InsightsPoint of View

From measurement to control

How a business turns operating information into better decisions

A dispatcher reviewing operating information while looking out toward the equipment yard.
By
Cam Myers
Content type
Point of View
Reading time
6 min read
Status
Author’s perspective
Topics
Operating Models

In brief

  • A useful measurement system connects a signal with an available intervention.
  • Feedback delays, normal variation, and the chosen objective shape the response.
  • Record expected outcomes and competing explanations before judging an improvement.

The argument

A dashboard can tell an owner that a truck is losing money every week. It cannot decide whether to change the route, raise the price, or stop taking certain jobs.

That gap deserves as much attention as the dashboard itself.

Businesses often invest heavily in collecting information and much less in designing the response. Reports arrive sooner. More numbers are visible. Meetings become better informed. Yet the same problems continue because nobody has connected a particular signal to a decision the company can actually make.

A useful measurement system helps the business change its behavior. It needs to show what is happening, support a sensible intervention, and reveal whether that intervention worked. In systems terms, that is a feedback loop. For an owner, it means the numbers lead to action and the next set of numbers helps judge the action.

The decision underneath

Which operating decision will this information help the business make?

Measure something the business can influence

Consider an illustrative midsized equipment rental company in the northeastern United States advertising across several service areas. Its marketing report shows that the cost per lead is falling. The agency recommends increasing the budget.

Meanwhile, the new inquiries are farther from the yard. Drivers spend longer on the road, delivery costs rise, and dispatch struggles to combine stops. The company gets more leads while earning less from each completed job.

The marketing measure is accurate. It is simply too far removed from the economic decision.

To decide where another advertising dollar should go, the company needs to connect leads with booked jobs, service location, revenue, and the additional cost of serving them. It may discover that the more expensive lead produces a better customer because the delivery fits an existing route or leads to repeated business.

That does not mean the advertising system needs to know every accounting detail before it can be useful. It needs enough information to distinguish decisions that are commercially different. A practical first step might separate nearby jobs from distant jobs and compare completed bookings rather than raw inquiries.

The right level of measurement follows the decision. More detail is useful when it changes what the business should do. Otherwise it adds reporting expense without improving judgment.

See the difference before trying to correct it

A total can hide the condition that matters. Average equipment utilization might look healthy while a few high-demand units are unavailable and another category sits idle. Cutting the entire advertising budget would be a poor response to a shortage confined to one equipment type.

This is an observability problem: can the business infer enough about what is happening inside the operation from what it can measure? The owner does not need that term to ask whether the report distinguishes a shortage of demand from a shortage of usable equipment.

There is a second question. Can the business do anything effective about the condition it observes? If it cannot change advertising, prices, purchasing, or allocation, a more detailed utilization report may have limited immediate value.

A measurement project should therefore identify an available intervention. If the report shows a profitable service area with spare capacity, who can redirect spending there? If it shows that a route is overfilled, who can change what the company promises?

The response must fit the cause. More advertising will not repair broken equipment. More equipment will not fix poor follow-up. An impressive dashboard cannot compensate for an incorrect explanation of why performance is falling short.

Feedback can arrive too late

Suppose customers inquire today, book next week, receive service later, and pay the following month. The final financial result is important, but waiting for it before making every marketing decision could leave the company reacting far too slowly.

The business needs earlier signals that have a tested relationship to the eventual result. Qualified inquiries and booked jobs can guide short-term decisions. Completed service and collected contribution can verify whether those earlier signals remain useful.

The timing of changes also matters. If management adjusts prices, advertising, and sales follow-up every two days, it may never learn which change helped. A slow result from last week's decision can arrive just as the business reverses direction again.

This is how a company can create instability while trying hard to improve. It responds to normal variation as though every movement requires correction.

Set a reasonable observation period based on the delay between action and outcome. Establish ranges that warrant investigation rather than treating every fluctuation as an instruction. Keep a record of the changes made, including when and where they applied.

Google's guidance on operational monitoring makes a useful distinction between observing internal activity and testing the behavior users actually experience.[1] The business equivalent is checking both that the booking application ran and that customers received valid bookings. A technically successful step can still produce a failed service outcome.

Build a way to learn from the intervention

Before changing an operating rule, write down what you expect to happen and why. That forces the reasoning into view. It also prevents a favorable result from being explained after the fact with whichever story happens to fit.

Where practical, compare similar locations, customer groups, or time periods while holding other conditions reasonably steady. A controlled test can be useful, but many businesses cannot run a clean experiment for every decision. They can still record demand changes, seasonality, capacity constraints, and other explanations that weaken a simple before-and-after comparison.

Be particularly careful when several departments claim the same improvement. A revenue increase may reflect pricing, customer demand, available inventory, and marketing together. Each department can have contributed without each owning the entire result.

The aim is to improve the next decision with the evidence available. False precision can be as damaging as no measurement because it gives management confidence in the wrong explanation.

Choose the economic target carefully

A feedback loop can perform exactly as intended and still pursue the wrong objective. A pricing system aimed at equipment utilization may keep every unit rented by accepting business that barely covers delivery and wear. A system aimed only at margin percentage may price out work that would have contributed useful dollars from otherwise idle equipment.

The objective needs to reflect the decision and the resources it consumes. When delivery capacity is scarce, contribution per available driver hour may be more informative than revenue per booking. When equipment is idle and delivery capacity is available, the relevant comparison changes.

That is why an operating measure cannot be selected once and forgotten. The business changes, and the constraint can move. Management should revisit whether the measure still points toward the outcome it cares about.

Giving an advertising algorithm more financial data helps only if the data describes the value the company actually wants to pursue. Automating a poor objective makes the error more consistent.

I would start with one consequential recurring decision rather than a wall of metrics. Identify the outcome, the signal worth watching, the action available, and the person authorized to take it. Decide how long the business should wait before judging the result and what would cause it to stop or revise the intervention.

Once that loop works, extend it.

A business becomes more capable when it can detect a meaningful problem, respond effectively, and learn from the consequences. The dashboard earns its place by helping that happen.

Source notes

Business examples and figures are illustrative and are not reported client results.

  1. Rob Ewaschuk. Monitoring Distributed Systems. Chapter 6 of Site Reliability Engineering, Google. The translation to business feedback and bookings is the author’s analysis.

For the leadership team

In practice

  • Start with one consequential recurring decision rather than a wall of metrics.
  • Name the signal, authorized response, observation period, and stopping condition.
  • Check whether the economic measure still reflects the business’s current constraint.

Questions worth asking

  1. What can the business change when this report identifies a problem?
  2. How will we distinguish the effect of our action from other explanations?

Weighing this decision now? Tell us what you are considering and we will say plainly whether we can help.

Discuss your business