The Swedish election results are coming out. It’s very close – one seat either way could make the difference.
As with all close elections, campaigns matter. Digital campaigns are an important part of that. And the use of digital ads are a piece of that.
Let’s start with looking at the use of Facebook/Instagram ads (data is only until 8th Sept, as FB doesn’t release data in realtime).
- Sweden Democrats were the highest spending page between 1st and 8th Sept (€101k)
- Centre Party (€95k) next, then the Social Democrats (€63k) and Christian Democrats (€53k)
- Centre Party spent most when you add all their candidates in. Also had the highest number of active advertisers (270+) and total ads run (3800+).
- Overall, the coalitions spent similar amounts. Left/centre parties spent €604k, right parties spent €647k.
Targeting strategies used by the parties:
- Almost everyone targets on educational level to some extent.
- The left coalition mostly targeted people with an interest in politics, social movements, environment, most parties also tried to reach people in ethnic minority groups (interests in Somalia, Iran, Iraqi Kurdistan)
- The Socialists went a bit mad targeting what they clearly thought were socialist-proximate interests (knitting, reggae, houseplants, St. Pauli FC, Lidl, LARPing)
- The Christian Democrats targeted people interested in “Villas”, “Boating” and “Beef cattle”
- The Moderate Party were trying to reach aspirational higher income voters – people looking for new property or in the market for a BMW X5 or Audi Q7 SUV.
- The Sweden Democrats only did a small amount of interest-based targeting (4% of ads), focused on reaching people who like cars and driving.
Most parties tried to “exclude” at least some audiences:
- The Socialists excluded people with an interest in the stock market from 40% of their ads and a third of their ads were not shown to people interested in “golf”.
- They also tried to avoid people with interests in more expensive outdoor sports (sailing, windsurfing, 4x4s), though also managed to accidentally exclude people with an interest in “Marathon (Greece)”
- The Centre Party was keen to avoid anyone interested in socialist-proximate causes (Cuba, Venezuela, Nicaragua) and right-proximate (Russia, Hungary, Poland). They also tried to avoid people who like camping.
Given their lack of interest-based targeting, most of their strategy for the Sweden Democrats was based on geography and reaching men (84% of their ads were targeted solely at men) and older ones at that (87% of their ads reached people over 40).
This appears to be the continuation of a trend we see where far right parties tend to run fewer ads and have more straightforward targeting methods. Some argue this is because FB’s ad delivery algorithm steers ads towards people most likely to respond. Others argue it’s because FB audiences lean right/older. Hard to tell without more empirical work.
Moving on to Google/YouTube (data includes the week to election day). Overall – a mixed pattern. The Greens and Social Democrats went big, but in different ways. Greens appear to have spent more on YT than on all FB properties. Social Democrats may have too (again, caveat: different dates). Unlike all the others, they put almost all of their money (€100k+) into search ads rather than YT.
Almost everyone else spent about a third to a half on Google/YT compared to their Facebook/IG spend. Worth noting we think this proportion is increasing across Europe over time (YT used to be a poor cousin compared to Facebook). We’ll all need to put more effort into understanding what happens on Google/YouTube.
- Both coalitions likely spent over a million euro each in the final week across Meta/Google ads.
- The availability of more targeting information helps with explaining the approaches of parties and candidates. It’s an improvement to be able to see who is and isn’t seeing certain types of ads.
- YouTube ads are still extremely hard to analyse. YT should require that political ads on their service include a transcript/subtitles.
- Way too many (20k+) ads for journalists and civil society to meaningfully analyse or hold to account. You just have to hope there aren’t too many problematic ones. Is that good enough?