Since the introduction of the Melodifestivalen tour in 2022 did some sort of final playoff exist for those songs that finished 3rd or 4th in their heat. Originally the format of that was to have 8 remaining songs compete in one final heat, with two songs coming through to head straight to the final.
The first introduction of any Melodifestivalen duelling system came in 2007, a year which saw Sanna Nielsen and Sonja Alden make it through two head to head matchups to get the last remaining spots in the final. This knockout tournament style was the way the competition worked until 2013, when the decision was made to make a hybrid of the two systems, where a conventional Melodifestivalen eight song duel ultimately resulted in two head to heads with the top four remaining songs to decide those top two places. In the first year of this did we find the one and only Melodifestivalen winner to come via this extra round.
This remained until 2015, when the duelling head to head battle system returned for the entire show. What was different was the 2015 Melodifestivalen final had an expanded 12 songs competing for the ticket to Eurovision – so only one round of duelling was needed to find the four songs needed for the final show.
This has been the system that the show has had since then until now. While duelling hasn’t been a constant in the history of this so called Andra Chansen (literally meaning Second Chance), it is what is most thought of as the tradition of the show. Why has SVT decided to completely overhaul this round of the biggest show?
To The Semi Final
The rebranded Andra Chansen is now to be known as a semifinal. How it will work is that the eight songs left searching for those final four spots in the final will compete rather in duels, but in two groups of four songs. While the running order for the show will be produced by SVT, the division of which songs are in which heat was randomly drawn in a streamed show this Tuesday lunchtime.
The division of the Semi Finals is as follows:
Semi 1: Tone Sekelius, Alvaro Estrella, Danne Stråhed, Anna Bergendahl
Semi 2: Theoz, Lillasyster, Lisa Miskovsky, Cazzi Opeia
What this randomly drawing does is create the risk for chaos. Not only is it possible for all the 3rd place songs from each heat to be drawn together, it is also possible for the viewers to know about it. That is because in this year’s show through the heats the scores are shown age group by age group, so the final score and finishing position is visible to all those watching. Neither of these happened in the previous system – it always used to be the case that 3rd place songs played off against 4th place songs, even if we had to speculate which were which.
This year’s show has avoided some of that drama, as there is at least one 3rd place/4th place song in each Semi. What we do know is that we do have in each semi the same songs competing against each other that met in the heat – both Alvaro Estrella and Tone Sekelius from heat 2 are in Semi 1, and Cazzi Opeia and Lisa Miskovsky from heat 3 are in Semi 2.
The rationale behind these structural changes are many. Firstly, Swedish newspaper Aftonbladet has claimed many artists and record labels have been against the nature of the duels, as it creates an overly competitive environment on the ground. One of the aspects of duelling is the media, and ultimately the viewers at home, end up pitting song against song and artist against artist. However, unsurprisingly, whenever I have been on the ground at Andra Chansen the atmosphere between competing artists is extremely cordial. While the duelling is for competitive fun, the battling environment that goes alongside it is not desirable.
Secondly, SVT has acknowledged that these duels struggle to fulfil their purpose as competitive fun. The project leader for Melodifestivalen at SVT, Anette Brattström, explained to SVT news that the expansion to groups of four should make it “more exciting for viewers”. This is especially true in those cases that the head for head format has a predetermined favourite.
Such is the plethora of data such as Spotify streams, YouTube likes and so on, as well as now the actual voting data from each heat being known, there are more duels that have obvious winners and losers than those that warrant real speculation. In the voting system as used from 2019 to 2021, where the 7 age groups and the televote made up the final scores, more than half of the 12 duels that took place resulted in a landslide 7-1 or 8-0 victory.
And finally, part of this rebrand is that the term Andra Chansen doesn’t really reflect reality. If you as an artist don’t succeed the first time, the way the show is set up is that you get time to alter your stage performance to make it stand out even more. In reality, there’s barely a performance that looks any bit different from the heat they were in, barring perhaps the odd song getting some extra pyrotechnics. Melodifestivalen songs come into each heat as inch-perfect as they can be nowadays, so the mindset of altering the performances is incredibly rare.
That is why the term Semi Final has now been introduced, as it is more fitting that the show is a stepping stone to the final rather than an opportunity to do it all over again.
My only request would be…could this penultimate round not be called the Semi Finals instead? After all, Saturday’s show will have two groups rather than one big one.
The Impact Of This Change On The Results
There’s one major win in having groups of 4 rather than groups of 2, and that is the huge reduction in negative voting. We have previously researched the presence of superfinals – extra runoffs in National Finals between a top two or top three after the other songs have been eliminated. We found then that as a general rule when the superfinal changed the outcome from the first round of voting, that was to allow the song that was ‘safer’ do well. We can define safer in terms of musical genre or image for example – with superfinal formats tending to favour conventional pop ballad style entries.
Melodifestivalen had the same problem, and it has been difficult in recent years for anything alternative existing in a Melodifestivalen Andra Chansen duel to advance. Fans will be screaming at me to name Loreen’s ‘Statements’ here, but I’m also naming rock entries such as ‘Echo’ and ‘Pretender’ that got competitively obliterated as soon as they reached their respective head to heads.
If we go further back in history to when Melodifestivalen required songs to win two duels to advance, there are shocking figures to show just how this negative voting can be riled up. Heading to 2007 first, where Magnus Uggla’s ‘För Kung och Fosterland’ received 149,331 votes in defeating Nanne Grönwall in his first duel. Then, up against Sonja Alden ‘För att du finns’ Magnus received a lower score of 142,540. That is no large difference, true, but what is staggering is how Sonja Alden’s televote score skyrocketed. She won her duel in round 1 with 70,863 votes, before more than doubling the number of votes to creep over Magnus’ score with 143,980.
It’s unsurprising that Magnus’ song was a brash attempt at a faux-patriotic Eurovision number, whereas ‘För att du finns’ is a gentle encaptivating ballad that is far less likely to be off-putting. The following year saw similar shocking swings in the two duels that band Nordman won. In the first duel the band received a whopping 316,118 votes, yet in the next duel they still won, but with a far less impressive 166,735 votes. It wasn’t that the first duel opponent was so poor quality that meant this happened, but the opposite. Up against Nordman that day were Carola and Andreas Johnson, two stars of infamy many thought couldn’t be beaten, and Nordman’s ‘I lågornas sken’ picked up huge support as the underdog in that matchup.
This negative voting should die in this new system. What that means is that the voting system for our current show means that acts like Lillasyster shouldn’t see viewers vote for the alternative option just because they don’t like rock music, and neither should Alvaro Estrella accumulate votes for simply being the easy listening radio pop alternative.
The catch though is that the way this voting system is set up is that you really need to succeed in getting points from all the age groups. The points will be given out via the 7 age blocs and the televote as per the heats, but this time the scoring count is 12-10-8-5. In the Eurovision Song Contest, the count used there amplifies the points at the top of the scoreboard – the favourites – hence we give out 12 points to the highest ranked of the top 10. Instead, the Melodifestivalen 12-10-8-5 is effectively the same as 1-4-6-8 – and it is the song that is placed last that has the biggest negative drag.
This isn’t a surprise for anybody who knows the mantra of Melodifestivalen and knows the desire that it should be Hela Sveriges Fest – and the desire of the production team will be to find acts that appeal to the entire nation. But this suggests that the acts that score highly with older voters but do really poorly with younger voters, or vice versa, are overly at risk of losing out.
We have examples of these types of songs in these Semi Finals. Lisa Miskovsky scored top scores from every group above 45 years old in her heat, but finished last with the 16-29 year old group. Lillasyster were 1st or 2nd with every age group under 60, but bottom three with the age groups above 60. Tone Sekelius finished 2nd with four different groups but then ended up in last place with the youngest age category.
Below I try to model what the scoring could look like. To do this I ignore any differences in relative strength between the heats and rank the songs as they are placed in their heat. If there is a tie, I use a suitable proxy (YouTube, Spotify or iTunes depending on age) to separate the scoring. Furthermore, I have needed some poetic licence as Theoz and Danne Stråhed qualified through a heat where the app did not work, so all we know is that they finished 3rd or 4th in televoting in that heat.
Firstly, the model for Semi 1:
Modelling for the first semi final of Melodifestivalen 2022
The first Semi qualifiers as this model suggests would be Anna Bergendahl and Alvaro Estrella. But this is close, with Tone Sekelius just one point behind Alvaro in that quest for the final spot. As I explained above, this is a model not a prediction and personally I suspect Tone’s last place with the 3-9 age group to be a blip rather than a repeating pattern, and if Tone could beat even just Danne in that age group it would be enough to go through.
Note here that the model has Anna Bergendahl as the only act not to get any last place results.
And now the model for Semi 2:
Modelling of the second semi final of Melodifestivalen 2022
This model shows Cazzi Opeia as the qualifier of this Semi, with a tie between Theoz and Lisa Miskovsky. In such cases as a tie-break, the deciding factor is the raw number of votes – which we would anticipate to be Theoz with his younger fan base. This is on paper a tighter heat – and while Lillasyster lie last by virtue of those results in the televote and over 60s, I am going off very little to suggest that Theoz has a better result there and if not Lillasyster would creep into the top 2.
Again, I reiterate that this is a model, not a prediction, and you need many pinches of salt before using this data. What I try and show here is that the way the scoring works in this show in particular means that to get to the final you need to appeal to young and old voters. Do that and you will get the opportunity to perform on the biggest stage of them all to a full size Friends Arena.
This new format from SVT has a thumbs up from me. Immediately after the draw the speculation about who would go through and who would make it was a million times more exciting than a regular head to head format. It helps to give energy and life to the show and that helps to bring focus to these songs – some of which I am sure need more than one listen for you to work out their quality.