Which KPIs measure the success of a smart city?

Count the devices that are online and almost any smart-city programme looks healthy; ask whether residents are better served and the picture can change completely. Good smart-city KPIs avoid this trap by measuring changes in the outcomes that matter (service quality, resource use, resilience and people's experience). The amount of technology deployed is not one of them. They exist to tell decision makers whether a specific intervention works.

That calls for a small, balanced set of measures for each programme, each with a baseline, a documented calculation method and an accountable owner, used alongside a recognised city-level framework for broader comparison.

The city-level frameworks

Two international frameworks are worth knowing. The U4SSC KPIs were developed under the United for Smart Sustainable Cities initiative coordinated by the ITU and are based on Recommendation ITU-T Y.4903/L.1603. They comprise 91 indicators across three dimensions (economy; environment; and society and culture), divided into core indicators most cities can report and advanced indicators for more mature initiatives.

The ISO pair work together. ISO 37120:2018 defines indicators for city services and quality of life, designed to allow comparable and verifiable measurement for any city regardless of size, and ISO 37122:2019 adds indicators specific to smart cities, intended to be used together with ISO 37120. ISO lists 37120:2018 as due for revision, so the current edition should be checked before adoption.

Both are valuable for strategy, benchmarking and reporting. They are less useful for steering an individual project, which needs measures tied to its own purpose.

Activity is not outcome

Activity measures have their place in tracking delivery. They should never be presented as evidence that a service has improved.

Making a figure trustworthy

A KPI is only as good as its definition. Each one needs a stated boundary, meaning what is included and excluded, a calculation method and reporting period, a named data source with a way of checking its quality, a rule for handling missing data and an owner who answers for the figure.

The details matter because numbers can flatter. A shorter average response time can hide unresolved difficult cases if the calculation includes only closed requests. An apparent fall in faults may mean only that a monitoring system stopped reporting. Publishing enough context alongside the figure prevents misleading interpretation.

Beyond the average

Averages hide distribution. An improvement concentrated in one neighbourhood does not represent the experience of the whole city, and where services affect access to essentials, the KPI set should show accessibility and service availability for different groups.