Strava Metro (sports-)recreational cycling (blogpart 2)

Strava usage among (Sports-)recreational cycling

Strava owes its fame primarily to the many cycling enthusiasts who use it. In 2020, 512 million bicycle rides were recorded worldwide with Strava (STRAVA YIS 2020). That's almost half of the 1.1 billion Strava activities. Use in different countries varies greatly, but public Strava Heatmaps already show that the Netherlands has one of the highest densities of use. The UK, several European countries and US states also have high Strava usage:


This is confirmed by research from the NTFU (Dutch Recreational Cycling Union). In 2017, 62% of Dutch 'fanatic sports cyclists’ used Strava. In 2021, the Netherlands would have between 850,000 and 1.2 million sports cyclists, two-thirds of whom characterize themselves as 'fanatic sports cyclists’. If 62% of these use Strava, that already amounts to at least ~350,000 users. Among 'non-fanatic cycling athletes', the percentage of Strava users is not known, but this will certainly add several hundred thousand users.


But, not all Strava cyclists are by definition 'sports-cyclists’. For example, some Strava users only do 'commuting rides' or purely recreational bike rides without any sporting purpose. But in the Netherlands, that appears to be the smaller proportion of cycling Strava users.


In the Netherlands, we (track-landscapes) have access to Strava data within a few municipalities, including Amersfoort. Also based on this, we can give an indication of the number of users in the Netherlands (assuming that Amersfoort does not deviate substantially from the Dutch average). If we extend the number of local Amersfoort users to the whole of the Netherlands, this would amount to 750,000 cycling Strava users in 2020. The question is how exactly Strava defines 'local users'.


Demographic characteristics


There are some opportunities to get insight into age and gender of Strava users. In Amersfoort, where we have access to Strava Metro, the age distribution does show a characteristic picture. Although the group 13-19 and 65+ are relatively small, within those groups the biggest growth is visible since 2017. The 35-54 age group is actually decreasing slightly. Especially among young people, the use is increasing strongly. Among older (cyclists) this increase is much more limited in Amersfoort, only ~4% of cycling Strava

users are 65+. This is considerably less than the Dutch average (Wielersportmonitor 2019), in which ~18% of cyclists are 65+.

Not much is known about the male/female ratio of Strava users in recent years specifically in the Netherlands. But in Amersfoort we can give an indication based on the extent to which all streets of the city were passed by unique individuals. 20% of all cyclists counted by Strava (in 2020) were women, 77% were men (and 3% were of unknown gender). 44% of counts were generated by people 35-54 years old. 65+ and 19- both provided 3% of counts.


It should be taken into account that Strava men, in 2020 cycled about twice as many kilometers (2028km) as women (1101km). This is also a reason that men are counted a bit more often in this comparison. Thus, the Strava cycling population is likely to be slightly more than 20% women. With that, it does not deviate much from the distribution of Dutch sports-cyclists, which consists of 28% women and 72% men (Wielersportmonitor 2020).

And by the way, the percentage of women therein is increasing significantly, from 21% in 2016 to 28% in 2020. This growth is also strongly visible within Strava use worldwide (YIS 2020), and even stronger in the Netherlands. The number of female Strava rides in the Netherlands increased by 110% in 2020 compared to 2019. For men, that increase was ‘only’ 54%.


Both within Strava users and in general in cycling: women are on a very strong rise.

Representativity of Strava Metro (Sports-)recreational cycling

Based on the number of cycling sports in the Netherlands (0.85-1.2 million), the percentage that uses Strava (up to 60%) and the estimated number of cycling Strava users (~750,000), you can conclude that the vast majority of Strava cycling users consists of 'sports cyclists'. That is the group it represents, the usage of space by these cyclists can therefore be visualized very well with the Strava data. And since ~a million Dutch people have this hobby -representing ~30-40% of all recreational cycling activities in the Netherlands (source, source)- mapping their uses (and thus interests) is very relevant and interesting.

One can wonder, though, if Strava users represent the 'fanatic sports cyclists' more strongly than the 'less-fanatic sports cyclists'. Our impression is that this is not very strongly the case. For reference, the average Dutch Strava cycling trip in 2020 was 44.3km (a 24.6km/hour) for men and 33.5km (a 21km/hour) for women. These figures are almost identical to, for example the activity tracker 'Endomondo'. In both Strava and Endomondo, mountain bike activities are included (in Endomondo 20% of activities, in Strava not exactly known). So Strava does not show itself as the more competitive/fanatical app user in this comparison. The averages are not very fast or long.


Also, in amount of bike rides, Strava does not necessarily seem more competitive/fanatical than other averages. In 2020, Strava men recorded an average of 46 rides and women 33, which are lower numbers than for example the 2019 Cycling Monitor, in which sports cyclists averaged ~70 rides per year. In the cycling monitor, however, respondents were only counted if they undertook >12 sport bicycle rides annually.


To what extent is the 'under-representation' of certain specific sub-groups, such as '65+ers' or 'women' relevant for representativeness? It is certainly conceivable that these cycling enthusiasts may have different route preferences and habits. But that is precisely why Strava Metro is interesting: you can map the space usage of different age groups and also men-women, separately. To the extent that there is a difference in route preferences, Strava Metro allows you to examine them.


It is therefore to be expected that Strava Metro gives a representative picture of the space/route usage of Dutch cyclists.


But what we cannot yet conclude with this is that the Strava activity 'cycling_leisure' properly represents all recreational bicycle traffic. Sporty recreational cyclists may have different preferences and (route) usage than non-sporty recreational cyclists. Between 2012 and 2017, in the recreational 'Endomondo' data, we were able to distinguish between more and less sporty recreational cycling trips (based on speed and distance). The three images below show route use of Endomondo-using leisure cyclists (mountain bikers not included) in the Utrecht city region. A subdivision has been made between true sporty cyclists (activities with average speed >25km/h), recreational cyclists without sporty characteristics (activities with average speed <20km/h) and an intermediate group sporty/recreational (activities with average 20-25km/h).

Differences in use of routes are visible in several places. For example, semi-paved bicycle paths, or narrow bicycle paths were used relatively less by sports cyclists. These prefer to ride on asphalt, on wider roads. Non-sporty recreational cyclists, on the other hand, have a stronger preference for such paths. Sports-cyclists also cycle longer distances, reaching areas further from their homes more often. In relation with this, long, continues roads are used a lot by these sports-cycling.


A similar subdivision can currently not be made with Strava, Strava will therefore mainly show the sporty side of this spectrum; sports-cyclists.


Activity tracking data sportief en recreatief fietsen

Strava Metro; comparison with bicycle counts

However, it is difficult to make further statements about representativeness of Strava 'recreational cycling activities' based on comparisons with actual local bicycle counts, because bicycle counts do normally not distinguish between recreational and utilitarian traffic. Bicycle counts then almost exclusively show utilitarian bicycle traffic, as this constitutes 96% of bicycle activities in the Netherlands (recreational ~4%). Moreover, counting points are more often located within cities than outside them, where recreational bicycle trips mainly take place.


The only exception to this is the Province of Flemish Brabant. Here (by the Directorate of Space and Mobility) bicycle counts have been conducted at 13 locations which also characterized whether it was a recreational or functional cyclist. We compared these counts in 2018 with data from activity trackers Endomondo (we do not have access to the Strava data of that region). The ratio of recreational to utility use on bike lanes appeared to be very similar to the ratios from the local bike count in the Endomondo data. The left bar of each count point shows the ratios passages recreational/utilitarian of the local count at a count point, the right bar shows that ratio within the Endomondo dataset. And here, the Endomondo dataset was about 100 times smaller than the Strava dataset of today.



Of course, this is just one comparison. Moreover, cyclists in Belgium might have different characteristics than cyclists in the Netherlands, and Strava-using cyclists might differ slightly from Endomondo-using cyclists. For example, in Belgium, the proportion of cyclists within 'recreational cyclists' is likely to be (even) higher than in the Netherlands. And that should benefit such a counting comparison. More in-depth information of this comparison can be found in this document.


Conclusions Strava Metro (Sports-)recreational cycling

Both the comparison based on user/activity characteristics, and the counting point comparison, show that activity tracking data can be 'representative' if it is clearly defined which types of cyclists are portrayed. In the case of 'leisure cycling', Strava mainly provides a picture of sports-cyclists, within which sub-groups of age and gender can be distinguished.


It would be an added value if within the 'leisure' Strava cycling activities, the less sporty activities would be a separate group (column in data frames). Because although this group is relatively underrepresented within Strava, Strava is also used for this purpose. For example, based on route distance and speed, such a distinction can be made well. Strava data can then also be used to break down various route use differences between recreational bike rides of a more and less sporty nature.


Other relevant divisions are also conceivable. For example, the difference between solo cyclists and group cyclists. Here too substantial differences in route choices can be expected and it is very relevant for designing bicycle routes. A distinction in use by 'local cyclists' or 'cycling tourists' would also be very relevant and interesting.


It will therefore become more interesting and possible to compare the Strava Metro 'leisure_cycling' cycling data, with data from recreational cyclists from other studies/sources.


We are also making an open call to contribute ideas: do you know of more counting points in the Netherlands or abroad where 'recreational cyclists' have been counted? Or other research into the spatial behaviour and use of sporty/recreational cyclists? We would love to do more research. Then we can even better explain the values and opportunities of Strava Metro.


Would you like to know more about the ways in which we translate Strava Metro data and other activity tracking data, insights into sporty/recreational cycling into spatial development opportunities? In Amersfoort mapped the use of cyclists, runners and mountain bikers with data from Strava Metro. We translated this into spatial proposals. In Utrecht we mapped recreational and utilitarian bicycle rides using data from Endomondo.


Curious about the ways we translate Strava Metro data and other activity tracking data, insights into sport-recreational cycling into spatial development opportunities?

In Amersfoort, we used data from Strava Metro to map the use of recreational and utility cyclists, mountain bikers and runners, for a vision of route-based recreation in the region. In Utrecht we used data from Endomondo to map the shared interests of recreational and utilitarian bicycle traffic, see for example the blog beautiful-fast cycleroutes.