Why Citizen Satisfaction Scores Fail Public Services

A passport office can score 4.3 out of 5 on its exit survey while its complaints inbox fills with people who missed flights waiting for renewals. Both numbers are real. Only one of them tells you anything useful about how the service actually performs.

That contradiction is not a measurement glitch. It is what happens when agencies borrow a retail instrument — the satisfaction score — and point it at a transaction nobody chose to have. Citizens don't shop for a driving licence renewal the way they shop for a hotel room. They show up because the law, a deadline, or a life event forced them to, and when it's finally over, relief gets mistaken for approval.

Why do citizen satisfaction scores mislead public agencies?

Citizen satisfaction scores mislead agencies because they measure how someone feels at the moment a transaction ends, not how much friction they endured getting there. Government interactions are rarely discretionary — people don't choose to contest a fine, register a birth, or renew a passport — so a rating captures relief at completion more than judgment on quality. The fix is not a better survey question. It's measuring effort, outcome, and time instead of sentiment alone.

This matters because satisfaction data drives budget, staffing, and political attention. A department that reports 88% satisfaction has no incentive to fix the parts of the journey that never reach the survey — the three phone transfers, the form rejected for a missing stamp, the appointment booked six weeks out. My colleagues covering this exact failure mode in why citizen satisfaction scores mislead public agencies go deeper into the mechanics; the short version is that the metric rewards the wrong behaviour, and agencies optimise for what gets measured.

What makes a government transaction different from a commercial one?

A government transaction differs from a commercial one because the citizen has no exit option. In 1970, economist Albert O. Hirschman published Exit, Voice, and Loyalty, a study of how people respond to decline in organisations they can either leave or complain to. His argument was simple: when exit is available — you can switch banks, switch airlines — dissatisfaction shows up as churn, and providers feel it fast. When exit is not available — you cannot switch passport authorities — dissatisfaction has nowhere to go except voice, or silence.

Most public services sit in that second category. There is one tax authority, one land registry, one municipal licensing office. Satisfaction scores were built for markets with competition and choice; NPS and CSAT assume the respondent could have gone elsewhere and didn't. Strip out that assumption and the score stops meaning what the dashboard says it means. It becomes a measure of mood on a given day, filtered through relief that the ordeal is finished, not a signal about whether the service actually worked.

Why does the peak-end rule distort how citizens rate public services?

The peak-end rule distorts citizen ratings because people judge an entire experience by its most intense moment and its final moment, not by the average of everything in between. Daniel Kahneman and colleagues demonstrated this in a well-known study of colonoscopy patients, published as Redelmeier and Kahneman's 1996 paper in the journal Pain: patients' retrospective ratings of a procedure correlated far more strongly with the pain at its peak and at its end than with its total duration or cumulative discomfort.

Apply that to a benefits application that took six weeks, four document resubmissions, and two contradictory phone calls, but ended with a friendly officer processing the payment in under a minute. The exit survey lands right after that friendly minute. The citizen rates the ending, not the six weeks. Multiply that across a call centre and the aggregate satisfaction score looks healthy while the actual journey — measured in effort, time, and rework — is quietly broken. This is precisely the kind of distortion that behavioural economics is built to expose, which is why we treat it as a core input in behavioural economics work rather than a footnote to survey design.

What should agencies measure instead of satisfaction?

Agencies should measure the things that predict whether a citizen will need to come back, complain, or escalate: effort, first-contact resolution, time-to-outcome, and error rate. These are harder to game and harder to inflate with a smile at the end of the call.

The clearest precedent for this shift comes from the commercial world. In a 2010 Harvard Business Review article, Stop Trying to Delight Your Customers, Matthew Dixon, Karen Freeman, and Nicholas Toman reported research conducted for the Corporate Executive Board showing that reducing customer effort was a stronger predictor of loyalty and repeat contact than exceeding expectations. Their argument, built on data from thousands of service interactions, was that customers don't reward being delighted nearly as much as they punish being made to work hard. That asymmetry — effort punished harder than ease is rewarded — is loss aversion showing up in service design, and it applies with even more force where citizens have no alternative provider to defect to.

For public services specifically, the useful metrics look like this:

• Effort score at the transaction level — how many steps, documents, and contacts it took to complete the service, captured immediately after the interaction, not weeks later.
• First-contact resolution rate — the share of cases closed without a callback, repeat visit, or escalation.
• Time-to-outcome — elapsed time from request to resolution, not from request to acknowledgement.
• Rework rate — how often an application, form, or claim is rejected and resubmitted for reasons the agency could have prevented with clearer guidance.
• Channel-switching frequency — how often a citizen has to move from online to phone to in-person to get one thing done, which is itself a proxy for broken channel flexibility.

None of these require asking anyone how they feel. They come out of operational data the agency already holds, which makes them cheaper to collect and far harder to distort than a five-point exit survey.

How do you build a citizen-effort measurement framework?

Building a measurement framework that captures friction rather than sentiment is a sequencing problem more than a technology one. The steps below reflect what tends to work when agencies move away from satisfaction-only reporting.

1. Map the journey by life event, not by department. A citizen renewing a business licence doesn't experience "the licensing department" and "the tax office" as separate entities — they experience one continuous task. Map the full journey across agency boundaries before deciding what to measure inside it. This is the same logic behind life-event based service design in government, and it is the step most agencies skip because it requires cross-department cooperation.
2. Instrument effort at each handoff, not just at the end. Every point where a case moves between systems, teams, or channels is where effort accumulates invisibly. Log time-in-queue, transfer count, and resubmission count at each handoff, not just total elapsed time.
3. Separate "closed" from "resolved." A case marked closed in the system and a citizen who actually got what they needed are not always the same thing. Track resolution from the citizen's side — did the outcome match the request — not just the case-management status.
4. Set effort and time thresholds before you look at satisfaction at all. Decide what an acceptable number of steps, contacts, and days looks like for each service, based on the service's own complexity, before you introduce a sentiment layer on top.
5. Add sentiment last, and pair it with the effort data on the same case. Once operational metrics exist, a short satisfaction question adds value because you can now see where sentiment and effort disagree — a high-effort case that still scored well is worth investigating for what the staff did right.
6. Report the distribution, not the average. A mean satisfaction score of 4.1 hides the citizens who scored it 1. Report the shape of the data — how many people had a genuinely bad experience — not just the central tendency.

Agencies that have gone furthest with this approach tend to start with a structured baseline before touching survey design at all. A CX maturity assessment across the twelve building blocks of experience management shows, quickly, whether the organisation is even capturing the operational data this framework depends on — because you cannot measure effort you never logged.

Is trust a different metric from satisfaction?

Trust is a different metric from satisfaction, and it is often the one that actually matters for public services. Satisfaction is transaction-level and short-lived; trust is cumulative and shapes whether citizens comply voluntarily, engage with future services, or assume the system is rigged against them. The OECD's work on trust in government, drawn from its recurring cross-country Trust Survey, treats reliability, responsiveness, and fairness of institutions as distinct drivers from any single interaction's outcome — a citizen can be dissatisfied with one specific renewal process and still trust the institution overall, or vice versa.

The practical implication is that a bad satisfaction score on one transaction is not automatically a trust problem, and a good one does not automatically build trust. Trust accumulates from patterns — did the agency do what it said it would, did the rules apply consistently, was the process fair to people without connections or fluent paperwork skills. That is a service-design question as much as a measurement one, and it is why journey-level effort data, tracked over time and across services, tells you more about the trajectory of public trust than any single satisfaction number ever will.

How should agencies act on the data once they have it?

Collecting better data is wasted effort if nothing changes in how the service runs. The agencies that get value from effort and outcome metrics share a few habits.

• Route the data to the people who can fix the step, not just to a quarterly report. A frontline team lead needs to see next week's rework rate, not a board needs to see last year's average.
• Treat every rework case as a design defect, not a citizen error. If thousands of people submit the same form incorrectly, the form is the problem, not the public.
• Publish the effort and time metrics alongside satisfaction, not instead of it. Transparency about how long things actually take is itself a trust-building move, aligned with the same reliability signal the OECD's research points to.
• Redesign around the worst-performing segment, not the average citizen. Elderly applicants, non-native speakers, and people without digital access usually carry the highest effort scores; fixing their path improves the whole system.
• Close the loop publicly. When a measured friction point gets fixed — a step removed, a wait time cut — say so. It converts a metric into evidence that feedback changes something, which is the single biggest driver of whether citizens bother giving it at all.

None of this requires abandoning citizen feedback. It requires putting it in its proper place — as one input alongside operational effort data, rather than the entire measurement system. The UK's Government Digital Service Standard reflects this shift already: services are assessed against user needs, task completion, and iteration based on evidence, with satisfaction as one signal among several rather than the headline score.

The number that should worry a permanent secretary

A satisfaction score that keeps climbing while complaint volumes stay flat is not good news. It usually means the survey is catching people at the moment of relief and missing everyone who gave up before reaching it. The citizens who never finish the form, never get through on the phone, or quietly stop trying to claim what they're owed don't appear in an exit survey at all — they simply vanish from the data, which is the most dangerous place for a public-service problem to hide.

Measuring effort, time, and outcome instead of sentiment alone won't make that problem disappear. But it will put it back where a permanent secretary can see it, argue about it, and fund the fix. That is the whole point of measurement in public services: not a better number to report upward, but a true enough picture to act on. Building that picture — journey by journey, life event by life event — is slower and less flattering than a satisfaction dashboard. It is also the only version worth trusting.

Renascence works with public-sector teams to rebuild citizen measurement around effort and outcome rather than sentiment alone — if your agency's satisfaction scores and complaint volumes have stopped agreeing with each other, our citizen feedback management practice is a reasonable place to start that conversation.