Ad intelligence API vs Meta Ad Library API: which should you build on

An ad intelligence API is a paid, cross-platform feed of structured competitor ad data. The Meta Ad Library API is Meta's free, official endpoint for Facebook and Instagram ads. Pick the Meta API for Meta-sanctioned political ad data. Pick an ad intelligence API for commercial creative across platforms, historical snapshots, and agent-ready responses.

Both interfaces answer the same narrow question: what ads is this advertiser running, and what do they look like. They part ways on who is allowed to ask, what the answer costs, and how much of the answer survives.

What is the Meta Ad Library API?

The Meta Ad Library API is Meta's sanctioned way to query the public archive that powers facebook.com/ads/library. It returns the same records the browser shows, as JSON, through one Graph endpoint at graph.facebook.com/v19.0/ads_archive. Every call needs an access token plus at least one search axis: search_terms for keyword matching, or search_page_ids to target a specific advertiser Page.

Access is gated in three steps. You register a Meta for Developers app, verify your identity with a passport or driver's licence, and accept Meta's Ad Library API terms. HyperFX's developer guide, published April 25, 2026, puts typical approval at one to two weeks, with political research routed faster.

Server-side jobs should use a system user access token with the ads_archive scope. Those tokens do not expire mid-campaign and can be rotated independently of any one developer's account, which matters once a scan runs on a schedule.

What does the Meta Ad Library API actually return?

For ordinary commercial ads the record is creative and metadata: ad copy, images and video, the advertiser's Page name and ID, the countries the ad ran in, the publisher platforms, the active status, and the language. That is enough to see what a competitor is saying and where they are saying it.

It does not return performance. Impressions, spend ranges, and demographic delivery breakdowns are exposed only for political and issue ads. Targeting fields are never exposed. CTR, CVR, and ROAS are never exposed. ScrapeCreators' ad library API guide, last checked September 1, 2026, states the boundary plainly: an ad library API can show which creatives ran and when, but it cannot tell you whether an ad made money.

Freshness cuts the other way as well. Ads appear within minutes of going live, and inactive ads disappear at the moment they leave the browser. Meta maintains no historical archive for commercial ads. If you want a timeline of what a competitor ran last quarter, you must snapshot it yourself, on a schedule, before it vanishes.

Pagination is manual. The endpoint caps a response at 1000 records, defaults to 25, and hands you a cursor in the paging block. Most production integrations loop until the next URL comes back empty, which quietly turns one question into dozens of calls.

What is an ad intelligence API?

An ad intelligence API is a commercial product that aggregates ad data across platforms and returns it in one schema. Where Meta answers for Meta, an ad intelligence API answers for Meta, Google, TikTok, LinkedIn, and others through a single key. The value it adds is not access to public data. It is structure, history, and throughput.

Vendors snapshot the public libraries continuously, so they hold the archive Meta does not keep. They normalise creative into primitives such as angle, hook, format, offer, and call to action. And they raise the ceiling on volume, because quota is pooled across customers instead of capped per app.

The spread is wide enough that job fit matters more than brand, which is why we ranked the field in our guide to the best ad intelligence tools for agencies rather than by logo size.

Coverage: where each API can and cannot see

The Meta Ad Library API covers Facebook, Instagram, Messenger, Audience Network, and Threads. That is deep but narrow. It has no read on Google Search ads, no read on TikTok's creative rotation, and no read on LinkedIn's B2B offer testing.

Google makes the gap concrete. Its Ads Transparency Center lets anyone browse a verified advertiser's ads, but as of September 1, 2026 ScrapeCreators could not find a general public Ads Transparency Center API in Google's developer documentation. If you need that data in code, you are buying it or scraping it.

Retention differs by platform too. LinkedIn's own help pages state that ads stay in its library for one year after their last impression, and that EU-targeted ads disclose extra targeting and impression fields. Coverage rules are neither uniform nor permanent.

Region changes the answer as well. Adrio's August 6, 2026 comparison notes that the European Union gets broader visibility into all ads than most other markets, so the same call can return a rich record in one country and a thin one in another.

Access, authentication and rate limits

Both routes ask for credentials, but only one asks for a human review. The gated approval on Meta's side is the first real cost of the official route, and it lands before a single request is sent.

Throughput is the sharper constraint. HyperFX measured Meta's standard access at roughly 200 calls per hour per app on a rolling window. A 50-brand scan returning 100 ads each is 5,000 records, which is at least 50 paginated calls, so a naive poller can burn hours of quota in one pass.

When teams hit that ceiling they usually reach for the scraping route. Our guide to automating Meta Ads Library scraping covers where it holds up and where it breaks, because the library's HTML and GraphQL responses change every few weeks.

Managed ad intelligence APIs sidestep the per-app cap by pooling quota across customers. The trade is that you now depend on a vendor's uptime, schema decisions, and pricing instead of Meta's.

What each one costs

The Meta Ad Library API publishes no per-request fee. You pay in engineering time: app registration, identity verification, token rotation, pagination logic, and schema versioning. For a one-off research project that is close to free. For a production pipeline it is a standing cost.

Managed providers price in credits or seats. ScrapeCreators starts with 100 free credits and sells 25,000 credits for $47, with most endpoints costing one credit and richer detail costing more. SearchAPI charges $40 per month for 10,000 Reddit searches. Enterprise ad intelligence contracts sit in another band again, which is why the decision belongs in front of the price.

Spend data is the clearest example of paying for something Meta will not sell you. Our competitor ad spend tracking guide walks through estimating budget from public signals when the platform only discloses ranges for political ads.

Which API should you wire into your agent stack?

If the question your agent asks is about Meta creative, the official API is the honest answer and the cheap one. If the question is what a competitor is doing across channels and what changed since last week, the Meta API cannot answer it, because it holds no history and sees one platform.

A split is common and sensible. Use the official endpoint for anything that must be Meta-sanctioned or political, and an ad intelligence API for the recurring competitive sweep that feeds your dashboards and your agent.

If you are wiring the official route into Claude, Cursor, or another MCP client, our guide to the Meta Ads Library API for AI agents walks through token scope, pagination, and the fields that survive a tool call.

The transport layer matters as much as the data source, and how MCP servers connect AI agents to ad platforms explains why a read-only server is usually the right default for research work.

How to migrate from the Meta Ad Library API

Migration is rarely a rewrite, because the two APIs answer different questions. Start by mapping the fields you actually consume today. Most teams find they read Page name, ad copy, media, country, platforms, first-seen date, and active status, and little else.

Then reverse that mapping onto the wider schema, and keep one internal ID that survives both sources. Vendor schemas change as well, so store the stable ad ID rather than the order a vendor happens to return rows in.

Run both sources in parallel for two weeks and diff the output. The diff tells you whether the wider feed is genuinely adding coverage or quietly dropping the Meta records you already trusted.

What changes over the next year

Official coverage keeps expanding. Adrio's August 2026 piece notes that the EU already sees broader access to all ads than other regions, and transparency rules tend to tighten rather than loosen, which generally means more fields in more markets.

The more interesting shift is where the query runs. Ad libraries were built as browsing surfaces for humans. Every major vendor now exposes the same data through an MCP server or an agent tool, so the unit of consumption is moving from a page view to a tool call.

That reframes the build decision. The question is no longer which dashboard renders the prettiest table, but which API returns a structured, attributable answer that an agent can cite without you re-parsing HTML.

Decision checklist. Choose the Meta Ad Library API when your data must be Meta-sanctioned, your competitors advertise only on Facebook and Instagram, and you can absorb a one to two week approval window (HyperFX, April 25, 2026). Choose an ad intelligence API when you need cross-platform coverage, ad history past the moment an ad pauses, or throughput above roughly 200 calls per hour per app (ScrapeCreators, September 1, 2026).

Neither route is a conversion report. Both tell you what a competitor chose to publish, not what it earned. Treat every record as a signal to test, and keep a human between the pattern and the budget line.

Frequently asked questions