Measuring the Organization

Idea one

Activity, and impact

Measurement is necessary because organizations are too complex to manage through individual observation alone. No director can personally see every credit-hold exception, every delayed repair, every customer contact about the portal, or every hour Support spends sending records manually. Measures create a shared view. But they can also create a distorted one.

The easiest things to measure are activities: calls answered, tickets closed, visits completed, invoices issued, meetings held. Activity measures are not useless. They tell you whether work is being performed, and they reveal capacity constraints, backlogs and failures of basic discipline. The problem begins when activity is mistaken for impact.

Support may answer ninety-five percent of calls within target. On the report, the team appears successful. But if customers are calling repeatedly because the portal still will not let them download service records, the fast answer may not have solved the customer's problem.

Activity asks, "Did we do the work?" Impact asks, "Did the work produce the outcome it was intended to produce?"

An organization that measures only activity can become very efficient at processing failure.

That is especially common where work is divided between functions. Finance tracks overdue debt and review speed. Operations tracks utilisation and response time. Sales tracks revenue retained. Technology tracks planned work delivered. None of these measures is unreasonable. Yet a company can achieve all of them while a customer's critical equipment stays down, nobody gives a clear update, and a repair that could have taken hours takes a day. Every function may be able to show a respectable report. The organization may still have failed.

The answer is not to abandon functional measures for one broad number called customer satisfaction — that would hide important differences. The task is to connect activity measures to the outcomes that cross functional boundaries.

Idea two

Start with the outcome, not the easiest report

A service manager may be pleased that engineers complete a high number of visits each week. But if repeated visits occur because the first engineer lacked the necessary part, diagnostic information, or authority, the completion figure may conceal customer inconvenience and avoidable cost. A high number of visits can mean strong capacity. It can also mean that the company is making customers wait through several handovers. The measure needs interpretation.

This is why good measurement starts with the outcome, not with the report that is easiest to produce.

Before choosing a measure, ask what the organization is trying to make true. For the portal, that customers can access accurate service records without needing manual support. For urgent repairs, that critical equipment is restored within an appropriate time while financial and safety controls remain effective. For credit control, that the company collects what it is owed without treating every operationally urgent case as a routine exception.

Once the outcome is clear, activity measures can be selected as supporting evidence rather than treated as the objective itself.

A single impact measure is rarely enough on its own. Time from verified failure to restoration reflects what the customer experiences, and a rapid repair achieved by releasing expensive parts to customers with no prospect of payment would create unacceptable financial risk. A repair completed quickly but unsafely would not be a success. So it needs supporting measures — the proportion of urgent exceptions decided within the agreed route, the value and outcome of releases, the number of requests returned because information was missing, the frequency of repeat exceptions for the same customer. Together, these measures show more than speed. They show whether the organization is restoring service while making proportionate decisions about risk.

And they must stay few enough to be understood. When organizations feel uncertain, they often respond by measuring more. Dashboards expand. Managers spend meetings explaining why one number rose while another fell. The volume creates an impression of control, but it can obscure the decisions that matter. A useful measure earns its place by helping someone decide, act, or learn. If a number does none of these things, it may be interesting, but it is not necessarily useful.

Idea three

A workaround can make a bad measure look good

Consider what happens when Support absorbs the portal defect well. If advisers manually send records quickly, customer complaints may fall. Leaders may conclude that the issue is under control. But the apparent improvement may depend on Support spending increasing hours doing work the portal was meant to remove. The customer may be receiving the record, through a process that is slower, more expensive, less consistent, and more prone to error than the intended service.

This is an important principle: a workaround can make a bad measure look good. The measure should reveal the workaround, not reward the organization for needing it.

So the useful figures are the ones that make the cost visible across the organization — the percentage of customers who successfully obtain records through the portal, the number of repeat contacts about the same request, the elapsed time from request to receipt, and the volume of manual work required to compensate for the system failure. They prevent the problem from appearing small simply because Support has become skilled at absorbing it.

The same risk applies to meetings. You could count meetings held, attendance rates, actions recorded. These are activities. The impact is whether decisions are made at the right level, actions are completed, repeated problems reduce, and work moves with fewer unnecessary handovers. A weekly meeting generating twenty actions while the same credit-hold issue stays unresolved for three weeks is not necessarily effective; a brief meeting that clarifies the escalation route and assigns a Finance deputy may produce fewer visible actions while creating a much better result.

Measures should not reward motion when the need is progress.

Which requires care with targets. Once a number becomes a target, people naturally begin to organize work around achieving it. That is not dishonesty. It is a predictable response to being judged. Measure advisers only on call duration and they end difficult conversations quickly. Measure engineers only on jobs completed and they avoid complex work. Measure Finance only on debt reduction and it resists justified exceptions. A measure changes behavior. The question is whether it changes behavior in the direction the organization actually needs.

Idea four

Numbers can mislead without being false

A misleading metric is not necessarily inaccurate. It may be calculated correctly, reported consistently, and presented with impressive precision. The danger is that it gives an answer to a narrower question than the organization believes it is asking.

Support answers calls within target. Finance has reduced overdue debt. Operations has increased utilisation. Technology has delivered its roadmap. Each figure may be correct. Together they may create the reassuring story that the business is under control while customers struggle to obtain records, urgent repairs wait for credit decisions, and employees use personal relationships to make work move.

The problem is not the number itself. The problem is the conclusion attached to it.

This often begins with proxies. A proxy is a number used because the thing that really matters is difficult, slow, or expensive to measure directly. You want to know whether customers receive helpful support, so you measure call handling time. You want to know whether engineers resolve problems, so you measure jobs completed. Proxies can be useful. No organization can measure every outcome perfectly. The danger comes when a proxy becomes treated as the outcome.

A short average call may mean advisers are knowledgeable and systems work. It may also mean advisers are ending difficult conversations before the underlying need has been met. One adviser confirms the portal issue is known, logs the contact and ends the call — the measure improves, the customer still lacks the record. Another verifies details, sends the record securely and explains what happens next — the call takes longer and the adviser may appear less efficient on the report. Treat call duration as the main measure of service and you reward the first behaviour and penalise the second.

Closure rates carry the same distortion. Closure is an attractive measure because it suggests completion. But a closed item is not always a solved problem. A ticket can be closed on a temporary answer the customer will call back about. A defect investigation can be closed because the cause is identified, though nobody has decided whether to fix it. A service job can be closed because an engineer attended, though the equipment needs another visit. In these cases, closure measures administrative progress rather than outcome.

So ask a more demanding question: what happens after the item is closed? Do customers contact you again about the same issue? Does the same equipment fail again? Does the request return as a manual task every month? A number becomes more useful when it is connected to the next part of the work.

Idea five

Local scores, and the cases at the edge

Misleading metrics also arise when people are measured on work they can control while the outcome depends on people they cannot. That division is understandable — it would be unfair to hold each person individually responsible for every result across the whole organization.

But the opposite arrangement can be equally damaging. If every function is measured only on its own narrow activity, it becomes rational for people to protect their local score even when doing so slows the wider journey.

Finance reduces overdue debt by holding accounts firmly — good discipline, and if the measure gives no visibility of justified urgent releases, critical equipment stays down where an exception would have protected the relationship and improved the chance of recovering the debt. Operations improves utilisation by keeping schedules full — and if every hour is allocated against the target, the company may appear efficient while becoming less responsive when customers most need it.

The answer is not to discard functional measures. Finance should not be judged only on repair speed. The answer is to place local measures beside shared measures that reveal the consequence of local optimisation. These do not make every team responsible for every action. They make it harder for one team to improve its own report by transferring waiting to another.

And then there is the last trap. Averages are useful because they simplify a large volume of activity. But they can conceal the cases that matter most.

Finance may respond to urgent requests in an average of two hours. That sounds reassuring. Yet the average may include many straightforward cases decided quickly, while a small number of genuinely critical requests wait eight or ten hours because the account is complex, the commercial input is missing, or the usual decision-maker is unavailable.

For the customer whose operation is down, the average is irrelevant.

Those cases may be few in number and high in consequence. They may involve important customers, safety documentation, contractual obligations, or customers whose frustration is growing with every contact. Leaders need to look beyond the central number. They should ask about the range, the exceptions, and the cases at the edge.

The point is not to turn people into numbers. It is to make sure that the numbers lead people back to the real work: the customer waiting for a repair, the adviser handling the same request for the third time, the warehouse supervisor holding a part, and the manager who needs to know whether the organization is truly moving work forward or merely recording that everyone has been busy.