We're going in circles

What Google's data tells us about our mobility

Berlin, May 2023. Author: Benno Bock

Google recently expanded its modal split time series to include data from 2022. This is a good time to make a data-based assessment of the modal shift. Because now we can see where the journey is heading with the modal shift according to Covid19 - for Germany, the data paints a restrained picture. Let us take a closer look at this data. In doing so, we should not forget a qualitative assessment of the new data jellyfish in terms of reliability and completeness.

Click here to go directly to the data in table form and to the data download.

The share price of the transport turnaround

Modal split 2018 - 2022 in major German cities
[in percent]

The trend: falling shares for sustainability

To get a feel for the development over time, we have averaged the modal split shares of the four main modes of transport in the 20 most populous cities in Germany. We can clearly see the Covid19 effects in 2020 and 2021: public transport shows a dip in use - at the same time we see increases in MIV (motorised private transport) and cycling. The trend reversal can then be heard in 2022. But is it enough? 2022 was not free of corona effects either, but the figures suggest that the use of the environmental alliance (i.e. cycling, bus and trains and walking together) will settle at a lower level than in 2018 after the pandemic.

Munich top, Bremen flop

A look at the time series reveals that in almost all major German cities there is an overall increase in the share of private transport between 2018 and 2022. Conversely, this means that the trend towards more environmentally friendly alternatives is decreasing. So there can hardly be any talk of a mobility turnaround where it is said to be most easily possible - in the big cities!

According to Google, the city of the 'bread roll button' performs worst: in Bremen, the share of MIV grew by almost 3 percentage points. Düsseldorf, Stuttgart, Cologne, but also heavyweight Berlin show significant increases in MIV. The cycling capital Munich stands out positively in the data: here Google notes a decrease in the share of private transport by 3.6 percentage points.

MIV share gains between 2018 and 2022
[in percent points]

Interim conclusion: Good to know, but is it true at all?

If the data is to be trusted, there is still a lot to be done in major German cities. That in itself should be common knowledge. But with Google's modal split, this can now be seen in black and white. Year fine. City fine. For everyone to look up or even follow, because next year is sure to come.

Modal-split à la Google

Where does the modal split data come from?

Catchment has taken the information processed here from the "Google Environmental Insights Explorer" - a portal that provides cooperating municipalities with important ecological key figures. Employees of municipal institutions can apply to Google for access. However, the data is only available to the general public as an "excerpt".

Therefore, this information should be treated with a certain degree of caution, but it represents the most comprehensive modal split data currently available in the public domain. In Europe, 7,624 locations are mapped. The modal split data goes back to 2018. Also good: All data are based on the same collection and processing method - so they can be compared over time and for different areas (albeit with limitations: see "A still young method").

Data providers: The many millions of users of Google services. All those who have ticked the "Google Location History" box can even view - and correct - their movement and modal split data in their own timeline via Google Maps. Modes of transport shown by Google in the Environmental Impact Explorer are car, motorbike, walking, bike, train, bus, tram and metro. The first two were grouped together above as motorised private transport and the last four as public transport (PT).

Completeness of individual modes of transport for the 20 cities

Still a young process

Particularly among the public transport data, there are certain gaps or presumably stochastic changes in the time series for individual modes of transport. Google itself writes about this: "These changes are the result of improved inference models that better distinguish between modes. Overall, these changes improve the accuracy and usability of the emission estimates in the long term." Indirectly, Google admits to taking the data with a grain of salt. The situation is similar with data completeness. For example, the bus or motorbike shares are not shown for every year. However, it can be assumed that every major city has a share of more than one percentile, which also rules out the possibility that these values are not listed because they are rounded down to zero.

Less cycling, more walking

If the difference between the last major mobility survey in Germany, the MiD 2017, is used, it becomes clear that there are some differences. For example, the bike shares in the modal split for Berlin, Hamburg, Munich, Stuttgart and Bremen are systematically underestimated by Google compared to the MiD. In our opinion, Google's recognition algorithm must be adjusted at this point, because parallel studies with Google Location History data show that this mode of transport in particular is corrected by the participants. Footpaths consistently occur more frequently in Google, because Google used factually tracked, firmly defined time and location data that are processed in the algorithm - regardless of how short the paths are.

Difference between Google shares and MID figures
[in percent points]

Similar to smartphone tracking surveys such as MotionTag, this survey method thus tends to record footpaths that in other, survey-oriented methods would rather be recorded as stages or path segments. Accordingly, the difference in the walking shares between MiD and Google may be due to the fact that these short path segments are simply forgotten by the respondents. The modal split of the MID only asks about trips that are defined by a main trip purpose (e.g. going to the bakery).

The modal split data from Google in tabular form

The data was collected for the 20 most important German cities over five years.