Ad spy tools: what they actually return, and what they cannot

An ad spy tool is a searchable archive of the ads other advertisers are running now. It returns creative, copy, landing pages and run dates. It never returns revenue or ROAS, and no tool can. Judge one on field coverage, refresh rate and depth per network, not on database size.

What do ad spy tools actually return?

Most tools return the same six fields. The creative file, the headline, the body copy, the call to action, the destination URL, and the date range on which the ad was first and last seen. The stronger ones add the country, the placement (Facebook feed versus Instagram Reels, for example) and the advertiser's page identity.

For video, a useful tool also returns structure: the first three seconds, the on-screen text, the caption, the aspect ratio and the length. That is the material a creative team can work with. A cover image and a headline is not.

The seventh field is the one buyers assume they are paying for, and it does not exist. None of these tools can return impressions, clicks, conversion rate or ROAS for another advertiser. Meta says so plainly: the Ad Library shows all active ads, and adds spend, reach and funding detail only for ads about social issues, elections or politics (Meta Transparency Center, updated 24 August 2023).

So every winning ad label inside a spy tool is a proxy. In practice it means the ad has been running a long time, or it collected visible engagement. Run duration is the honest signal, because nobody keeps paying for a loser for sixty days. Engagement is the weaker one, since it reflects comments as much as conversions, and plenty of advertisers turn comments off.

That is why an ad that has been live for eleven weeks is worth more of your attention than an ad with 4,000 reactions from last Tuesday. One is a budget decision repeated 77 times. The other is one good day.

Which networks can you spy on, and how deeply?

Meta is the deepest public archive and the only one with an official API a marketer can query at scale. Meta's Ad Library covers all active ads across its apps, archives EU-delivered ads for one year after their last impression, and keeps political and social issue ads for seven years.

Google is narrower than most buyers expect. The Ads Transparency Center is a searchable hub of ads served from verified advertisers, launched on 29 March 2023 (Google, announcing the new Ads Transparency Center). It is genuinely good for confirming that a competitor runs search, YouTube or display creative. It is not a performance database, and its history is short.

TikTok has two surfaces. The Creative Center shows top-performing creative by industry and region, and a commercial content library exists because EU rules require one. Both are strong on formats and hooks. Neither publishes spend on commercial ads.

LinkedIn, Pinterest and Snapchat publish considerably less, and where an archive exists it tends to lean political or regulatory rather than commercial. Native, display and push networks have almost no public archive at all. That gap is exactly where paid spy tools earn their subscription.

Regional rules matter more than most buyers realise. The EU archives exist because of the Digital Services Act, which is also why an EU-targeted ad often has a clean, dated record while the same advertiser's US campaign has none.

How do ad spy tools get their data?

There are four collection routes, and every tool you compare is a different mix of them.

The first is the public transparency archive. It is free, structured and shallow, as described above. Any tool can read it, which is why so many tools look identical on Meta.

The second is an official API. Meta's Ad Library API is the big one, and its scope is narrower than the marketing copy implies: it is built for deeper analysis of social issue, election and political ads, plus ads delivered to the EU, plus active public branded content (Meta Transparency Center). Teams that need commercial creative at scale end up on a broader ad intelligence API instead, which our comparison of an ad intelligence API versus the Meta Ad Library API sets out in detail.

The third is scraping and panel data. Scrapers read the same public pages more often, with retries and normalisation. Panels add clickstream and app usage signals from real devices. Both give breadth. Both break when a platform changes a payload or starts blocking a range of addresses.

The fourth is affiliate and network feeds. Native and push networks supply their own creative feeds to the tools that pay for them, which is why a Native or Push module can cost as much as a complete social tool.

This is also why the same ad shows three different run dates in three different tools. Each one records first sighting, not launch. A tool that started polling yesterday reports a shorter run than a tool that has been watching the advertiser for a year. When two tools disagree by weeks, you are looking at crawl history, not a data error.

How much should an ad spy tool cost in 2026?

Published prices in 2026 run from about $9 to about $249 a month. BigSpy starts at $9, Trendtrack at $47, AdSpy at $149 with no trial, and AdPlexity sells traffic modules separately at $149 for push or YouTube and $249 for native (Trendtrack's 2026 roundup of 14 tools). A separate 2026 guide puts paid spy tools broadly at $30 to $200 a month (AdLibrary.com).

Price tracks three things: how many networks are covered, whether exports and API access are included, and how many seats you need. It does not track accuracy. A $249 tool with stale run dates is worth less than a $47 tool that refreshed this morning.

If you are an agency, do the seat and export maths before the demo. A plan that looks cheap for one analyst gets expensive at eight, and export caps are where a $99 tier quietly becomes a $299 tier. Ask for the row limit per month in writing.

Free trials are not evenly distributed either. Some tools offer a seven-day trial, some a three-day one, and the two most established Meta specialists sell a subscription with no trial at all. Budget a paid month as the real evaluation cost, and run the test below inside it.

How do you test an ad spy tool in ten minutes?

Ten-minute field test, adextract checklist, October 2026: pick one advertiser you already know is advertising, check whether the tool shows their ads from the last 48 hours, open two landing pages, export 20 rows, then repeat the whole test on the one network you actually spend money on.

Step one exposes freshness faster than any feature list. If a tool cannot show you an ad that started yesterday, its crawl cadence is measured in weeks, and every recently launched filter it offers is decoration.

Step two catches the most common data failure: a destination URL that resolves to a homepage because the ad pointed at a tracking redirect the tool never unwrapped. If two landing page clicks dead-end, budget an hour of manual repair per research session.

Step three is about relevance. Search a product term your competitors use and count how many of the first twenty results actually mention it. Spy tools that index by advertiser name rather than ad text will hand you twenty unrelated results and call it coverage.

Step four is the one buyers skip. Most tools are strong on Meta because Meta is the easiest public archive, and weak on LinkedIn, native or push, which is often exactly where an in-house team buys. Testing only on Meta tells you nothing about the network you care about.

Why database size is the wrong number to buy on

A billion creatives sounds decisive until you deduplicate. The same video often appears as five records with different flight dates, different countries or slightly different thumbnails. Group by creative file and the count falls hard. That is how a tool claiming 164 million ads across 223 countries and 88 languages (AdSpy, per Trendtrack) and a tool claiming one billion ads across nine platforms (BigSpy) can return the same forty competitors in your niche.

Freshness beats size for one reason: creative dies fast. If your category refreshes every fortnight, a database three weeks behind is showing you last quarter's strategy. Our piece on spotting ad creative fatigue before ROAS drops runs the same logic from the other direction.

Depth beats breadth for a second reason. Forty well-labelled fields on one network, with clean run dates and unwrapped destination URLs, beats four networks of cover images with no structure. You can query the first. You can only look at the second.

One practical check before you trust a count: ask what a record is. If a record is an ad flight rather than a creative, the headline number is inflated by every restart, and the number you actually care about, distinct creatives per advertiser per week, is not on the pricing page.

Where spy tools stop, and the agent gap begins

A spy tool is a browsing surface. A person types a keyword, filters, sorts, screenshots and pastes into a deck. That works at one competitor. It fails at forty, and it fails completely when the output has to be reviewed every week by someone who did not do the browsing.

An agent needs a queryable layer with stable fields, not a dashboard. That is what an MCP server provides: the agent asks for ads by advertiser, by network, by first-seen date, and gets structured rows back. Our guide to choosing an MCP server for competitor analysis covers the questions to ask before you wire one in.

Exports are the usual bottleneck. Most spy tools allow a few thousand rows a month on the entry tier, which is a week of monitoring for a mid-size agency and a day for anything automated. If the plan is an agent that polls daily, the export cap is the real pricing axis, not the database size.

This is the part most roundups skip, because most roundups are written by browsing tools. If your deliverable is a weekly competitive brief rather than a lunchtime scroll, the buying question is not which tool has the biggest database. It is whether the data can leave the tool at all.

Which ad spy tool fits your buying motion?

E-commerce and dropshipping: buy for creative volume and TikTok depth, and accept that most of what you see is other dropshippers testing the same products. The $9 to $49 tier is enough to start, and the free tiers are enough to learn the interface.

Agency: buy seats, exports and a client-ready surface, and check the per-seat maths before you commit. If AdSpy is on your shortlist, our breakdown of AdSpy alternatives maps four routes by buyer type.

In-house performance team: buy depth on the networks you actually spend on, and treat Meta coverage as table stakes rather than a reason to buy. The criteria in our competitor ad monitoring tool guide are the ones procurement will ask you about.

Agent-first teams: buy the data layer and budget seats only for the two or three people who do the weekly read. Everything else should arrive as a query, not as a login.

What to do next

Write down the three questions your research has to answer this quarter. Then test each candidate against those three questions only, on your own network, with an advertiser you already know. Ten minutes per tool, same checklist, same day.

Before you commit to a spy tool at all, check whether the free archive you already have is enough. If what you need is spend rather than creative, no spy tool sells it, and our guide to estimating competitor ad budget from public ad data covers what you can actually defend in a client meeting.

And if the deliverable is a weekly brief rather than a browsing session, price the data layer first and the interface second. Interfaces are easy to replace. A field schema your workflows already depend on is not.

Native inventory is the case where the mechanics differ most, because there is no public archive to read. Our guide to native ads spy tools covers the Taboola, Outbrain and MGID tools, what each one costs on vendor pages checked in September 2026, and which claims do not survive a trial.

Frequently asked questions