The dashboard at a guidance office showed a rising line that looked like good news: more cases were closed in less time every week. The team celebrated until a worker reviewed conversations that returned. Many people received an automatic link, marked the request resolved because they could find no other option, then came back days later with the same question, sometimes through a different channel. The dashboard counted closures. It did not count returns, confusion, or the time someone spent rebuilding a conversation the system had already declared finished.
Speed is attractive because it offers a clear signal. A short queue, immediate reply, or large number of completed tasks can look like proof of progress. But progress means more than moving items from one column to another. It means a person understands what is happening, can take the next step, and does not carry afterward a problem the service has called resolved. When a metric sees only the moment of closure, it can reward the transfer of work to people with fewer resources to ask again.
Making the context clear#
The difference between activity and outcome becomes visible when a team follows a case through its full journey. A response may be sent in thirty seconds yet leave an access question, confusing document, or need for human support untouched. To know this, look at what happens next. Does the person contact the service again, use a different channel, abandon the process, receive an understandable explanation, or need another organisation to repair the effect? These questions do not reject efficiency. They place it inside the experience the service is meant to improve.
The difference between activity and outcome becomes visible when a team follows a case through its full journey.
One practical adjustment is to replace the word “closed” with more honest categories. A case may be answered, referred, confirmed by the person, resolved, or awaiting follow-up. The categories should not become a new bureaucracy of boxes. They should distinguish sending a message from completing help. At the guidance office, staff added one short question after certain replies: “Were you able to take the next step?” It was not a lengthy survey. It gave a signal about whether speed had helped or merely accelerated an incomplete exit.
What changes in practice#
It matters who defines success. If only the funder of a tool or the owner of an indicator decides, the result may reflect an administrative need rather than service quality. Including frontline staff, users, and receiving organisations reveals costs that do not appear in a dashboard. A worker may see that fast messages create more clarification calls. A community group may notice that a referral with no appointment leaves volunteers containing frustration. Situated experience is not an obstacle to measurement. It is part of what deserves to be measured.
An employment centre tested a reopening review. Every Friday it selected five cases marked complete that had returned through any route within two weeks. It did not hunt for blame or demand an individual explanation from each person. It compared the first message, the requested next step, and the reason for return. Many reopenings followed guidance to a website that required a digital certificate without explaining alternatives. The improvement was not to answer everything more slowly. It was to add a sentence about in-person appointments and an accessibility check before closing.
Deciding with information#
Repair measures complete this picture. When a recommendation sends someone down the wrong route or a criterion excludes a person, how long does the service take to notice, acknowledge, and correct it? An organisation that measures only output can hide these costs in the invisible work of people handling complaints. Recording repair time, number of affected cases, and clarity of communication does not punish unavoidable error. It helps decide whether apparent speed rests on a debt that will grow later.
Complexity is not a reason to leave everything open. Some questions are simple and benefit from a quick answer. The key is to separate by consequence rather than internal convenience. A question about opening hours can close after clear information; a decision affecting support, housing, or health needs stronger checking. Designing these differences avoids two extremes: a heavy review for everything or treating every case as if the same automated script could resolve it.
Reviewing and adjusting#
Dashboard language shapes behaviour too. If a board celebrates only “average closure time,” staff learn that asking once more is a delay. If it displays confirmed understanding, reopenings, and corrections beside that number, another conversation appears: what kind of reply genuinely saves work over time, where is extra support needed, and which task should not be automated? Metrics are not neutral. They teach which behaviour is valuable. Choosing them is a service decision, not statistical decoration.
Qualitative reading needs protection so that numbers do not become a new way to ignore people. Five carefully read cases can reveal a pattern that an average hides: people leaving after not understanding a term, families repeating data, or workers manually correcting an output every day. A short meeting can combine that reading with one volume measure and end with a small, dated, owned change. There is no need for an endless committee to discover that a sentence, route, or threshold is returning work to the person who came for help.
Transparency for affected people is part of progress. When a service uses an assisted response, it can explain that the first guidance is help in finding a route, not the final word on a case. It can say how to return, what minimum information helps, and how long a review will take. This clarity does not weaken the system. It reduces guesswork, keeps people from feeling guilty for insisting, and lets staff distinguish a resolved inquiry from an abandoned conversation.
A healthy review includes what a team decides not to measure. Counting how often somebody opens a notice, for example, may look useful but can encourage increasingly insistent messages without improving understanding. Before adding an indicator, ask what behaviour it will reward, who may be harmed, and whether a less intrusive measure exists. Sometimes the most responsible datum is the one not collected because it does not help anyone receive a better answer.
Progress is not the line that rises on a dashboard while doubts return beneath it. It is a service’s capacity to resolve clearly, recognise what it did not resolve, and repair what it sent in the wrong direction. The guidance office did not abandon fast responses. It learned to accompany them with a human route, an honest follow-up category, and a review of returns. The result looked less brilliant on a slide and became more useful in the lives of people who had to do something after reading a message.
This week, choose one speed metric and place five returning cases beside it. Ask what happened between the first response and the second request: was information missing, was the route inaccessible, did nobody know they could return, or did the issue need another kind of help? Choose one concrete repair and review it in two weeks. If the number of closures drops slightly while people understand more and return less with the same doubt, you have not moved backward. You have begun to measure progress where it can actually be felt.
