A support team reduces the time spent on each customer query. The dashboard improves, yet customers continue contacting the company about the same issues. The shorter interactions may reflect a better process, or they may mean that a problem is being divided across several contacts. The headline number cannot establish which explanation fits.
This is the difficulty hidden inside familiar sayings about incentives. A clear reward can focus attention, but the measured activity may represent only part of the result the organisation wants. Improving that activity does not automatically improve the whole service.
An incentive therefore needs to be examined as a decision arrangement. What does it make attractive? Which responsibilities remain outside the calculation? How will the organisation know whether the response is useful? Those questions provide a better basis for design than assuming that a reward determines behaviour in a simple, predictable way.
What Does ‘You Get What You Incentivise’ Leave Open?
The saying encourages managers to examine consequences rather than rely solely on stated intentions. That is useful. If a support agent is praised for closing cases quickly and criticised for long conversations, the agent receives information about what the organisation values in practice.
The response is not predetermined. An experienced employee may continue investigating because they care about resolving the issue. Another may ask for guidance about complex cases. A third may focus narrowly on the measured target. Differences in skills, workload, judgement, and management support can all affect what happens.
For the support team, first define the intended improvement. The business might want customers to receive an accurate answer without unnecessary waiting or repeated effort. That objective includes speed, but it also includes resolution and the customer’s experience across contacts.
A handle-time measure can contribute to understanding that service. It might reveal an awkward system, a knowledge gap, or a repetitive step worth improving. The difficulty begins when the figure is treated as a complete definition of good performance.
The distinction matters because a measure can be useful without being suitable as an individual target. Data that help managers investigate a queue may become less informative when employees face strong consequences for every variation. The design should reflect what the number can reasonably establish about the work and the person being assessed.
The Proxy and the Goal Need Separate Checks
A proxy is a measurable indication of something harder to observe directly. Case closure is easy to count; a customer’s problem being resolved can require more investigation. That does not make the count worthless. It means the relationship needs testing.
Examine recent cases. Did customers return? Were they contacting the company about the same issue or a different one? Was the original answer correct? Did the employee have access to the information needed? A shorter contact can be a genuine improvement when a tool removes unnecessary work. It can be a deterioration when the customer is asked to start again elsewhere.
Changes in the case mix also matter. An employee handling simple requests should not automatically be compared with someone investigating difficult faults. A team’s average can improve because complex work has moved to another queue, leaving the total customer effort unchanged.
DORA’s guidance on software delivery metrics warns against treating a single measure as the whole goal and against comparisons that ignore context. Its setting is software delivery, but the measurement problem is relevant here: a useful indicator should support investigation and improvement rather than replace an account of the system.
The support team can use several kinds of evidence without turning each into another bonus target. Resolution checks, representative case review, repeat-contact patterns, and customer feedback can help interpret the headline figure. Their role is to test whether the apparent improvement means what the business thinks it means.
‘What Gets Measured Gets Managed’ Can Narrow Attention
Measurement makes some work easier to discuss and compare. Unmeasured work may remain consequential: explaining an unusual case to a colleague, correcting an article, identifying a recurring defect, or helping a customer understand a difficult decision.
If the team counts only completed contacts, time spent on those contributions can appear unproductive. Staff may need to choose between improving the underlying service and meeting an individual throughput expectation. The organisation should resolve that conflict rather than rely on employees to absorb it privately.
This does not require counting every contribution. A larger scorecard can create additional administration while preserving the same narrow assumptions. Managers also need to examine work qualitatively and understand why time was used in a particular way.
Research on intrinsic motivation and extrinsic incentives found that both could help predict performance, with differences related to the type of performance and the incentive arrangement. That is a more useful account than assuming a formal reward eliminates every other motive.
For the support team, recognise contributions to service quality while making their place in the workload explicit. Someone assigned to improve guidance should not then be assessed as though every hour remained available for contact handling. The responsibility, capacity, and evaluation need to describe compatible work.
Distinguish Adaptation From Deliberate Misrepresentation
People can respond to a target without intending to deceive. An agent might shorten explanations because managers repeatedly emphasise speed. That behaviour may be unhelpful, but it differs from knowingly recording an unresolved case as resolved to obtain a reward.
The distinction requires evidence. What instructions applied? How was the measure explained? Did employees receive examples of acceptable handling? Were managers aware that the target was difficult to meet alongside the service standard?
A design problem does not excuse every action. Deliberate falsification or repeated disregard of a clear responsibility can require an appropriate management response. At the same time, addressing one employee’s conduct should not close the investigation into an arrangement encouraging the same problem across the team.
Look for mechanisms rather than labels. A reopened case might count as a new case, making premature closure attractive. A performance threshold might penalise every long contact, regardless of complexity. A ranking might discourage staff from helping one another. Each finding suggests a specific change to consider.
Keep responsibility distributed accurately. The employee can explain their actions, the manager the instructions and assessment, and the measure’s owner the assumptions behind it. Assigning all responsibility either to individuals or to the incentive system would leave part of the account missing.
Test a Change Before Extending It
A revised incentive should start with an explicit expectation. The support team might reduce the weight placed on handle time, introduce representative quality review, and provide agreed capacity for recurring-problem investigation. The proposed effect is better resolution without an unmanaged increase in waiting.
Consider how the revised arrangement could fail. Quality review might become inconsistent between managers. A repeat-contact measure might penalise employees for defects they cannot fix. Team rewards might make individual contribution harder to assess. These possibilities should inform the trial rather than be discovered only after a full rollout.
Keep the interpretation of exceptions explicit. A long contact may reflect a complex case, an avoidable process, or a need for coaching. The revised arrangement should allow the manager to investigate which explanation fits rather than automatically reward or penalise the duration. Otherwise the organisation risks replacing one crude rule with another.
Review the customer’s journey across teams as well. A support target may improve because work moves to a specialist queue whose effort is not included. The trial needs enough visibility to identify that transfer and decide whether it represents a genuine improvement or simply a different location for the cost.
Involve staff in examining the design. They can often explain which cases the data omit and how a rule would operate during a busy shift. Their input is evidence about feasibility, not a guarantee that every preferred arrangement is suitable.
Keep a baseline and record material changes in demand, staffing, and case mix. If results improve after several changes, avoid attributing the entire result to the incentive alone. A system correction or additional capacity may have contributed.
The review should compare the intended outcome with the observed response. Did customers need fewer repeated contacts? Were answers accurate? Did waiting become worse? Could managers explain performance fairly? These findings can support adjustment, wider use, or a decision that the proposed incentive is not worth retaining.
Keep the Reward Open to Revision
An incentive that once made sense can become less suitable as the service changes. Automation may remove simple cases, leaving staff with a more complex workload. A new product may increase investigation time. Continuing to use the same threshold can create a different pressure from the one originally intended.
Review the underlying assumption as well as the target. Is the measure still related to the result? Can employees influence it within their responsibilities? Does meeting it create costs elsewhere? Are managers applying the same principles consistently?
The organisation should be able to explain a revision without presenting it as an admission that all measurement failed. Evidence-based adjustment is part of maintaining a useful arrangement. Staff need to know what changes, why, and how expectations will be handled during the transition.
Incentive sayings are useful prompts to examine what decisions the organisation encourages. Their limits appear when they are treated as complete predictions of behaviour. A sound design keeps the goal visible, tests the proxy, protects relevant responsibilities, and checks what actually happens when consequences attach to the number.
