Fairness has always been a hot topic in Eurovision debates, with different stakeholders emphasising different dimensions of the term. In our work, we focus on one specific and measurable aspect: fairness in qualification chances to the Grand Final.
Our lens is narrow but essential. We ask: How can we ensure that the allocation of countries and performance slots in the semifinals does not systematically advantage or disadvantage certain entries? Despite the EBU’s ongoing efforts to promote fairness – including the use of pots based on historical voting patterns and a televised allocation draw – our analysis suggests that structural imbalances can still arise.
Drawing on historical data and mathematical optimization methods, we explore where bias may persist and how a more data-driven approach could help level the playing field.
The Current Semifinal Allocation
We focus on two main drivers of an uneven playing field: the voting block effect, which is determined by the allocation of countries to semifinals, and the positioning effect, which is given by the running order.
In one of the earliest and most influential studies, Yair (1995) analysed 18 years of voting data using social network analysis and identified three voting blocs: the Western Bloc, the Northern Bloc, and a diffuse Mediterranean Bloc. These blocs were not merely statistical artefacts — they reflected persistent patterns in how countries awarded points to each other.
Since then, both the composition and influence of voting blocs have evolved. In response, Eurovision organisers have introduced measures aimed at reducing bloc voting and favouritism. Chief among these is the introduction of two semifinals and the pot-based allocation system, which groups countries with historically similar voting patterns and distributes them across different semifinals. As the official Eurovision 2023 guide explained:
“The countries…are divided into pots based on their historic voting patterns and then chosen at random…that traditionally award each other points are less likely to end up in the same Semi-Final…”
The pots for the Semi Final Allocation Draw in 2025
While this system successfully separates larger blocs, such as the Nordic bloc (Denmark, Finland, Iceland, Norway, Sweden), it can inadvertently create new fairness concerns. For example, smaller or unbalanced blocs — like Greece and Cyprus — may still frequently end up in the same semifinal. Even though they are always placed in the same pot, they have been in the same semifinal 7 out of 15 times since 2008. Or consider a four-country bloc in which three members are placed together, but the fourth is isolated in the other semifinal. From that country’s perspective, the draw may feel anything but fair. To make matters worse, some countries consistently receive disproportionately low support from certain others – even when performing with comparable quality – meaning that not all countries outside of voting blocks are equally neutral.
The draw also influences running order advantages. Producers can use the final performance slots to boost certain entries or place others early to neutralize potential advantages. However, this lever is constrained: the allocation draw also randomly assigns countries to perform in either the first or second half of the show – limiting the producers’ flexibility for half of the entries and adding another layer of luck to the process.
The running orders for the First and Second Sem Final of the 2025 Eurovision Song Contest
Take Montenegro’s placement in the Eurovision 2025 semifinals. As part of the broader Balkan bloc – which includes Slovenia, Croatia, Albania, Armenia, and Serbia – Montenegro saw its regional allies carefully split across both semifinals, in line with the allocation system’s goals.
If we consider the complete list of countries and running order, however, the semifinals tell a different story for Montenegro. Montenegro was assigned slot 2, a notoriously difficult place to qualify from. Meanwhile, Serbia, a fellow bloc member, drew a much more favorable late slot – one of the final two positions – often associated with a qualification boost. To make matters worse, Montenegro finds itself surrounded by countries such as Ireland, the United Kingdom, and Lithuania, which have historically awarded it few, if any, points.
In short, despite a balanced bloc distribution on paper, Montenegro faces both a performance disadvantage and an unfavorable allocation — a combination that significantly reduces its qualification chances, according to our analysis.
Building a Fairer Contest: Our Optimization Model
To explore this further, we developed a mathematical optimization model to evaluate and improve semifinal fairness, focusing on the two types of biases mentioned above: the voting bloc and the positioning effect. Each country is exposed to the sum of those two biases and our objective with this model is to minimize the difference in this bias across countries, which we call the bias gap. In short, we aim at minimizing the largest difference in received bias between any two countries in the same semifinal.
To be more precise, we let the total bias a country receives, biasCountry, be equal to the sum of the biases it historically received from other countries within the same semifinal, plus the historical position effect given by the running order. In the case of Serbia and Montenegro in the real semifinal allocation, this bias is e.g. biasSerbia = +3.21 and biasMontenegro = −12.32 [editor’s note. these are not points, as the model ranks 1-10, but it is in the same magnitude of points if that helps you understand the real-world implication].
The gap in these biases is 15.53, highlighting Serbia’s advantage given the more favorable position despite similar friendliness faced from other countries in this semifinal. The idea of our model is to determine semifinals where no two countries face a large gap in biases. To do this, we compute the maximum bias gain for each semifinal as:
and use a mixed-integer program to minimize these bias gaps, see Barz et al. (2015).
We estimate the biases between countries and the position effect by a regression model utilizing historical Eurovision voting data, enabling a data-driven approach to fairness. For the semifinals, we only detected the position effect when performing in the first three positions or the last two positions, which are those that feel the most bias. This solution with our optimized semifinals also leave the choice of who is in the middle to artistic experts, so that producers can still curate a show that has dynamics and different varieties of music throughout each evening of entertainment.
Bias Gap Reduced: What a Data-Driven Draw Can Deliver
Applying our model to the 2025 Contest, our optimized semifinal assignment yields a significantly lower bias gap, where no two countries would have a bias gap greater than 6.10. While the bias gap for the real semifinals exceeds 30 if calculated in the same way as above.
By using data to choose which countries compete in which Semi Final, and from there further choose the opening and closing songs of each Semi Final, we can reduce the bias of a country’s qualifying chances.
Unlike the current draw process, which prioritizes separating countries in the same voting bloc, our model takes a holistic view: it weighs all advantages and disadvantages simultaneously to reduce disparities in total bias. In some cases, separating bloc members may actually increase imbalance.
Consider a hypothetical case where half of the countries form a tight voting bloc and the other half vote evenly and impartially. Rather than forcing a split, it might be fairer to group the bloc into one semifinal and the impartial countries into the other — resulting in more internally balanced groups and a smaller overall bias gap.
Furthermore, instead of assigning the Big 5 and the remaining countries sequentially and then determining their positions, the model considers all aspects simultaneously and aims to ensure fair visibility for all participants.
Our model offers a concrete way to address imbalances in semifinal qualification by incorporating voting history and running order effects into a unified fairness framework. We recognize that Eurovision is not just a mathematical problem — artistic decisions, production logistics, and viewer engagement all play vital roles in shaping the show. Our aim is not to override these elements, but to provide a tool that supports them. By making structural biases visible and manageable, we hope to contribute to a fairer allocation process that complements, rather than constrains, the creative vision of the contest.
References
Barz, C., Ren, J., Sahin, O., Saure, D. (2025). “The Cost of Showmanship: Increasing Fairness in Eurovision’s Semifinal Allocation”. Working Paper.
Yair, G. (1995). “Unite Unite Europe”: The political and cultural structures of Europe as reflected in the Eurovision Song Contest. Social Networks, 17(2), 147–161.
About the Eurovisions Academic Conference
The eighth edition of the Eurovisions Academic Conference takes place at the University of Basel on Tuesday 13th and Wednesday 14th May this year. The event is free to attend both in person and online. This year’s theme is (Un)Fairness in Music – and the Eurovision Song Contest and there will be keynote speakers and panel discussions throughout the event.
Find out more on eurovisions.eu