The Meta Ad Library is the most underused public-interest dataset in U.S. political analysis. It records every ad on Facebook and Instagram that Meta has classified as "about social issues, elections, or politics" since the library opened, alongside the funding entity disclaimer, the spend, the ad count, and the geographic distribution of impressions. It is open. It is downloadable. It is updated continuously.
What follows is an analysis of the lifelong U.S. report that Meta makes available as a downloadable CSV through the Ad Library Report tool. The snapshot used here is dated May 3, 2026, covering all ads classified as social issues, elections, or politics from the library's inception through that date in the United States. The methodology can be re-run with any other date range; instructions are below.
A note on the date range, because it matters. The user-friendly question "what does Meta political ad spending look like over the last 30 or 90 days" is answerable in the same tool, but with a much smaller dataset that does not capture the multi-cycle pattern. The lifelong dataset is the right window for understanding how the system actually works. The 30 and 90 day windows tell you which races are heating up right now. Both are useful. This post focuses on the structural narrative the lifelong file makes obvious.
The headline numbers
From the Meta Ad Library Report, U.S. lifelong, May 3, 2026 snapshot:
- Total spend on ads about social issues, elections, or politics: roughly $5.39 billion across the entire history of the library
- Total advertisers with measurable spend: 186,470
- Total ads in the library: 17,570,858
That's almost $5.4 billion of paid Meta inventory, identified by the platform as politically or social-issue relevant, since 2018. To put that in scale: it is roughly half of what AdImpact projects for all political advertising across all media in the 2025-2026 cycle alone (AdImpact's 2025-2026 political projection). The Meta library is one slice of the overall picture, but it is the largest digital slice with public spend disclosure.
The shape of the spend is what matters next.
Concentration: a few hundred wallets do most of the work
The lifelong file shows a strong long-tail distribution. From the snapshot:
| Threshold | Advertisers | Cumulative spend | Share of total |
|---|---|---|---|
| $10M+ lifetime | 55 | $1.22B | 22.5% |
| $1M+ lifetime | 774 | $3.04B | 56.4% |
| $100K+ lifetime | 5,904 | $4.47B | 82.9% |
| $10K+ lifetime | 27,956 | $5.15B | 95.4% |
| $1K+ lifetime | 90,092 | $5.36B | 99.3% |
| All measurable | 186,470 | $5.39B | 100.0% |
Read it carefully. Less than 6,000 advertisers - 3.2 percent of the total list - account for 83 percent of all Meta SIEP spend in the U.S. since the library opened. The remaining 180,000 wallets, almost all of them small candidates, micro-PACs, local issue groups, and trial campaigns, account for less than 18 percent. The system is loud, but the loudness is not evenly distributed.
That is the first ideologically neutral fact the data exposes. Whatever you think of the politics, the structural truth is that political messaging at scale on Meta is dominated by a few hundred well-funded wallets, not by a grass-roots ecosystem.
Top-line spenders
The actual top of the lifelong list (using the May 3, 2026 snapshot, lifetime spend, U.S. only) is dominated by presidential campaigns and a few advocacy operations. The top entries by lifetime spend, abbreviated for length:
- Joe Biden (Biden Victory Fund): ~$79.7M lifetime, ~72,500 ads
- Mike Bloomberg (Mike Bloomberg 2020 Inc): ~$63.3M, ~181,000 ads
- Donald J. Trump (Trump Make America Great Again Committee): ~$58.3M, ~312,000 ads
- Joe Biden (Biden for President): ~$53.0M, ~54,400 ads
- Kamala Harris (Harris for President): ~$45.4M, ~49,100 ads
- WorkMoney: ~$40.0M, ~15,700 ads
- PragerU: ~$36.0M, ~42,600 ads
- Donald J. Trump (Donald J. Trump for President, Inc.): ~$33.2M, ~301,700 ads
- Kamala Harris (Harris Victory Fund): ~$32.2M, ~15,900 ads
- Facebook (Meta running its own SIEP-classified ads): ~$30.4M, ~810 ads
The top ten in raw spend is exactly the kind of list the headlines would tell you to expect. The interesting structure starts immediately after.
The disguised-brand pattern
A surprising amount of the political spend on Meta does not run under names that read as political. The library's "disclaimer" field, which is the legally-required "paid for by" line, frequently exposes a parent organization that the public-facing Page name does not.
A few examples lifted directly from the lifelong dataset, with no commentary added:
- "Energy Citizens" is a Page that has spent more than $22.8M in lifetime SIEP advertising. The disclaimer reads: American Petroleum Institute.
- "America's Plastic Makers" has spent more than $16.7M. Disclaimer: American Chemistry Council, Inc.
- "Stop Republicans" has spent more than $26.6M. Disclaimer: Stop Republicans.
- "AFP Action" has spent close to $20M. Disclaimer: Americans for Prosperity Action, Inc., which is the political arm of the broader Americans for Prosperity network.
- "Working America" has spent more than $15M. Disclaimer: Working America, which is the AFL-CIO-affiliated community advocacy organization.
- "Sandy Hook Promise" has spent more than $18M, with a disclaimer matching the organization name.
- "PhRMA" (the pharmaceutical industry's main trade association) has spent more than $20M.
- "Hillsdale College" has spent more than $16M, well above what most colleges spend on Meta political-issue ads.
The disclaimer field is the enforcement mechanism. Meta requires it for all SIEP ads, and the field is what lets a researcher trace a Page like "Energy Citizens" back to its actual funder. Without that field, the Page name alone would obscure who is actually paying.
That mapping has consequences. A reader scrolling Facebook sees "Energy Citizens" as a citizen brand. The library shows it as the American Petroleum Institute. Both descriptions are accurate; both are also legally permitted. The library is the only widely-available place to see them at the same time.
Where the money lands
The locations file in the same dataset breaks spend by U.S. state. Top states by lifelong SIEP spend (as of May 3, 2026):
| State | Lifelong spend | Share of national |
|---|---|---|
| California | $584.9M | 10.9% |
| Texas | $382.3M | 7.1% |
| Florida | $362.2M | 6.7% |
| Pennsylvania | $309.2M | 5.8% |
| New York | $299.6M | 5.6% |
| Michigan | $254.8M | 4.7% |
| Georgia | $215.0M | 4.0% |
| North Carolina | $214.6M | 4.0% |
| Arizona | $197.6M | 3.7% |
| Ohio | $180.4M | 3.4% |
| Illinois | $175.1M | 3.3% |
| Wisconsin | $155.3M | 2.9% |
Two patterns are obvious. First, the largest states by population (California, Texas, Florida, New York) consume a steady share of the spend regardless of whether they are competitive. Some of that is straightforward: campaigns spending nationally cannot avoid imposing a population-weighted cost on populous states.
Second, and more interesting, the seven traditional swing states (Pennsylvania, Michigan, Wisconsin, Arizona, Georgia, North Carolina, Nevada) collectively absorbed roughly 26.6 percent of all U.S. SIEP spend on Meta in the entire history of the library. That is a population-weighted overrepresentation. The combined population of those states is about 16 percent of the U.S., yet they absorbed about 27 percent of the political ad spend. The asymmetry is real and it is structural to American electoral geography. A nationally-strategic campaign cannot help but route its money there.
What other ideas the same dataset exposes
The CSV is much richer than the headline numbers. A short list of what else can be extracted with no additional data:
Spend efficiency by funding entity. Mike Bloomberg's 2020 campaign ran 181,000 ads on $63M; the Trump MAGA Committee ran 312,000 ads on $58M; Biden Victory Fund ran 72,500 ads on $79M. Different campaigns deploy radically different ad-volume strategies for similar budgets. The CSV makes this directly visible by simple division.
Multi-page funding networks. Several of the largest funding entities (Trump MAGA Committee, Biden Victory Fund, Harris Victory Fund) appear under several distinct Page names. Aggregating by disclaimer rather than by Page name surfaces the actual money flow. A surface-level scan that only groups by Page name systematically undercounts coordinated network spend.
Per-state advertiser breakdowns. The dataset includes a sub-folder of state-level advertiser CSVs. Each state file shows every advertiser whose ads reached that state. Comparing per-state spend to the national list reveals which advertisers are nationally-funded but locally-focused (a typical pattern for state-level issue advocacy) versus which are locally-funded and national in distribution.
Deceptive issue advocacy at scale. Cross-referencing Page name against disclaimer field exposes the disguised-brand pattern at scale. A simple analytical pass would flag any Page whose name contains civic-sounding language ("Citizens," "Americans," "People," "Stop," "Save," "Protect") and whose disclaimer maps to a corporate or trade-association funder. The library is the canonical primary source for that kind of investigation.
Long-tail micro-spend. Roughly 96,000 advertisers spent under $1,000 over the entire life of the library. Many of them are local races, hobbyist organizers, or single-issue micro-campaigns. The bottom of the distribution is a useful index for where political grassroots is spending its money in small dollars, separate from the institutional layer at the top.
Year-over-year comparison. The lifelong file is a single snapshot, but Meta's report tool lets you download the same data at any custom date range. Running the equivalent CSV for the last 30 days, 90 days, calendar 2025, or any other window and diffing against the lifelong file produces a year-over-year change report, segmented by advertiser, disclaimer, and state.
Regional and city-level focus. While the public report breaks out by state, the underlying ads carry impressions data at finer granularity in the live library UI. Researchers willing to script against the public-facing Library can produce city- or media-market-level rollups for any window.
How to run the same analysis on a 30 or 90 day window
The lifelong file used here is the same kind of file that Meta makes available for any custom date range. Steps:
- Go to facebook.com/ads/library/report.
- Set country (United States, or any country supported by the library's SIEP transparency).
- Choose a date range. Meta supports lifetime, last 7 days, last 30 days, last 90 days, and arbitrary custom windows up through the most recently completed 7-day period.
- Download the report. It arrives as a ZIP archive containing an
advertisers.csv, alocations.csv, and aregions/folder of state-level CSVs. - Process the CSVs with whatever tool you prefer. A few lines of Node, Python, or SQL on the advertisers file produces top-spender lists and concentration ratios. The locations file produces the state breakdown.
That same workflow is the right way to compare current cycle activity against the lifelong baseline.
Other platforms with comparable transparency requirements
Meta is the most-developed public ad library, but it is not the only one. If you are reviewing political ad spending across the major platforms, these are the comparable disclosure surfaces.
Google Ads Transparency Center. Google publishes political and issue ads in its Ads Transparency Center, with a dedicated political tab. Spending data is shown in ranges (for example, "$200 to $300") rather than precise figures, with impression ranges in similar bands. Verification requires a U.S. government-issued ID, and ads must include a "paid for by" disclosure. (Google blog announcement on the transparency report for political ads, Google Cloud guide on the dataset)
X (Twitter). X reinstated political advertising in 2023 after a multi-year ban under prior management. The platform requires identity verification, "paid for by" disclosures, and inclusion in its Ads Transparency Center for at least seven years. Coverage and update cadence have been criticized by external researchers, including reporting that X's political ads library went weeks without updates during the 2024 cycle. (Bloomberg coverage of X's ads library lapses)
LinkedIn. LinkedIn launched its Ad Library in compliance with the EU Digital Services Act and tightened its transparency framework effective April 1, 2026. The framework introduces persistent visible "Sponsored" labels and a "Why this ad" disclosure showing targeting criteria. LinkedIn also removed inferred-political-affiliation, health-group, and several sensitive demographic targeting parameters as of April 2026. (LinkedIn engineering blog on the Ad Library)
TikTok. TikTok does not allow paid political advertising on its platform at all, a policy in place since 2019 (TikTok Politics, Government, and Elections policy). Under the EU DSA, TikTok has committed to a commercial Commercial Content Library covering all advertising, with EU enforcement actions in 2025 driving subsequent updates (European Commission on TikTok DSA commitments).
EU Digital Services Act, Very Large Online Platforms. Article 39 of the DSA requires designated VLOPs to maintain a public ad repository with persistent ad records, advertiser identification, targeting parameters, and reach data. As of 2026, this applies to Meta, Google, X, LinkedIn, TikTok, Pinterest, Snapchat, and several other services. EU-resident researchers can use these repositories for any platform that has been designated. (TechPolicy.Press on DSA impact on advertising)
The cumulative effect of this collection of disclosures is that anyone with a few hours and basic data-processing skills can audit how the major platforms are paid to deliver political and issue messaging in any country with applicable transparency rules. The Meta dataset is the most usable today. The others are getting better, and the EU DSA is forcing the slowest of them to catch up.
What this means for a working researcher
The honest version of the takeaway: the public ad-transparency datasets are useful precisely because most journalists, candidates, and citizens still treat them as too tedious to query. The Meta library has been open since 2018. The data is in CSV form. Pulling it and running a basic analysis is a single afternoon of work. The findings above came from one such afternoon and a snapshot file that anyone can re-download.
The longer-term implication is that the discourse around "dark money in political ads" is now much more accurately described as "money that is not dark, but is buried in CSVs no one is reading." The disclosure works. The bottleneck is on the analysis side, not on the platforms.
If you are a journalist, a campaign opponent, a watchdog organization, an academic, or a candidate running for any office, building the habit of pulling the latest Meta Ad Library report - and the equivalent reports from Google, X, and LinkedIn - is one of the cheapest investigative reflexes available. It does not require a subpoena. It requires a CSV reader.
Related reading
- The Best MCP Servers By Industry for the broader infrastructure shift toward agent-driven querying of datasets like this
- Why Top Quant Desks Outperform And How U.S. Tax Code Multiplies The Edge for a parallel pattern of public-data analysis applied to financial markets
- What Actually Fixed My Claude Code Sessions for the structured-research methodology behind this kind of investigation
- A 24.8 Percent AI Crypto Portfolio, Back-Checked for the same back-check discipline applied to a different domain
Fact-check notes and sources
- Source data: Meta Ad Library Report, U.S. lifelong dataset, snapshot dated May 3, 2026. Field definitions per the report's published README.
- Public details on the Meta Ad Library tool surface: Meta Transparency Center research tools, Meta Ad Library policies
- Google political ad transparency: Google Ads Transparency Center, Google blog on the transparency report, Cloud guide on the dataset
- X transparency posture: X Transparency Center, Bloomberg on X's lapses in 2024 cycle updates
- LinkedIn Ad Library: LinkedIn engineering blog, LinkedIn DSA disclosures
- TikTok political-ads ban and DSA commitments: TikTok policy on Politics, Government, and Elections, European Commission on TikTok DSA commitments
- Aggregate political spending projections: AdImpact 2025-2026 political projections, Brennan Center on online political spending in 2024, OpenSecrets online political ads coverage
- DSA framework on advertising transparency: TechPolicy.Press on DSA impact, Campaigns & Elections on ad-library transparency reports
This post is informational. The numbers above are derived from a single downloaded snapshot of a public dataset; the live library updates continuously, so figures will drift. Disclaimer-field interpretation reflects what is shown in the dataset and does not imply legal conclusions about the entities named. Treat all assertions about specific advertisers as grounded in the disclosure they themselves submitted to Meta, not in editorial judgment.
