available languages: english July 20, 2021

Today, we are going to take a look back at the German Federal Election of 2017 and reflect upon what has changed since then in the world of social media – and how German politicians are leveraging it for electoral success.

But first, let’s look at our Social Media Dashboard for the German Federal Election 2021 in collaboration with Der Tagesspiegel and few more infographics to enhance all our digital literacy.

Key takeaways

  • The AfD still has the social media edge due to its continued dominance of Facebook and followers sharing behaviour. However, the Greens and FDP are closing in looking across all platforms and all parties are vying equally for the social media space.
  • The most shared content comes from YouTube, on both Twitter and Facebook. It thus deserves attention, given the importance of cross-platform spread of information.
  • Anti-party and anti-candidate hashtags, as compared to topical/issue hashtags, dominate the 2021 campaign thus far, unlike in 2017.
  • Amongst these, AfD supporters are pushing hardest against the Greens, the CDU fellowship is pushing most against the AfD, and the Greens electorate leads in anti-CDU posts. There is little campaigning against the SPD and their lead candidate Olaf Scholz is not much mentioned.
  • Users making anti-AfD posts are focused on this campaign, the Green’s anti-CDU push is less single-minded, while the anti-Green posts include warning flags for suspicious accounts.

Dashboard infographics
Social media and the governing coalition: Not for lack of trying

In 2017 the AfD party dominated social media, but looking at the total number of recent Facebook, Instagram and Twitter posts by the major candidates, every party wants to contest the social media space – explore our dashboard in cooperation with Der Tagesspiegel and discover for yourself!

Infographic Exploration
Hashtags and Beyond

WordClouds allow us to quickly assess how often different terms are used. Here, you can see the most used hashtags from our data. This includes Facebook, YouTube, Twitter, and Instagram data from election, party and candidate searches (for our search criteria see here):

Here it looks like #Baerbock is doing quite well, while #Scholz is barely visible – but quantity is not everything. Let’s look deeper. This network map shows what other hashtags are used alongside #Baerbock:

Several of these terms have clear negative connotations, such as #baerbockverhindern (Stop Baerbock), #baerbockplag and #baerplag (reflecting a plagiarism controversy). Things are not looking as good for Baerbock as our first glance made it seem…

Next time: We will more deeply explore the words used around the candidates and what they mean.

Deep dive
Monitoring social media for #BTW2021 – Looking ahead by looking back

Before we zoom in on specific issues and actors, let’s take a closer look at what we can learn from the 2017 elections. Much has changed in Germany, and within social media, since then. By looking back, we can see developments over time and adjust our approach, ensuring we are not monitoring 2021 as if nothing has changed since the last election.

This is noteworthy:

  • The rapid growth of social media, coupled with data privacy concerns, favours early and targeted social media monitoring in 2021. To ensure coverage of more social media, we will coordinate and collaborate with others monitoring other social media.
  • As several key platforms are centred around easily shared images, the spread of electoral and misinformation campaigns across platforms warrants our attention. Moreover, we cannot ignore new, smaller, players that can also be vectors of disinformation.
  • Manipulation has become more sophisticated and often subtle – lies embedded in truths, nudging through language rather than outright deceit. This has made older approaches to classifying hate speech and disinformation less accurate. Thus, we track these phenomena, respectively, by monitoring and analysing how comments and threads escalate in emotional intensity, and by watching how the party’s and candidate’s agendas spread and shift.

Which platforms matter? 

Part of targeted social media monitoring during electoral campaigns is choosing which platforms matter based on the questions you want to answer. Social media is not all the same. Different platforms cater toward and are popular with different target groups and demographics, confronting the political parties with the challenge of tailoring their campaign to the platform and goal.

  • Even if Twitter receives a lot of attention, the size of its user base in Germany is only about one-sixth that of Facebook. Yet, given that around 25% of certified Twitter accounts belong to journalists, tweets and trends on Twitter are often picked up by mass media, magnifying their impact. Therefore, despite having fewer users, Twitter could be an important tool to set the agenda for political parties.
  • Instagram is another case: despite having fewer overall posts, it rivals other platforms in engagement (e.g., likes) and its photos and short clips often make it onto other platforms. YouTube has a similar dynamic. Audio-visual content may thus play a greater role in 2021 – neither Instagram (perhaps due to its newness) nor YouTube featured prominently in social media monitoring and analysis in 2017.
  • The increased ease of sharable content also highlights that while in-platform spread has traditionally been a focus for disseminators, allowing them to ‘game’ a platform’s algorithm, cross-platform spread (of misinformation) needs increased attention. Its impact is potentially more serious, given jumping platforms may mean that more people and different demographics are reached.

Facebook and Twitter shares of content from other platforms

In-platform sharing still dominates Facebook, but YouTube is the platform to watch for cross-platform spread.


Smaller platforms can influence certain parts of the electorate as well. Indeed, especially far-right groups and parties increasingly shun established social media outlets, accusing them of being in cahoots with established parties and media. It is important to monitor the share of users abandoning established platforms to join the newly emerging space of like-minded persons on other platforms, such as Telegram, gab.ai, and perhaps soon GTTR.

We will monitor for this “right flight” – or its absence – in the coming months.

Most shared links in our Facebook data

The AfD is still on top but targets you on Telegram – and a couple of right-leaning influencers made it into the top 14.

Is the online space a level playing field?

The ‘rules of the game’ for social media are found in algorithms that award the tweets and posts that receive the most engagement. Engagement – the ‘currency for attention online’ – is measured in retweets, comments and mentions on Twitter, as well as likes/reactions, replies and shares on Facebook, YouTube and Instagram.

  • When looking at social media campaigns, the most attention and engagement is usually accomplished when posts appeal to emotions, regardless of whether these are empathy and solidarity or outrage and fear. This path from emotion to engagement to attention immediately benefits certain parties and their campaign styles. Even bots and trolls contribute to this path, initiating community-building and inducing a band-wagoning effect. Some messages will perform better in these metrics even if they would not resonate far in the real world.
  • No matter whether inauthentic or authentic, the generated emotions push the posts with the most reactions to the top, granting them more influence in setting the agenda. While we cannot peek inside social media algorithms, we can break apart the message and messengers that trend.
  • As appeals to emotion are not inherently wrong, we are most concerned with appeals to hate or attempts to extreme polarisation. Yet classifying such things is hard, and even when possible, may require data one cannot legally retrieve or store (e.g., replies on Facebook). Thus, it is often best to monitor trends in behaviours and aggregate comments in threads, for instance, monitoring escalating discourses surrounding certain issues, actors, and their political affiliations.
  • When looking at 2017, the AfD was especially successful in generating engagement with users on social media platforms. By receiving the most attention both on Facebook and Twitter, it thus became more likely that their issues would reach larger audiences and be magnified through other media sources picking up their trending topics as the most prominent on social media.

For this year’s election, however, the AfD should be less able to dominate the social media discourse. As shown in the infographics above, the social media landscape has become a more equally contested electoral battleground.

Conclusion: The heat is on

This contestation can be seen by returning in conclusion to our hashtag style analysis from earlier. #Baerbockverhindern is not the only often seen hashtag centred on being opposed to something rather than being for something. Others include #grueneverhindern, #laschetverhindern, #noafd, #niemalscdu, #niemalsspd, and so on.

These consistently outperform issue-oriented hashtags, unlike in 2017. Their intensity of usage also correlates somewhat negatively with the polls – when a party looks to be doing well, the pushback comes. All indications are that this election is starting off heated.

Next time: We will examine if this competition also shows signs of turning negative. Let’s look at the position of anti-party hashtags in the overall hashtag ranking from Twitter:

The CDU/CSU, Grüne and AfD are all getting pushback with hashtags opposing them in the top 20. The top anti-SPD hashtag ranks 168th: apparently, they are not perceived as a threat, alongside (more expectedly) the FDP and Die Linke.

Let’s look at how these hashtags trended on Twitter overtime:

Anti-CDU hashtags seem to be trending while both the Greens and the CDU have an event-driven spike. Anti-AfD hashtags pushes are stable. Perhaps good for the Greens: Events can be somewhat controlled, whereas trends are harder to reverse.

Finally, let’s look at who is pushing against whom so we can identify where to watch for clashes in the coming weeks: