How much can digital data reveal about our cities?
We have more data about our cities than ever before. This can tempt us to focus primarily on answering questions for which data is available. However, we shouldn't lose sight of the pressing issues beyond that, says Esra Suel, professor of Urban Analytics.
Every day, millions of journeys, transactions, sensor readings and location traces generate new data about how cities work. Mobile phones track how people move through neighborhoods in real time. Satellite and street imagery capture urban change at a scale and frequency that would have required large teams of surveyors a generation ago. Digitalization has transformed urban research and policy. Questions that were once impossible to study because they were too expensive, too slow or too difficult to observe are now within reach.
And yet, some of the most urgent challenges facing cities remain difficult to address. Housing affordability. Reducing emissions from transport. Long-term social change. Who cities work for, and who they leave behind. Digitalization has not (yet) resolved these issues. In some ways, it has made them harder to address.
Data shapes what we study
As new datasets become available, research gravitates toward the questions they can answer — not always the questions that matter most. This is particularly visible with high-frequency data. When our best datasets update daily, hourly or seasonally, we become very good at studying problems that move at the same pace: congestion, footfall, transit delays, short-term changes in activity. But cities are also shaped by slower processes: densification, demographic change, climate adaptation, shifting land use. These unfold over years or decades. They do not show up as clearly in the digital traces produced by everyday urban life. If researchers study problems with the data at hand, some of the most important questions start appearing less often in research.
Big data is not always information rich
More data is not always better data. Take mobile phone data. They arrive with a compelling promise: millions of people, continuously observed. But volume is not the same as information content. From a mobile phone dataset, you often know where a device is — but rarely who is carrying it. Income, household structure, gender, caring responsibilities — this information is frequently missing or incomplete. Yet these are precisely the variables that explain why people travel the way they do. Traces often look identical whether a trip was a commute, a school run, or a care trip. If we observe fewer trips, we cannot easily tell what has changed: are people shopping less, socializing less, combining activities, or shifting to other locations? Without additional information, these interpretations require assumptions that are difficult to verify.
Mobile phone data is one example, but the pattern runs across much of the new digital data landscape — high in volume, often low in the kind of information that drives understanding. This is why surveys and censuses still matter. They are costly, slower, and typically smaller — but they were built to answer the questions we need answered. No amount of scale substitutes for that.
Urban data is increasingly privately owned
Traditional urban datasets were collected by public institutions to support research, policy and informed public debate. Censuses, travel surveys, public registers — were built for public purpose, open to researchers, policymakers, and the public.
New types of urban data operate differently. Companies collect it and sell to customers, sometimes at very high prices. Access is increasingly a business decision, and data is increasingly treated as a product, not a public resource. But cost is not the only issue — these datasets were rarely designed to represent a population. They reflect the users of particular devices, apps and platforms. Some groups are overrepresented; others are systematically underrepresented or missing altogether. As a result, private companies are increasingly shaping whose lives appear in the data, whose problems are studied, and whose interests are represented.
The real opportunity
None of this is an argument against digital data. Its scale, continuity and granularity are real advances — and we are learning a great deal by working with it.
The opportunity lies in combination. Digital data works best not as a replacement for what came before, but as a complement to it. Mobile phone data offers scale and continuity that traditional surveys cannot match. Surveys offer what mobile data cannot: the who and the why. Used together, deliberately, they can answer questions that neither could answer alone. There are also new possibilities for collecting the kinds of information we still need: app-based surveys, continuous cohort studies, administrative records enriched through digital tools. This is not traditional data collection as it used to be. It is traditional data collection rebuilt for what is now possible.
We cannot realize the true potential of this moment by simply following the data wherever it leads. We need to start with what we want to know and then decide which data, traditional or new, gets us there.
