Party attention: searches, Wikipedia and media against the polls
Polls measure support. This page measures attention alongside it: how often a party is searched on Google, viewed on Wikipedia and mentioned in online media. Both are set side by side week by week, nationally and, for Google searches, by state. Attention is not support: people also search out of rejection or curiosity.
The media presence series (GDELT) will be added with the next automatic run; until then the page shows Google searches and Wikipedia.
Share of attention and poll rating by party
The shares show how attention is distributed across the eight parties for each source (each column sums to 100 per cent). Next to them is the average of the Sunday question for the federal election.
| Party | Polls | Google searches | Wikipedia | Search minus polls |
|---|---|---|---|---|
| AfD | 28.0% | 54.7% | 28.9% | +26.7 |
| CDU/CSU | 19.9% | 15.7% | 8.3% | -4.2 |
| Grüne | 15.3% | 2.5% | 4.5% | -12.8 |
| SPD | 12.6% | 7.9% | 6.0% | -4.7 |
| Linke | 11.6% | 9.7% | 15.2% | -1.9 |
| FDP | 4.5% | 2.2% | 7.4% | -2.3 |
| BSW | 3.3% | 7.2% | 27.7% | +3.9 |
| Freie Wähler | 1.0% | 0.0% | 2.1% | -1.0 |
AfD takes 54.7% of the search interest across the eight parties and stands at 28.0% in the polls. Largest surplus of search share over poll rating: AfD (+26.7 points); largest shortfall: Grüne (−12.8).
How are the attention shares calculated?
For each source the weekly values of the last four weeks are averaged per party and divided by the sum of all eight parties. The poll rating is the mean of the latest federal poll of each pollster within 28 days.
Trend: poll rating and attention
Choose a party and a source: the line shows the polling average, the area shows attention as an index (highest week in the period = 100).
How is the attention index calculated?
The weekly value of the selected source is divided by the highest week in the selected period and multiplied by 100. The line shows the mean of the latest federal poll of each pollster within 28 days.
Does attention run ahead of the polls?
The comparison uses changes over four weeks: poll rating in points against the percentage change in attention. The value r ranges from −1 to +1; from about 0.25 a relationship is visible, below 0.2 it is treated as absent here. The shift states at how many weeks' distance the relationship is strongest.
| Party | Google searches | Wikipedia |
|---|---|---|
| AfD | r 0.25 · 2 weeks behind | r 0.20 · 2 weeks behind |
| CDU/CSU | no relationship | r -0.21 · 2 weeks ahead |
| Grüne | no relationship | r 0.23 · 3 weeks ahead |
| SPD | no relationship | r 0.21 · 4 weeks behind |
| Linke | r -0.22 · 5 weeks behind | no relationship |
| FDP | no relationship | r 0.25 · 5 weeks ahead |
| BSW | r 0.25 · 5 weeks ahead | r 0.28 · 5 weeks ahead |
| Freie Wähler | too little data | r -0.20 · 8 weeks behind |
The strongest relationship is BSW with Wikipedia views (r 0.28, attention 5 weeks ahead of the polls). 3 of 15 combinations reach at least 0.25; in 2 attention runs ahead of the polls, in 1 it follows them.
How is the lead or lag calculated?
For every week the change over four weeks is formed for both series. The Pearson correlation is calculated for shifts from eight weeks back to eight weeks ahead; the table names the shift with the largest absolute value. Overlapping four-week windows make the values look more certain than they are, so only clear values should be read as a relationship.
Which party is searched most where
Each party's share of search interest in the state (last 90 days, each row sums to 100 per cent). Search is the only one of the three sources with state resolution. Each row leads to the state page with polls and results.
| State | AfD | CDU/CSU | Grüne | SPD | Linke | FDP | BSW | Freie Wähler |
|---|---|---|---|---|---|---|---|---|
| Baden-Württemberg | 58.0% | 17.2% | 2.8% | 9.2% | 5.5% | 2.6% | 4.6% | 0.0% |
| Bavaria | 58.6% | 17.5% | 2.7% | 8.0% | 6.2% | 2.1% | 4.2% | 0.7% |
| Berlin | 44.2% | 16.8% | 3.6% | 11.7% | 14.4% | 3.2% | 5.9% | 0.0% |
| Brandenburg | 57.2% | 16.3% | 2.8% | 8.3% | 8.3% | 1.9% | 5.2% | 0.0% |
| Bremen | 53.4% | 16.9% | 2.7% | 10.0% | 8.2% | 3.1% | 5.7% | 0.0% |
| Hamburg | 55.9% | 15.3% | 2.7% | 10.1% | 8.2% | 2.6% | 5.1% | 0.0% |
| Hesse | 57.2% | 17.2% | 2.8% | 10.1% | 5.5% | 2.6% | 4.5% | 0.0% |
| Lower Saxony | 52.5% | 18.1% | 2.8% | 12.9% | 6.5% | 3.0% | 3.6% | 0.6% |
| Mecklenburg-Western Pomerania | 56.3% | 15.4% | 2.7% | 10.0% | 6.4% | 2.6% | 6.5% | 0.0% |
| North Rhine-Westphalia | 56.3% | 18.2% | 1.8% | 9.2% | 7.4% | 2.6% | 4.5% | 0.0% |
| Rhineland-Palatinate | 59.7% | 16.3% | 2.8% | 8.3% | 5.5% | 2.0% | 5.4% | 0.0% |
| Saarland | 52.2% | 18.9% | 2.7% | 11.9% | 6.4% | 2.4% | 5.5% | 0.0% |
| Saxony | 59.7% | 17.2% | 1.8% | 7.3% | 6.4% | 1.4% | 6.1% | 0.0% |
| Saxony-Anhalt | 63.1% | 15.5% | 1.8% | 6.4% | 4.6% | 1.4% | 7.2% | 0.0% |
| Schleswig-Holstein | 58.0% | 16.3% | 2.8% | 9.2% | 6.4% | 2.6% | 4.6% | 0.0% |
| Thuringia | 61.7% | 15.2% | 1.8% | 7.3% | 5.4% | 1.4% | 7.1% | 0.0% |
How is the search share by state calculated?
Google Trends reports for each state how search interest is distributed across the compared parties over the last 90 days. Two groups of five are linked through the AfD as a common yardstick; CDU and CSU are added together.
Reach on open platforms
Public counters of the party profiles, recorded daily since 1 October 2026. Only platforms whose figures can be read without logging in are covered: Bluesky, Mastodon, Telegram and the views of the latest YouTube videos. Instagram, TikTok, X and Facebook do not release their figures freely; the profiles are linked. Growth can only be shown once our own time series is long enough.
| Party | Bluesky | Mastodon | Telegram | YouTube: views per video | Videos in 30 days | Profiles |
|---|---|---|---|---|---|---|
| AfD | – | – | 23,360 | 76,439 | 11 | InstagramTikTokYouTubeXFacebookTelegram |
| CDU/CSU | – | – | – | 4,251 | 29 | InstagramTikTokYouTubeXFacebook |
| Grüne | 40,592 | – | – | – | – | InstagramXFacebookBluesky |
| SPD | 11,677 | – | – | 3,405 | 9 | InstagramYouTubeXBluesky |
| Linke | 24,041 | 22,515 | 12,300 | 76,186 | 15 | InstagramTikTokYouTubeXFacebookBlueskyMastodonTelegram |
| FDP | – | – | – | 7,744 | 15 | InstagramYouTubeXFacebook |
| BSW | – | – | – | 14,775 | 15 | InstagramTikTokYouTubeXFacebook |
| Freie Wähler | – | 138 | – | – | – | InstagramXFacebookBlueskyMastodon |
YouTube: median views of the channel's latest 15 videos. CDU and CSU are added together. A dash means: no profile recorded in Wikidata or no public counter.
Questions about the analysis
- How is attention measured?
- From three freely accessible sources: search interest on Google (Google Trends, parties as a topic rather than a search term), the views of the party articles in the German-language Wikipedia, and the share of articles in German online media that mention a party (GDELT). All series are aggregated to weekly values.
- How is the relationship with the polls calculated?
- For each week the change over four weeks is formed: in points for the polling average, and as the percentage change of the four-week mean for attention. The Pearson correlation between the two series is calculated, additionally with a shift of up to eight weeks in either direction.
- Does more attention mean better poll ratings?
- No. Attention also arises from scandals, party conferences, resignations or debates about a ban. The analysis shows whether the two move together, not why. A correlation is not evidence of a cause.
- Why are state values only available for Google searches?
- Wikipedia views and media mentions cannot be split by state. Google Trends reports search interest by state; shown is each party's share of the searches for all eight parties in the state.
- Why are Instagram, TikTok and X missing?
- These platforms only release follower and view counts through restricted or paid interfaces. Only open platforms are therefore recorded; the other profiles are linked.