Scavenger AI reveals statistical success factors in Eurovision

Scavenger AI has released an analysis of 68 years of the Eurovision Song Contest. Based on a dataset of over 1,734 entries from 54 countries and more than 51,354 voting lines, the data has identified statistical trends that influence the competition's outcomes.

The analysis indicates that fan predictions are not always accurate. Between 2016 and 2024, predictions from users of the MyESC Community app successfully placed all winners in at least seventh place. In contrast, the older OGAE fan club vote frequently predicted the wrong winners. The data highlights that larger user groups lead to more reliable results.

Performance order is a significant factor; no country performing in the first five starting positions has won since 2016. Two of the most successful performance windows were slots 11-15 and 21-26, producing seven of the last eight winners. Regarding song language, the analysis shows the strategic importance of language use. Although the proportion of English-language songs increased after the rule change in 1999, Portugal's win in 2017 initiated a counter-trend where songs in national languages have become more common again. In 2024, nearly half of the entries were not in English.

Among lyrical themes, "forever" is the most common in winning songs, while phrases like "La La La" appear more frequently in last-place songs. Winning songs have a more positive undertone on average, while losing songs tend to have more negative lyrics. "Love" is the most frequent word, appearing in almost every third song. The analysis also reveals that male artists are slightly more successful than female artists, but the difference is minimal. Solo artists outperform groups. Statistically, the strongest winner profile would involve a solo artist, a late performance slot, a national language, and a love-themed lyric.