01
What happened
Before the ban, the 62 political and politically adjacent pages we track launched about 114 political ads a week, recording 4M cumulative EU-wide reach weekly, with 38% of that reach from party and politician pages. Since the ban they launch about 51 a week reaching 1.6M EU-wide. That is 45% of the former weekly ad-launch rate, and every ad that remains runs with no political label or political-ad targeting restriction. Only the DSA's general ad disclosures remain.
What remains comes from unions, business federations, advocacy organisations and, above all, media outlets owned by political organisations. Not one post-ban ad comes from a party or politician page in the tracked registry.
02
Political ads per week
Weekly launches of classifier-identified political ads by tracked pages, the last twelve complete weeks. The pre-election ramp is visible from mid-August.
ad launches
EU-wide delivered reach
03
The party is silent; its ecosystem is not
Every tracked page with a documented party affiliation leans the same way: 8 organisations affiliated with Socialdemokraterna launched 242 unlabeled political ads since the ban, reaching 7.3M EU-wide, while the party’s own pages ran none. That is the grey zone in one line.
Post-ban ad launches and EU-wide delivered reach grouped by the verified type and documented affiliation of each tracked page. Reach is cumulative across ads, so the same person may appear more than once.
| Page type | Pages | Ads | EU reach |
|---|---|---|---|
| Union / advocacy organisation | 22 | 1,614 | 45.8M |
| Organisation-owned media | 5 | 867 | 19.9M |
| Government authority | 1 | 77 | 15M |
| Company | 2 | 42 | 243.3K |
| Registry affiliation | Pages | Ads | EU reach |
|---|---|---|---|
| No registry affiliation | 22 | 2,358 | 73.6M |
| Socialdemokraterna | 8 | 242 | 7.3M |
04
Who runs the unlabeled campaign
Post-ban political advertisers by EU-wide delivered reach. The right-hand figures show each advertiser’s audience delivered in Sweden: share female and share aged 55 or older. The demographic shares use only Swedish delivery cells. The business-side campaign reaches older men; the environmental and rights campaigns reach women.
Not in the ranking: Valmyndigheten, 77 voter-information ads reaching 15M. A public authority telling people how to vote is political messaging by the classifier’s definition, but it is not the campaign, so it is listed here rather than ranked against it.
05
How the ads were bought
Meta reports the audience settings selected by the advertiser as well as the audience ultimately reached. These are different measures: targeting describes intent; the demographic figures in the advertiser ranking describe delivery.
Denominator: 2,600 post-ban political ads with targeting data.
Who paid for someone else’s ads
Every ad in the DSA repository declares who paid and who benefited. For most, they are the same organisation. In 3 cases they are not: Scream Mediabyrå AB paid for 6 ads benefiting Fackförbundet ST, reaching 2.9M.
| Payer | Beneficiary | Pages | Ads | EU reach |
|---|---|---|---|---|
| Scream Mediabyrå AB | Fackförbundet ST | 1 | 6 | 2.9M |
| Stiftelsen Fritt Näringsliv | Timbro förlag och tidningen Flamman | 1 | 1 | 1.5K |
| Timbro | Timbro och Flamman | 1 | 1 | 674 |
Read this with care
Meta no longer classifies any EU ad as political, but the Digital Services Act still obliges it to publish every ad delivered to EU users, with delivered reach, demographics and targeting. Vipima collects that repository daily for 62 tracked Swedish pages. For pages already in that registry, the collection is complete since 2025-08-17, spanning the final pre-ban weeks and the entire post-TTPA era (15,123 ads to date), and Vipima preserves it beyond Meta’s one-year retention.
Each ad is classified as de facto political messaging or not by a language-model classifier measured continuously against a 144-case hand-labeled evaluation set (95.4% at three iterations per case). Charity fundraising and member recruitment do not count as political; issue advocacy, electioneering and mobilisation do. Reach figures are EU-wide cumulative ad reach: the same person is counted across multiple ads. Demographic shares use Sweden-only delivery cells. A split-keyword capture-recapture exercise suggests the registry finds roughly 90% of currently advertising political pages. That estimate is indicative, not a census: the two discovery samples are not fully independent, and the registry applies documented editorial exclusions. Ads whose only political content is in images or video are a known blind spot. More on methodology.
Political-ad figures cover ads launched since 2025-10-10 unless noted.
Every election gets the same treatment: repeated archive collection before the vote, classification with published accuracy, and a public report. The elections covered and coming up are on the home page. Working on one of them? Get in touch.
The next report, the day it lands
One email per published election report. No other use of your address, unsubscribe anytime (privacy).