The Digital Services Act requires very large platforms and search engines to publish a repository of every ad they serve in the EU, not only political ones. Each platform built it its own way, with different fields, filters, countries and history. This guide sums up what the law requires, what each of the fourteen libraries Rastriq extracts publishes, and what is changing in November 2026.
It applies to very large online platforms and search engines (VLOPs and VLOSEs) and to all commercial advertising, not only political ads. The personal data of people who saw the ad must not appear.
| What it must publish | Detail |
|---|---|
| Ad content | Product, service or brand and subject of the ad. |
| Advertiser and payer | On whose behalf it is shown and who paid, if different. |
| Period | First and last date it was shown. |
| Targeting | Whether it targeted specific groups and the main parameters, including exclusions. |
| Reach | Total recipients and, where applicable, a breakdown by Member State of the targeted groups. |
| Retention | For as long as it runs and until one year after. |
Commission consultation, 4 November to 2 December 2026. The European Commission is collecting input for technical guidelines to make repositories easier to search and verify. It is not a new law or a single European archive: the obligation already exists.
European repository for political ads. A different regulation: its pilot opened on 12 October 2026 and covers political advertising only.
Uneven compliance. Independent reviews in 2026 find repositories with very different fields, search and history across platforms. That is why each actor marks in quality_flags what its source does not publish.
In Rastriq, the Microsoft, Amazon, Apple and Booking.com actors share one base block of fields, and the AliExpress, Pinterest, Snapchat, Temu and Zalando actors another, so each group stacks into one table. Meta, Google, TikTok, LinkedIn and X use their source's names.
| Library | Scope | Who paid | Reach | Targeting |
|---|---|---|---|---|
| Meta (Facebook, Instagram) | 40 countries | — | Spend and impression ranges in the EU | — |
| Google Ads Transparency | 38 countries | Verified advertiser | — | — |
| TikTok Ad Library | 33 countries | — | Estimated audience and impressions | Country, age and gender |
| LinkedIn Ad Library | Global | Yes | Impressions by country | Yes |
| X (Twitter) | EU-27 | Ad account and agency | Impressions, reach and frequency by country | Full |
| Microsoft (Bing) | EU + EEA | Yes, when Microsoft publishes it | Impressions by country | Types, with exclusions |
| Amazon | 9 EU stores | Yes | Reach by country and store | Methods |
| Apple App Store | 25 EU countries | Seller legal entity | Not published | Audience refinement |
| Booking.com | EEA | Yes | Impressions by country | Type |
| AliExpress | EU-27 | Seller and registered company | Impressions by country | — |
| EU-27 | Yes, when Pinterest publishes it | Reach by country, in ranges | Age, gender, interests and device | |
| Snapchat | EU-27 and Turkey | Yes | Impressions by country | Age, gender, language and device |
| Temu | EU-27 | Yes | Impressions | — |
| Zalando | 28 markets | Yes, when Zalando publishes it | Reach | Contextual or personal |
What Rastriq's actor returns from each library, according to its published README (12 November 2026). "—" = the actor does not return it.
A public, searchable archive of the ads a very large platform serves in the EU, with content, advertiser, payer, dates, targeting and reach per country. It is required by Article 39 of the Digital Services Act.
No. It includes all of the platform's commercial advertising. Political ads also have their own European repository, piloted since October 2026.
Those designated as very large online platforms or search engines (VLOPs and VLOSEs). Rastriq extracts fourteen today: Meta, Google, TikTok, LinkedIn, X, Microsoft (Bing), Amazon, Apple App Store, Booking.com, AliExpress, Pinterest, Snapchat, Temu and Zalando.
Few platforms offer a complete official API. Rastriq's actors on Apify turn each repository into a dataset you launch by API and export to JSON, CSV or Excel.
While they run and until one year after they were last shown. For a longer history you need to extract them on a schedule.