Why it’s important to collect data on how people actually travel.

Working with sustainability means working with data. We collect it, analyse it, and model it. But not all data is created equal. In fact, all data is—at least partly—wrong. The key question to ask is: “Is this data right for the purpose at hand?”

One thing that consistently impresses me about corporate sustainability professionals is the sheer breadth of knowledge required to do the job. From environmental science to communications, from behavioural change to supply chains—you’re expected to understand it all. So, no, I don’t expect sustainability strategists to be commuting data experts. But there is one thing I do want you to know about commuting data: if you want to understand how people commute, modelled data or mobile network traces won’t do the job. You need data about how people actually travel.

Data for insight

If your goal is to understand commuting patterns—to identify where and how to encourage more sustainable modes of transport—you need insight. And insight doesn’t come from perfect data (because it doesn’t exist); it comes from the right kind of data.

That means data that tells us how people currently commute, including their mode of travel, the distance they cover, and the time they spend, and also preferably the route. It also means understanding where the biggest potential for change lies. This isn’t about just counting cars or calculating emissions. It’s about supporting decisions: what measures should we take, where should we focus our efforts, and what will actually make a difference?

It’s about enabling informed decision-making. Should you invest in bike infrastructure? Subsidise public transport? Run a mobility challenge? To answer these questions, you need to understand how your employees actually travel. And that brings us to a crucial distinction: modelled data versus actual data.

What is modelled data?

We focus on the case of commuting data for workplaces. Modelled data is essentially an informed estimate. It relies on statistical methods to predict how people travel, based on general information like where a company is located, how many people work there, and what kind of work they do. Modelled data can also be data from mobile networks on movement of mobile phones or other passively collected data (e.g. smart cards). Mobile phone network data for example, can show how many phones from a particular operator move between two areas and at what times. But beyond that, everything else is modelled similarly to classic modelling - it is not actual data on how people travel.

This type of data is very useful for different purposes, and when nothing else is available, can give you broad understanding of travel. However, when you're focusing on a single company or workplace, it quickly becomes too general. It lacks the detail and context needed to design meaningful, tailored measures, and it cannot be used to follow up on measures, because it is not actual data on how people travel.

To improve accuracy, modelled data is often based on and/or calibrated using actual data (from travel surveys). But a one-off survey—like a basic online questionnaire—isn’t enough to do this well. Researchers are now exploring how to better align mobile phone network data with more detailed GPS tracks collected through travel survey apps. If you're curious about that sort of thing, I’m always happy to nerd out over the details 😅

What is actual data?

Actual data shows how your employees really travel—when they leave, how they get to work (route and mode of transport), and where they’re coming from. Capturing this perfectly for all employees is basically impossible. But the most reliable way to get close is through well-designed travel surveys, either collected via mobile apps or through online questionnaires. While paper forms or phone interviews are technically possible, they’re rarely used in workplace settings today.

To be meaningful, the survey needs to reflect the full employee population. That means it must be neutral and unbiased. If the data collection is linked to a campaign aimed at nudging behaviour or encouraging change, it will attract only certain employees—and that skews the results. Surveys that passively collect travel data through an app, without requiring user interaction, tend to be more objective, collect more accurate data and can also collect data over a longer time span.

Even the best survey won’t give you perfect data. Not everyone will respond, and you’ll need to clean the data and account for gaps. But that’s part of data management—not modelling. What you end up with is grounded in real behaviour, not just assumptions.

Why actual data matters

When working with commuting at the company level, actual data is required if you want to make meaningful improvements to mobility options. Modelled data can give you a rough sketch. But if you want to design specific measures that work for your employees and you want to follow up on these, you need data that reflects actual behaviour.

Commuting behaviour varies significantly between companies, even those located in the same area and operating in the same industry. That’s because there are many factors influencing commuting choices, including cultural norms, organisational routines, and individual preferences—alongside geographical location and job sector.

The closest you can get to actual data is through travel surveys. These need to be properly designed to collect objective and representative data that can establish a reliable baseline and support meaningful follow-up measurements over time.

In short, if you want to act, you need to understand. And to understand, you need data that reflects reality—not just approximations.

All data is wrong. That’s okay—but the best solution for understanding and doing something to improve commuting travel is app-collected travel survey data.