Transparency regulation handed marketers a gift most have not opened: every major ad platform now operates a public library of the ads running on it. Your category's entire paid conversation - every claim, offer, format and refresh, from every competitor - is readable without an account, without their permission, in something close to real time. This guide is the systematic version of reading it: which libraries expose what, the fields worth extracting, the longevity heuristic that approximates performance data, and the weekly diff that turns browsing into intelligence.
The open secret
The libraries exist because regulators demanded ad transparency, and platforms built them properly: searchable by advertiser, filterable, dated. What makes them intelligence rather than trivia is a structural fact about advertising - paid media is where companies state their strategy in public, in money. The claims a rival pays to repeat are the positioning they believe converts; the formats they invest in reveal their production capacity; the refresh rhythm reveals their creative operation's health. Reading the feed is reading the strategy - and unlike most competitive surfaces, this one updates continuously and lies rarely, because every entry cost its author real budget.
The five libraries and what each exposes
| Library | Covers | Distinctive value |
|---|---|---|
| Meta Ad Library | Facebook, Instagram, Messenger | Deepest archive; variations visible; start dates |
| Google Ads Transparency Center | Search, YouTube, Display | Search-ad copy and video creative in one place |
| LinkedIn Ad Library | The B2B motion: who targets which professional segments | |
| TikTok Creative Center | TikTok | The short-video frontier; trend formats early |
| Shopping surfaces | Product listings and promos | Ecommerce offer and pricing behavior |
Coverage is the point of listing them: a rival quiet on Meta may be spending heavily on LinkedIn, and a single-library read produces confident wrong conclusions about their investment. The full sweep across libraries is precisely the kind of parallel, repetitive read that runs as one mission rather than an afternoon of tabs.
The fields that matter
The difference between browsing and surveillance is extraction. Per competitor, per library, pull: the active creatives and their claims (clustered into the angle vocabulary your angle map uses), formats (static, video, carousel - the production-investment tell), start dates (the longevity input), variation counts (many variants of one concept means a test in flight; a dozen unrelated ads means spray), and offers (discounts and trials - pricing strategy visible in the feed before it reaches the pricing page). Structured this way, reads accumulate into time series, and time series are where every interesting conclusion lives.
Longevity: the performance proxy
The weekly diff
A single read is a snapshot; the value concentrates in the deltas. The weekly diff answers four questions per rival: what is new (fresh creatives - and a cluster of new ads on one theme is a launch or a pivot announcing itself), what was retired (killed ads are failed hypotheses - free negative knowledge about your shared market), what got refreshed (new variants of a proven concept - doubling down), and what moved (spend expanding to a new platform or format). This is the highest-signal page in the whole competitive intelligence program, and it stays readable in two minutes precisely because it reports change rather than state.
Routing what you find
Surveillance earns its cadence only when findings reach their consumers - the pillar's routing discipline applied here. Angle occupancy feeds the creative loop's map and the creative testing queue. Claims feed the battlecards - what the rival tells your shared prospects is exactly what sales must answer. Offer patterns feed pricing conversations. And the momentum read - expanding, contracting, pivoting - feeds the competitor analysis. As a scheduled mission from the Competitive Intel shelf, the whole loop - sweep, extract, diff, route - costs minutes a week and reads like a wire service for your category's paid strategy.
Frequently asked questions
Are ad libraries legal to monitor?
Yes - they exist for transparency and public reading is their intended use. Every major platform operates one: Meta, Google, LinkedIn and TikTok all publish the ads running on their surfaces.
What should be extracted from an ad library read?
Structured fields per rival: claims clustered into angles, formats, start dates, variation counts and offers. Structured reads accumulate into time series; screenshot galleries do not compound.
How can performance be inferred without metrics?
Longevity: advertisers kill what does not work, so long-running ads passed their owner’s own scrutiny. Weighting reads by run time separates proven claims from experiments.
How often should libraries be swept?
Weekly, as a diff: new, retired, refreshed and moved. Creative operations work on weekly rhythms, and the changes carry more strategy signal than any static read.
What do you do with the findings?
Route them: angles to the creative testing queue, claims to sales battlecards, offers to pricing, momentum to the teardown. Surveillance that ends in a folder is a hobby.
Sources
Every playbook on this blog ships as a runnable mission.
Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.