top of page

Strava Metro pedestrian data (2021, blogpart 4)


The rise and popularity of activity trackers was initially primarily due to runners. Although the application of activity tracking apps is becoming broader, the runner was and is the primary target group. Even with Endomondo, a widely used activity tracker that does not focus on a specific user group, about 50% of all recorded activities consist of 'running'.

Global figures are not known (to us) but in the Netherlands running is the most practiced sport, after fitness; ~16% of the Dutch population runs at least 12 times a year (2014). With lower regularity, many more Dutch people seem to run 'occasionally', ~31% of Dutch people do run at least once a year. Running is certainly more popular in urban regions than in non-urban regions. The Netherlands is estimated to have around 2.5 - 5 million runners, depending on the running frequency you set as a prerequisite. In 2014, ~85% did so 'unorganized', in the outdoors, mostly alone and sometimes with a friend/family member. Only ~15% did so in a running group or at a sports club.


Despite the huge number of practitioners -in public space- the runner has hardly been of meaning in spatial/recreation policy. Whereas walkers and cyclists in the Netherlands have been able to count on specifically signposted networks, facilities and infrastructure for decades, there are hardly any municipalities/provinces that take the needs of runners seriously. While the number of annual running activities (~150 million in 2014) comes close to the annual number of sport/recreational cycling trips (~190 million in 2016).


Strava usage among pedestrians

In the Netherlands, research has been conducted among participants of running events into the use of activity tracking apps (study Eindhoven, study Dam to Damloop). Approximately half of the runners appear to use an activity tracking app.


At the 2019 Amersfoort Half Marathon, where we have Strava Metro access, 36% of 21km runners used Strava, and 24% of 10km runners. The increase in Strava use is highly visible among participants of running events, at the DamtotDamloop in 2014 ~50% of participants used an app, of which 5% used Strava and 35% used Runkeeper.


Strava may be more known as a "cycling app," yet the number of running activities worldwide is almost as high as the number of cycling activities; approximately 480 million by 2020. That ratio has remained fairly even worldwide in recent years.


If the number of 'local users' of Amersfoort translates to the Netherlands there would be approximately 800,000 running Strava users in the Netherlands. The question is how exactly Strava defines 'local users'.


The age distribution of running Strava users shows an interesting picture: usage among teenagers (13-19) has increased very strongly since 2020. The age category 35-54 actually decreased. 55-65 also increased slightly. Representativeness seems to have improved in this respect; ages where use was relatively low have increased. Among 65+ers, little running is done with Strava, but this is generally recognizable in running. The KNAU/Mulierinstituut also conducted research on runners in 2014, the age distribution does not seem to differ greatly from Strava using runners. However, different units and age cohorts were used making it difficult to compare in one graph.


Walking

Although compared to cycling and running, walking/hiking is still relatively little recorded with Strava, this seems to be changing rapidly. In 2020, three times as many walking activities were undertaken/recorded within Strava worldwide than in 2019. This made walking by far the fastest-growing activity within Strava. Within Strava Metro, walking is currently still conflated with the activity running. Within Strava, walking appears to be done primarily with recreational motive, and to a lesser extent as a functional/transportation motive.


The potential of walking as an activity, split off from running, is undeniable to us. Recreational walking is by far the most frequently undertaken leisure activity in the Netherlands, with about 800 million activities per year. In many places in the Netherlands, walking is the backbone of tourist-recreational economic activity. But, knowledge of recreational walking behavior and use of walking routes, is still very limited.


Strava Metro representativeness pedestrian data


Several institutions have conducted research in the Netherlands on motivations for using activity tracking apps (most complete). The study sees several nuance differences in profiles and motives of app users vs. non-app users. However, the differences are not very large. It does show that older people use apps/wearables relatively less often.

Starting runners appear to see apps more often as a 'motivator' or 'health monitor' while more fanatical runners are more interested in the competitive functions of apps. Because different 'types' of runners appear to have such different reasons for using apps, the use of apps within different 'types' of runners is not very different.


It does, however, raise the question of whether Strava's distinctive competitive features (being able to compare times on segments) attract primarily competitive/experienced/fast/developed runners. The fact that Strava usage among 21km participants in Amersfoort was higher than Strava usage among 10km participants does support this thought.


Globally, men using Strava walked an average of 6.7 kilometers, and women 5.5 kilometers (Strava YIS 2019). Those are pretty much the same average distances as for users of other activity trackers like Runkeeper and Endomondo. They are not, on average, excessively long runs. And we would see those if Strava was only used by avid runners.


Running vs. Walking

The main limitation regarding the representativeness of the Strava (running) data, lies in the fact that running and walking are mixed in one dataset. This mainly removes the possibility of analyzing walking as a separate activity. For the vast majority of the dataset consists of running activities. In the Netherlands this will be at least 80%-90% of activities.


From previous studies with data from the app Endomondo, we know that this distinction is very relevant. Runners and recreational walkers have substantially different route preferences, spatial experiences and habits. In several Dutch cities -such as Utrecht, Rotterdam, The Hague- as well as smaller residential areas, recreational walking activities show a substantially different use of space/route than running activities. Runners more often walk along clearly followable, directive route structures, where as few stops as possible. They run on paths with a good surface, with a 'passable' width. Good asphalt roads, including bicycle paths and country lanes, are also frequently walked by them. Recreational walkers more often choose the smaller paths, and avoid traffic lights and crossings to a lesser extent. Narrow and irregular paths (farmland paths, towpaths, etc) are much more preferred than runners. The image below shows for example route use of runners and recreational walkers, who started their run from Utrecht. In many specific places, substantial differences are visible that seem to result from described differences in preferences:


Hardlopen en wandelen vanuit Utrecht obv Endomondo data
Hardlopen en wandelen vanuit Utrecht obv Endomondo data

But the difference in route use is not only the result of the different preferences. As with cycling, with running the distance to be covered is very decisive for route use/choise. With longer runs, other areas and routes come within reach. This strongly influences which directions we take, and which areas can be reached. Within the Endomondo dataset, running activities from Utrecht (starting point of activity is within urban contour) averaged 6.6 kilometers and walking activities from Utrecht averaged 4.3 kilometers. As a result, the picture of running has colored more intensively than walking (despite the legends being corrected for number of activities), and running activities are more prevalent in the green areas outside the city. Such differences were observable from all residential areas of Utrecht.

This influence of distance is also visible within a type of activity, for example running. In the figure below, running activities are divided by distance of the activity, in the city of Rotterdam. For short running distances, specific destinations are popular. As distances of activities become longer, use on certain routes increases strongly, and on others much less. For recreational walking, the same principle applies, but specifically sometimes on other route routes.

The interesting thing about these distance subdivisions is not only that the use of space varies greatly; it also shows the usage patterns of different types of runners, who also have different motives and preferences. And that is precisely what you want to take into account when designing routes, trails and surroundings.


Data hardlooproutes in Rotterdam Endomondo
Data hardlooproutes in Rotterdam Endomondo


Conclusions Strava Metro pedestrian data


Strava's 'run' data can currently be used to show a picture of the use of runners, especially in urban areas. This is a very interesting form of use; running is one of the largest types of recreational activities in the Netherlands, and probably in the rest of the world as well.

Strava Metro can play a crucial role in this; it is (one of) the most widely used activity trackers, and its use for walking activities is growing rapidly.

But to increase the usability of the Strava Metro 'walking' datasets, it is necessary to separate running activities from walking activities. They differ (too) much in spatial preferences and usage, and the proportion of walkers is still low in these datasets. Especially in and around urban areas this use is still buried under the much larger share of running activities.

In specific areas, which are more strongly known as 'out of town walking destinations', the proportion of walking activities will probably be higher. However, because it is not possible to identify the proportions there either, the application of Strava Metro 'foot' data is even more difficult here. Mixing the activities then leads to an unclear 'hybrid'.

If 'running' and 'recreational walking' are separated, this will benefit the representativity of 'running', and the potential of the group 'recreational walking' can be further investigated.

Within running, it could be further investigated which specific types of runners the Strava data represents to a greater or lesser extent. For the same reason, it would be valuable if within "running" and "walking" some subdivisions are added. Especially interesting would be a distinction based on activity distance, but also weekday vs. weekend activity. Based on the images of use that then emerge, other subdivisions and specifications will likely prove interesting.


Curious about how we translate these kinds of analyses into visions, strategies and design proposals?

Also in Brussels, data analyses from Endomondo, together with data from Runkeeper, were used to promote sportive use of public space. In Amersfoort, Strava Metro was used to show the (hard)running use of city and landscape.

תגובות


bottom of page