Everyone sees a different TikTok feed. For the first time, data from almost 700 German users are available for analysis of the „black box“ TikTok algorithm – the largest study to date.
Even at the height of the German federal election campaign, political content is only a marginal phenomenon on TikTok, a new analysis of the social video platform shows. Considerung all videos displayed to nearly 700 real German users, less than four percent of all clips in the average TikTok feed were explicitly political.
The data was collected and evaluated in collaboration with the public broadcaster BR, the Berlin Weizenbaum Institute, and the University of Zurich. According to Weizenbaum researcher Jakob Ohme, this is by far the largest set of real German user data that has been collected so far. The analysis therefore promises particularly realistic insights into the content shown to TikTok users. It also demonstrates that videos posted by politicians and political parties achieve a lower reach than those posted by influencers and other unofficial channels.
Left/green filter bubble
Until the 2024 European elections, only the far-right party AfD was considered successful on the platform. With its “Reclaim Tiktok” campaign, the Left Party in particular gained a large following among predominantly young users. Many election analyses identified this reach as a major reason for the success of the Left Party and AfD, especially among young voters.
Data analysis shows that the spectrum of left-wing/green users on TikTok is not exclusively addressed by the Left Party. Participating users were asked to share their voting intentions in the federal election. Voters for the Left Party, SPD, and Greens were also shown many videos from the other two parties—but rarely from channels affiliated with the AfD, CDU, or FDP. This suggests that TikTok users fall into at least two major political camps.
Below, we present a summary of the most important findings of the research.
TikTok is no longer just a place for right-wing populists. For the first time, the survey focuses on analyzing the left-wing space. Although the data set does not allow us to compare the influence of all parties on TikTok, we were able to identify more than 400 official accounts of politicians and parties as well as around 100 influencers to the left of the CDU/CSU in all participant’s feeds.
Politicians, parties, and influencers primarily reach those who share their opinions. There is always some mixing, and each voter group also has seen isolated videos from all other parties. We therefore cannot detect any absolutely distinct filter bubbles in our data. However, the political content on TikTok is not truly diverse.
There is a left-wing / green user bubble. Which politicians’ and parties’ accounts users follow and which party they voted for in the federal election is strongly linked to the content they see on TikTok. However, the boundaries between the Left Party, the Greens, and the SPD are fluid: for example, those who follow many Left Party accounts and voted for the Left Party usually also see Green or SPD content.
Politics is a marginal phenomenon on TikTok. Only 3.7 percent of videos in the average feed have a hashtag that indicates political content, and only one percent each are from a party or political influencer account—even though most participants have stated a high level of interest in politics. In our data set, men are shown political videos twice as often as women on average, but they also follow more party accounts than women.
Politicians and political parties do not dominate TikTok – even during election campaigns. Across all parties, at least 1.3 percent of videos in the average feed are from political influencers and 1.2 percent of all videos are from political parties. Only a sample of influencers with the widest reach were examined – in reality, political influencers and other political accounts are likely to be even more prominent in feeds.
Method and data set
In February, BR, Stuttgarter Zeitung / Stuttgarter Nachrichten, Weizenbaum Institute, and the University of Zurich called on TikTok users to donate their data. A total of 681 people donated their TikTok activity history and voting preferences to us.
The donors are predominantly female, between 20 and 30 years old, and well educated. Very few of them live in East Germany. Around 80 percent of the people surveyed vote for the Green Party or the Left Party. Many also vote for the SPD with their first vote.
The data set is therefore not representative. However, it provides a better insight into the political content consumed by real users on TikTok than any other survey available to date, as well as the first detailed analysis of left-wing and green content on TikTok.
To make the donors’ feeds comparable, we calculate the percentage of certain content (e.g., party videos) out of all TikTok videos that a person has viewed since January 1, 2025. With these percentages per person, we can see how a typical or average feed is composed – either across all donors or within individual groups of party supporters.