Ad Spy Tools: How to Find Any Competitor’s Ads Across Every Network

Almost every major ad platform now publishes what its advertisers are running. Meta keeps a public library of every active ad across its products.^{[1]What is the Meta Ad Library and how do I search it?facebook.com} Google runs an Ads Transparency Center where you can look up an advertiser by name or website and see what they have served.^{[2]Ads transparency – Advertising Policies Helpsupport.google.com} The information is free and sitting in the open. The problem is that checking ten platforms by hand, one at a time, is nobody’s idea of a good afternoon.

This session was a live walkthrough of what happens when you stop doing that manually and build the aggregator yourself, plus two other tools that came out of the same impulse.

Why Competitor Ads Are the Cheapest Research You Will Ever Do

A live ad that has been running for years is not a guess. It is a campaign somebody keeps funding, which means it earns more than it costs. That makes ad libraries the closest thing to free market research in this business.

• Longevity is the signal – Creative that has run since 2023 has survived every optimization pass a real budget can buy
• Structure over copy – What matters is the offer shape, the hook and the page speed, not the exact words
• Format tells you budget – Which layouts an advertiser commits to reveals where the platform spend actually goes
• Cross-platform presence – Once you identify one competitor, checking every network shows you which channels they proved out

The Search Language Problem

Most people search an ad library the way they would search Google, by typing the niche. That returns organic noise, because advertisers rarely name their product in the creative.

• Search the call to action, not the category – Advertisers use urgency language that ordinary posts do not
• Home services follow a pattern – Service type paired with an intent phrase surfaces the actual campaigns
• Insurance markets sideways – The hook usually describes a benefit or a scenario rather than the policy itself
• Time pressure is a tell – Phrases implying speed or a deadline almost always belong to paid creative

Reddit: Reading Threads That Were Never Organic

Search any local service in a city and you will find threads that look like genuine questions. Some of them are, and some are marketing. The tell is consistent enough to check in about a second.

• The pattern – A believable question, then a helpful-sounding reply that happens to name a specific business
• The test – Genuine questions rarely collect domain names in the comments, planted ones almost always do
• Why anyone bothers – Those threads get indexed, so the business borrows search visibility it never had to earn
• What it tells you – Wherever you find this happening, somebody has decided the niche is worth the effort

Scoring Thousands of YouTube Videos Instead of Browsing Them

Google Ads lets you target specific YouTube videos and channels as managed placements.^{[4]About targeting for Video campaignssupport.google.com} That is enormously useful and completely impractical by hand, because finding the right videos means reviewing thousands of them.

• Scrape wide first – A few dozen keywords produce a video pool far larger than anyone would review manually
• De-duplicate automatically – Overlapping keywords stop inflating the list once matches are stripped out
• Let AI do the filtering – Every video gets scored against a description of who you want to reach
• Then sort by what matters – View counts, publish dates and the channels producing consistently in the niche

Why Your AI Filter Returns the Wrong Videos

The most useful part of the session was watching a filter fail live and get fixed. The first scoring pass ranked sales-training and recruitment videos as highly relevant for a consumer insurance niche.

• The cause – The prompt described the subject matter, and industry content is topically identical to buyer content
• The fix – Describe the person you want to reach rather than the topic they relate to
• Exclusions do real work – Naming what you do not want is often more effective than describing what you do
• Raise the threshold too – Tightening the cutoff cuts the list dramatically and quality improves at every step

The Lifetime Value Math Nobody Runs First

An automated competitive report on auto insurance prompted the bluntest point of the call, and it applies well beyond insurance.

• The asymmetry – A carrier that keeps a customer for years values that customer in the thousands, while an affiliate is paid once per lead
• What that means at auction – The incumbent can outbid you permanently and still profit
• Where your edge has to come from – Creative angles, better information, or tooling the big players do not have
• The check to run first – Compare your payout against the incumbent’s lifetime value before you commit budget to a vertical

Two Shortcuts That End Accounts

Both came up naturally, one from an audience question and one from the conversation that opened the call. Neither is worth the risk.

• Misrepresentation is treated as egregious – Google states that accounts will be suspended upon detection without prior warning, with reinstatement only in compelling circumstances^{[5]Misrepresentation – Advertising Policies Helpsupport.google.com}
• Chasing accounts is not a business – When replacing banned accounts becomes the operation, the margin is gone
• Unsolicited texting is expensive – The TCPA allows recovery of up to $500 per violation, which a court may treble to $1,500 per violation for willful or knowing conduct^{[6]47 U.S. Code § 227 – Restrictions on use of telephone equipmentlaw.cornell.edu}
• It compounds per message – Those figures are per violation, which is what turns a campaign into a headline
• Detection keeps improving – Platforms are increasingly automated about finding this, so the window narrows every year

Build the Tool You Will Actually Use Tomorrow

The through line of the session was that all three tools started as personal annoyances rather than product ideas.

• The spy tool – Built because checking competitor creative across networks by hand was eating hours every week
• The video scraper – Built because placement targeting is only useful if you can find the placements
• The video editor – Built because editing recordings manually was the bottleneck in an entire content pipeline
• The pattern – Each one solved a problem that already existed, which is why each one gets used