Ad library api

The ad library API ecosystem exists because of regulation, not because platforms wanted to build it. Three regulatory waves drove the buildout. Wave one (2018-2020): post-2016 election scrutiny in the US drove Meta to launch the Ad Library API as a political-ad transparency tool, with similar disclosure requirements in the UK, EU, and Brazil. Wave two (2022-2023): the EU's Digital Services Act (DSA) classified major platforms as Very Large Online Platforms (VLOPs) and required them to maintain queryable repositories of all advertising shown to EU users, including commercial ads. Wave three (2024-present): Brazil and other jurisdictions began ratcheting their own commercial-ad transparency laws, expanding the geographic scope of mandated disclosure.

The result: every major platform now ships an ad library API, but each one covers exactly the data the regulators forced them to expose - and nothing more. Meta's commercial API covers EU and Brazil because those jurisdictions require it. TikTok's API covers EU only. Google covers EU plus global political. LinkedIn covers EU only. The pattern is identical: platforms expose the minimum mandated data in the minimum mandated geographies and resist any further disclosure as competitive intelligence.

Practical consequence: don't expect API coverage to expand unless new regulatory mandates force it. Plan your pipeline around the current scope rather than betting on future expansion.

Meta's Ad Library API was the first to ship and has set the de facto template that TikTok, Google, and LinkedIn followed. Common architectural patterns across all four: business verification gating, OAuth-based access tokens, country-anchored queries via ISO codes, advertiser-ID-anchored search (cleaner than keyword), JSON response shapes with paginated results via cursor, per-token rate limiting at ~100-200 calls per hour.

If you've integrated with the Meta API, integrating with the others is a smaller lift than the docs suggest. The mental model is the same; only the field names and authentication flows differ. The cross-platform learning curve is steep on the first platform and shallow on the next three.

Where the platforms diverge: TikTok exposes more targeting transparency than Meta does (the Commercial Content API includes audience-demographic breakdowns for commercial ads in EU). Google exposes spend data more aggressively on political ads but is stingier on commercial. LinkedIn has the cleanest targeting transparency of any platform because B2B targeting is inherently more visible in the UI. Use the platform-specific strengths when designing your analysis layer.

A multi-platform pipeline has five layers. Adapter layer: one thin module per platform that hits the API, handles auth, manages rate limits, and returns normalized JSON. Normalization layer: maps each platform's response to a unified internal schema (platform, advertiser_id, creative_url, started_at, etc.). Storage layer: persists normalized data in a queryable store - usually Postgres or BigQuery for analytical work. Enrichment layer: applies post-fetch enrichment (hook tagging, creative classification, brand-aware tagging) - usually LLM-based in 2026, often using vision models on creative_url. Query layer: powers the actual end-user surface (dashboards, alerts, exports).

The hardest part is the adapter and normalization layers. Each platform's API changes its quirks roughly quarterly - field renames, response-shape tweaks, rate-limit adjustments. Adapters need to be resilient to these changes, which usually means defensive parsing (handle missing fields gracefully, log unknown fields rather than crash) and version-pinned API calls where the platform supports them.

The easiest part is the enrichment layer, paradoxically. Modern vision-and-language models (Gemini 3 Pro and similar) can tag creative metadata - hook archetype, visual style, CTA position, brand mark placement - from a single API call per ad. The cost is roughly $0.01-0.03 per ad; for a 10,000-ad weekly sweep, the LLM enrichment cost is $100-300 per week, dwarfed by the engineering cost of the other layers.

Internal: Facebook Ad Library API goes deeper on Meta's specific API; Foreplay deep dive covers how the leading commercial product handles cross-platform integration.

The build case is strongest when: you're a vendor building a product that itself depends on ad library data as core infrastructure; you have unusual data shape or coverage requirements that paid tools don't serve; you need full control over the data pipeline for compliance or data-residency reasons. The maintenance cost is significant - 1-2 engineering FTEs for a serious cross-platform pipeline, primarily absorbing platform API changes and adding new platforms as they come online.

The buy case is stronger for most internal marketing teams. Foreplay, Atria, Motion, AdSpy, and Minea all aggregate across multiple platforms (with varying coverage) and resell access at $50-500/mo per seat. The pipelines are professionally maintained and the tagging layers are mature. For most teams the buy decision pays for itself within a quarter when you cost out the engineering time you'd otherwise spend on adapter maintenance.

Hybrid pattern that works well in practice: use a paid tool for breadth (cover the full competitive set across all platforms at acceptable depth), use the official APIs directly for depth (custom analysis on your top 3-5 priority competitors where you need full fidelity). The combined cost is $100-1000/mo and covers the bulk of competitive intelligence use cases without requiring you to maintain the full pipeline yourself.