.
They may simply be responding rationally to the system leadership created.
Before blaming employees for optimizing a metric, leadership should ask whether the organization taught them to optimize it.
That makes KPI design more than a reporting exercise.
It is part of organizational design.
Instead of only asking:
Can we measure this?
We should also ask:
What behaviour are we likely to create if we make this number important?
This Happens Everywhere
The sales call example is easy to understand, but the same problem can appear almost anywhere in a business.
Marketing
You want more customers, so you measure leads.
Marketing finds ways to produce more leads. Lead volume increases dramatically, but qualification deteriorates and the sales team spends more time sorting through people who were never likely to buy.
The lead dashboard looks fantastic.
Revenue does not move.
Customer Service
You want customers to get help quickly, so you measure tickets closed.
Employees learn that closing tickets is rewarded. Difficult problems get closed prematurely, customers reopen tickets and the number of completed tickets rises while the actual customer experience gets worse.
Operations
You want employees to be productive, so you measure utilization.
People become extremely good at filling timesheets and keeping themselves busy.
Whether that activity is producing the most valuable outcome for the business is a different question.
Project Management
You want projects completed on schedule, so you closely monitor percentage complete.
Now everyone has an incentive to describe their portion of the project as slightly further advanced than it really is.
The schedule can remain green right up until everyone suddenly discovers that the project is late.
Safety
You want fewer workplace incidents, so you reward teams that report zero incidents.
Unfortunately, you may have created two ways to improve the number:
- Have fewer incidents.
- Report fewer incidents.
Only one of those makes the workplace safer.
Start With What You Actually Want
A useful way to think about controls is to separate three different things.
1. The Outcome
What does the business actually want?
For a sales organization, that might be profitable customers who stay with the company.
2. The Behaviour or Process
What activities do we believe contribute to that outcome?
Perhaps consistent prospecting, good qualification, useful sales conversations and reliable follow-up.
3. The Measurement
What information can we use to understand whether those things are happening?
Calls, meetings, qualified opportunities, proposals, sales, gross margin and retention could all tell us something.
The danger begins when the third item replaces the first.
A measurement is evidence about what is happening.
It is not the reason the business exists.
Leading Indicators Are Useful, but They Need Context
Businesses understandably like leading indicators because outcomes often arrive too late.
Revenue tells you what already happened.
If revenue falls significantly this month, the underlying problem may have started several months ago.
That makes it useful to look further upstream:
Calls → Meetings → Qualified Opportunities → Proposals → Sales → Gross Profit → Retained Customers
You do not want to wait until the end of that chain before realizing there is a problem.
But there is a trade-off.
The further a KPI is from the outcome you actually care about, the more carefully it needs to be interpreted.
Increasing calls is easy to measure quickly.
Increasing profitable, satisfied, long-term customers takes much longer to observe.
You could dramatically increase the first number without improving the last one at all.
Leading indicators should therefore be treated as signals, not guarantees.
One Number Rarely Tells the Whole Story
Businesses are systems.
Changing one thing can affect something else.
That makes individual metrics particularly dangerous when they are viewed in isolation.
A salesperson could have fewer calls and fewer meetings while producing larger deals, a higher close rate, better margins and stronger customer retention.
Is the low call volume a performance problem?
Maybe.
Maybe not.
You need more of the system before you can understand what the number means.
The same problem exists throughout a company:
| If You Only Measure | You Might Miss |
|---|---|
| Revenue | Gross margin, receivables and customer retention |
| Leads | Qualification, cost per opportunity and eventual revenue |
| Production output | Defects, rework and waste |
| Tickets closed | Reopen rate and customer satisfaction |
| Utilization | Actual value created and unnecessary work |
This is where balancing metrics can help.
You do not need 50 KPIs to run a good business.
You simply need enough context that one number cannot create a completely misleading picture of reality.
Not Every Measurement Belongs on the Owner’s Dashboard
As businesses become more sophisticated, there is a tendency to keep adding measurements.
Someone discovers that a number can be measured, so it gets added to a report.
Eventually the owner receives a dashboard containing dozens of numbers, charts and colours.
More information does not automatically create more understanding.
It can be useful to distinguish between three types of measurement.
Core Controls
These are the relatively small number of indicators leadership regularly watches because they provide information about the overall health of the business.
Revenue, gross margin, cash position, qualified pipeline, customer retention or on-time delivery might belong here depending on the company.
Operational Metrics
These help managers understand and operate a particular system.
A sales manager may need more detail about calls, meetings, conversion rates and follow-up activity than the owner needs every week.
Diagnostic Metrics
These are measurements you investigate when something appears wrong.
If the close rate suddenly falls, leadership might examine lead sources, salesperson performance, response times, pricing, competitive losses and dozens of other measurements.
Those numbers can be extremely useful without requiring a permanent position on the owner’s dashboard.
The objective is not to see every number.
The objective is to know where attention is needed.
A Good Control Should Cause the Right Question
Suppose gross margin falls from 31 percent to 24 percent.
The control has already done something valuable.
It got your attention.
But the number does not necessarily tell you what to do.
Perhaps estimating has deteriorated.
Maybe material costs increased.
Maybe labour productivity fell.
Maybe the company took on a different mix of work.
Maybe one unusually bad project distorted the month.
Maybe the data itself is wrong.
A poor approach to controls says:
The number changed. Take this predetermined action.
A better approach often says:
The number changed. Look here.
A good control does not always give you the answer.
It helps you ask the right question. The 5 Whys is one simple way to follow that question deeper until you reach something useful to fix.
Employee turnover is a useful example: the number can tell you people are leaving, but it cannot tell you whether the cause is compensation, a manager, one employee, a broken system or the owner.
Metrics Have a Shelf Life
There is another danger with business measurement: once a KPI exists, it tends to live forever.
Imagine a company where salespeople are simply not prospecting enough.
Management starts tracking outbound activity.
For the next six months, measuring calls is extremely useful. Behaviour changes, prospecting becomes consistent and the original problem disappears.
Three years later, the business may still be obsessing over the same call target even though the constraint has moved somewhere else.
Maybe qualification is now the problem.
Maybe capacity is constrained.
Maybe close rates are excellent and there is no reason to increase call volume.
The business changed.
The measurement did not.
Controls should evolve with the systems they are intended to monitor.
Periodically ask:
- Why are we measuring this?
- What decision does this measurement help us make?
- Does it still tell us something useful?
- Has the underlying problem changed?
- What behaviour does this metric create?
- Would we lose anything important if we stopped measuring it?
A system that continues simply because “that is how we do it” has stopped being a useful system and started becoming bureaucracy.
The Dashboard Is a Map, Not Reality
This problem becomes more important as the owner moves further away from the day-to-day work.
Eventually, leadership experiences more of the business through abstractions:
- Dashboards
- Reports
- Forecasts
- KPIs
- Management meetings
- Financial statements
That is necessary.
You cannot build a scalable company where the owner needs to personally observe every process.
But abstraction creates distance.
A dashboard can say everything is green while the employees doing the work know that a process is falling apart.
Both can be true.
The dashboard is measuring what someone decided to measure.
The people doing the work are experiencing the entire system, including the things nobody thought to put on the dashboard.
This is one reason leadership cannot completely separate the head from the hands.
Measurement should complement judgment, observation and conversation.
It should not replace them.
The Owner’s Measurement Trap
This is particularly important for owner-led businesses.
The owner often begins with too much direct involvement.
They know what is happening because they are involved in nearly everything.
Eventually they realize this is unsustainable and begin delegating.
Then something uncomfortable happens.
They no longer know everything.
The natural response is to demand more reporting.
That instinct is not wrong. A business that operates through delegation needs controls.
But the owner can move from one extreme to another:
Personally involved in everything → Managing everything through numbers
Neither is the objective.
The goal is to build enough visibility that leadership can understand where attention, judgment and intervention are actually required.
How to Design Better Business Controls
Before turning a measurement into an important KPI, ask a few questions.
- What outcome are we actually trying to create?
- Why do we believe this metric tells us something about that outcome?
- How far removed is this measurement from the outcome?
- What behaviour might people adopt if we make this number important?
- Could someone improve this number while making the actual business worse?
- What other measurement would reveal that?
- Who actually needs to see this measurement?
- What decision should this measurement help us make?
- Is this a permanent control or a temporary diagnostic measure?
- Should this number trigger an action or simply trigger a conversation?
Those questions create a much stronger control system than simply asking what data is available.
Controls Should Concentrate Judgment, Not Eliminate It
There is a temptation when building systems to try to remove human judgment.
If we can create enough rules, enough KPIs and enough dashboards, perhaps the business can simply run itself.
That is usually the wrong objective.
The purpose of a good business system is not to eliminate judgment.
It is to concentrate judgment where it has the most value.
An owner should not need to spend valuable time determining whether someone made 47 or 50 calls yesterday.
The system can track routine activity.
Human attention can then be directed toward larger questions:
- Why is our close rate declining?
- Are we pursuing the right customers?
- Why is gross margin deteriorating?
- Is this process still serving us?
- Where is the business becoming constrained?
- What should we invest in next?
That is where leverage starts to matter.
As an owner builds leverage through people, systems and technology, they no longer personally perform every action in the company.
They need controls that allow them to understand the business from a distance.
But the controls are there to help leadership see reality.
They are not reality themselves.
Never Confuse the Measurement With the Thing Being Measured
Goodhart’s Law does not tell us to stop measuring.
Quite the opposite.
A growing company needs good measurement because an owner cannot manage a larger organization through intuition and personal involvement alone.
But every KPI is ultimately a proxy for something the business actually cares about.
The moment improving the proxy becomes more important than improving the underlying outcome, the measurement begins to lose its value.
Build controls.
Measure the business.
Use leading indicators.
Create accountability.
But keep asking what those measurements actually represent and what behaviour you are creating by making them important.
A KPI is an instrument.
A dashboard is a map.
A control is a signal.
Use them to know where to look, and then apply judgment to what you find.
Sources and Further Reading
Goodhart, Charles A. E. “Problems of Monetary Management: The U.K. Experience.” 1975. Goodhart’s original observation was developed in the context of monetary policy and later became known as Goodhart’s Law.
Strathern, Marilyn. “‘Improving Ratings’: Audit in the British University System.” European Review, Vol. 5, No. 3, 1997, pp. 305-321. This paper contains the widely used generalized formulation, “When a measure becomes a target, it ceases to be a good measure.”
