The peculiarities of a “landslide” #eurovision victory

For most Eurovision obsessives (a couple of merit badges above being a fan), Salvador and Luisa Sobral’s victory in Kyiv was a wonderful thing:

  • A country that hadn’t preivously ranked better than sixth stormed to victory on their 50th attempt
  • Their (non-English) national language without any significant diaspora or neighbourly support
  • No pyros. No costume changes. No key change.

Just a beautiful song, exquisitely arranged, and performed with élan and heart. Whether you loved, liked or loathed Amar Pelos Dois, most appreciate the sweetness of Portugal’s maiden victory: may they not have to wait another half a century for number two!

In 2017 the exception yet again somewhat disproves the rules: any delegation can win the Eurovision Song Contest. In the last decade or so we have seen another underperformer win comprehensively (Finland 2006) an allegedly “unpopular” Big Five country with a clear victory (Germany 2010), and a second victory after four decades for a country that’s struggled to qualify under the semi-final system (Austria 2013).  We are not arguing that it is equally easy for every delegation to win, but it is entirely plausible. And it will be interesting to see how this impacts the entries competing at Lisbon 2018.

This article drills down a bit into Portugal’s victory, and looks at data from the most recent detailed televote data we could find to examine how an individual participating broadcaster’s televote scores can be constructed—sometimes in surprising ways.

We wanted to understand how televote scores happen in the semi-final era. Suffice to say, we struggled to do so. Read on.

Jury’s Out

Amar Pelos Dois earned the top 12 points from 18 countries’ juries: Armenia, Czechia, France, Georgia, Hungary, Iceland, Israel, Latvia, Lithuania, the Netherlands, Poland, San Marino, Serbia, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom.  With 45 other countries voting in the Grand Final, that’s earning the maximum jury score from 40 per cent of the available delegation juries.  Impressive!

The jury scores are generated the same across the delegations. Each jury has five members, who each ranked the Grand Final entries from favourite (1st) to least favourite: 25th or 26th, depending upon whether their entry was in the Grand Final or not.  These rankings are added together and the lowest total score gets 12 points. Remember: when combining rankings the lowest score is the best. Let’s take a look at the 18 juries that awarded their top mark to Portugal:

Country # of jurors with Portugal #1 Other jurors’ rankings
Armenia 5
Czechia 4 (3)
France 2 (7,5,2)
Georgia 4 (3)
Hungary 3 (5,4)
Iceland 5
Israel 4 (3)
Latvia 4 (3)
Lithuania 5
Netherlands 3 (2,3)
Poland 3 (3,4)
San Marino 3 (2,6)
Serbia 2 (2,2,3)
Slovenia 3 (2,5)
Spain 3 (2,3)
Sweden 1 (8,7,6,2)
Switzerland 2 (3,6,5)
United Kingdom 2 (3,5,7)

Of these 18 countries only three (Armenia, Iceland and Lithuania) had unanimity in terms of ranking Amar Pelos Dois as all five jury members’ first favorite; another four juries had Portugal first with four outof five jurors. Only a slight majority of jurors from the other 10 juries had Portugal first. The Sobrals’ victory, while unambiguous, was not as massive as the impressions given during the voting sequence back in May.

Kyiv’s jury results sequence (Source: YouTube/Eurovision)

As an aside, only one Swedish juror ranked Portugal as their favourite. Next Swedish jury favourite Australia was also ranked first by a single  juror. The other Swedish jurors ranked It Don’t Come Easy 5th,8th, 4th and 6th; compared to 8th, 7th,6th and 2nd for Amar Pelos Dois. Adding up the five rankings for each entry and you get aggregate scores of 24 for Australia and 24 for Portugal. With only one first place ranking each, that 2nd ordinal ranking for Portugal was critical…just not critical to Portugal’s overall victory.

With jury memers offering ordinal rankings rather than raw scores, there’s no indication of how much #1 is preferred over #2—it could be razor thin or massive: a juror may loved their first choice, and struggled to rank their next four preferred entries, for example. Similarly, each juror’s top 3 could be a toss-up.  The rankings tell us who, but they don’t tell us how much.

Now let’s look at the televote results from 2017’s Grand Final. As much as we can, anyway.


There is a great deal of transparency with respect to the jury scores that is lacking on the televote side. The raw televotes from each country are not released; instead we get the points award from the televote after the broadcast and the televote rank. It would be excellent if the EBU released the granular televote data, even after a brief embargo.  So we do not know as much from the 2017 televote scores. Each jury has five members, which effectively weights each juror’s scores identically. These are rankings, not scores. Televotes produce actual scores and the number of voters is much much larger.

One reason we do not have access to the raw numbers at a delegation level might be related to the relative size of participants’ televotes. An entry will get the same 12 points  from winning the Maltese (population around 440,000) or Russian (population around 144,000,000) televote. A hundred million more people, but the same amount of points to award.

This is a good policy: it means that the Contest is not always determined by the televote results in the seven largest population participants whose populations add up to around 80 per cent of the total potential voting public: Russia, Germany, France, the UK, Spain, Italy and Poland. Quite literally the public’s vote in Malta would be worthless if their point allocation were aggregated into a pan-European televote, which would render their public’s opinion, proportional to population, as marginal. While releasing figures in terms of raw scores might underscore the differential, releasing the percentage of televotes received by each entry in each voting country would not.

For 2017 we know that Portugal won the televote in 12 countries: Austria, Belgium, Finland, France, Germany, Iceland, Israel, Lithuania, Netherlands, Norway, Spain, and Switzerland. But we have no indication of how much more popular Amar Pelos Dois was compared to the second ranked entry in each country. Given that some of the jury rankings were not skewed heavily towards Portugal it is likely that some of the televote rankings were much closer or much wider, since we are dealing with thousands or millions of voters in each televote, rather than 5 “voters” in a jury.

Kyiv’s televote results sequence (Source: YouTube/Eurovision)

In a 26 song Grand Final with 42 participating countries—bearing in mind you cannot vote for your own entry if yours is in the final—each country’s televote scores can be distributed among 25 or 26 countries. If these were distributed equally, each entry would receive 3.84% (if your entry isn’t a finalist) or 4% (if your entry is a finalist) of the televote in a given country. This is an almost impossible scenario, by the way.  Then these percentages would be converted into a ranking and the top 10 songs awarded their portion of the 58 points on offer.

If only we had some recent data to examine. Oh wait…define “recent”

Ciao ragazzi!

After scouring the internet, the most recent Grand Final televote distribution we could find was from Baku 2012 and Italy. Baku had 42 entries and a 26 song Grand Final.  Here’s the top 15 televote scores within Italy that year:

Rank Country Percentage of Televote Televote “Score”
Romania 28.32 12
Albania 12.92 10
Moldova 11.22 8
Serbia 7.39 7
Russia 5.23 6
Macedonia 3.66 5
Ukraine 3.13 4
Sweden 3.12 3
Turkey 2.79 2
Germany 2.43 1
Spain 2.38
Lithuania 2.36
Greece 2.28
Cyprus 1.86
Iceland 1.74

We excluded the bottom 10 rankings because we were already dealing with tiny percentages (i.e less than 2%) with several of the top 15 ranked entries.

For 2012 the Italian public only gave five entries more than four per cent support: Romania, Albania, Moldova, Serbia and Russia. Eventual winner Sweden was in 8th with a mere 3% of the televote!

In fact, the top three entries collectively took 52.46% of the Italian televote. Interestingly, the 10th and 11th ranked entries—Germany and Spain—were only .05% apart, a tiny amount.  In terms of relative popularity the entries ranked seventh through fifteenth were only marginally different: given an average (mean) percentage per entry in Italy would have been just under four per cent, half the top 10 were below the mean.

The Baku Grand Final was one where the jury and televote rankings were combined to generate a top 10 overall. With weak jury support, Romania (Mandiga’s Zaleilah) received 7 points, behind Albania’s 12 points (Rona Nishliu’s Suus), Russia’s 10 points (Buranovskiye Babushki’s Party for Everybody) and Germany’s 8 points (Roman Lob’s Standing Still). Germany’s 10th place televote rank indicates their jury score was very high.

Euphoria’s televote victory was 11 points over Party for Everybody, 343 versus 332 points. An 11 point difference from 41 countries is less than one third of a point—so one of the most sweeping Eurovision victories of all time was a marginal one compared to the second most popular (popular in the literal sense) entry that year. However the juries had Russia 11th, two hundred points behind Loreen, 296 versus 94 points.

If we had more delegation level televote scores from 2012 we could put together a more nuanced understanding of the results.

Even more transparency

It is worth remembering that the 58points scoring system is designed to skew scores in favour of the top two ranked entries. Effectively the top score is worth 50 per cent more than the third ranked entry (12 versus 8 points) and 20 per cent more than the second ranked one (12 versus 10 points). This reduces the chances of an outright tie, in terms of total points. The first Eurovision tie in 1969—under a different, less nuanced scoring system—led to five countries sitting out the following year’s Contest. The next tie, in 1991, occurred under the 58points scoring system with detailed tie-break protocol, though under today’s tie-beak protocol the 1991 runner-up would have won.

In 2017 we nearly had a tie, of a sort.  Salvador Sobral’s televote and jury scores were nearly identical: 376 and 382 points respectively—Salvador nearly tied himself. Amar Pelos Dois  was around 40 points ahead of Beautiful Mess, in terms of overall public support, but more than 100 points ahead with juries. If the televote and jury scores were combined using the same approach as 2012, some have argued that Portugal would have achieved the all-time greatest victory. We’ve not crunched those numbers, but you can check the ESCXtra folks’ math if you wish.

Releasing the percentage distribution for each country’s televotes would provide a comparable level of transparency to the jury scores. We agree releasing them immediately after the Grand Final has the potential to be something of a distraction: the winning team should always be given their unfettered moment of celebration. An embargo of  perhaps 30 days seems to strike a good balance between having a pause and maintaining the engagement of the general public.

The EBU offers its member broadcasters an online transparency tool. Releasing the granular televote percentage distributions would be an easy way to even further align the Contest with what are espoused values of the Union: the free flow of information, transparency and public scrutiny.

The current EBU general statute states that the one of the purposes of the EBU is to contribute in the field of broadcasting to:

1.2.3. enhancing the freedom and pluralism of the media, the free flow of information and ideas, and the free formation of opinions

As well, the EBU Code of Ethics states that:

In its interaction with Members, employees, public institutions and commercial partners, the EBU has to operate with the utmost transparency, effectiveness, legality and neutrality. It must be able to withstand the most rigorous standards of public scrutiny.

Free flow of information. Utmost transparency. Rigorous standards of public scrutiny. We would argue these are place already with respect to the jury scores: we know who the jurors are, their qualifications for nomination to their jurys and their comprehensive rankings of entries. We very nearly have it with respect to the public’s scores: we have the top 10 point allocation (and comprehensive ranking) for all three stages of the competition. Releasing the percentage televote scores would be a great next step.