Facebook Ad Library Search Tricks Most Marketers Miss

The search mistake that hides most winning ads

Most people open the library, type a competitor's name, click the Page in the dropdown and scroll. That shows you one Page. It doesn't show you the brand.

On the Ad Radar board (2,443 active Meta ads with $50k+ in estimated spend, October 2026), we can see where each ad sends its traffic. 1,463 ads, 62% of the board, point to a website that also receives ads from at least one other Page. Brands run winners through creators, partner Pages, review-style Pages and advertorial "news" Pages. Searching the brand Page alone misses that half of the picture.

For Javvy, the brand Page runs fewer than half of the winning ads that go to its store. Ten Pages do the rest. If you only searched "Javvy" and clicked the Page, you would study 7 ads and miss 10.

Trick 1: run both search modes on purpose

The search box does two different things depending on what you click.

Run the advertiser search first to audit the main Page. Then run a keyword search for the brand name, the product name and the offer to catch the Pages around it.

Trick 2: use quotes for exact phrases

Typed without quotes, a multi-word query matches the words in any order and anywhere in the ad. Wrapped in double quotes, it matches the phrase exactly. The API makes the same split explicit with two search types, KEYWORD_UNORDERED and KEYWORD_EXACT_PHRASE, documented in Meta's Ad Library API reference.

Exact phrase is how you find every brand running the same offer or the same claim structure. Here is how often some common phrases appear in the written copy of winning ads on our board:

Offers live in the written copy. Authority and story live in the voiceover. Search "doctor" in the library and you find 60 of the 154 winners that use the word. The other 94 only say it out loud.

Trick 3: remember the library can't hear

Keyword search reads written fields and advertiser names. It doesn't read what people say in a video. That is the single biggest blind spot for hook research, because the hook of a video ad is almost always spoken.

Of the 1,682 video winners on our board with a transcript, 1,679 open with a spoken line that doesn't appear anywhere in their headline or primary text. The written copy carries the offer. The voice carries the hook.

If you want to search hooks, you need transcripts. Either transcribe the ads you save, or use a tool that searches scripts. Ad Radar searches headlines, body copy and video transcripts in one box.

Trick 4: search the offer and the domain

Two keyword searches that find Pages the brand search misses:

• The offer, in quotes. If the brand's product page shows "Buy 2 Get 1 Free", search that exact phrase. Partner Pages copy the offer word for word.
• The domain. 163 ads on our board write a URL into their primary text. Searching "brandname.com" as a keyword surfaces Pages that paste the link in the copy, which is common on advertorial-style Pages.

Trick 5: pick the country on purpose

The country filter is not cosmetic. Two rules:

1. Search where they sell. "All" mixes markets and languages. For a US brand, pick the United States.
2. Switch to an EU country for history. Ads that delivered in the EU stay in the library for one year after their last impression, while other commercial ads disappear when they stop, according to Meta's Ad Library tools page. If a US brand also sells in Germany or France, searching those countries shows inactive ads you can't see from the US. EU ads also show reach and the age, gender and location targeting.

Trick 6: filter media type, then read versions

Filter to images or videos separately. The two behave differently. On our board, 43% of image winners carry two or more variants, against 25% of video winners. Image ads are cheap to multiply, so brands stack headline and text versions on the same Library ID.

When a card says the ad has multiple versions, open it and click through every version you can see. Write down what changes between them. If only the headline changes, the brand is testing angles. If only the image changes, it is testing visuals behind a copy it trusts. We go further in variant count as a scaling signal.

Trick 7: use the date filter as a time machine

The date filter narrows results to ads that delivered impressions in a window you choose. Set it to last November and you see who was running during Black Friday, as long as those ads are still active or delivered in the EU. Combine it with active status "All" when you search an EU country.

Trick 8: read the platform icons and the label

Each ad shows the platforms it runs on, which tells you whether the brand is buying Facebook, Instagram or the full set of placements for that creative. Since late 2025 the library also adds a "Low impression count" label to ads with fewer than 100 impressions (PPC Land). Skip those: they are tests that never got budget.

A search routine you can copy

For each competitor, in this order:

1. Advertiser search on their main Page, United States, active only. Count ads, note the oldest start date.
2. Keyword search for the brand name and the product name, no quotes. List every other Page that appears.
3. Exact phrase search for their main offer in quotes.
4. Keyword search for their domain.
5. Switch to an EU country they sell in, status "All", to see what they stopped.
6. Filter to video, open the five oldest ads, transcribe the first sentence of each.

Log what you find in one place. Our swipe file guide has a tagging system, and the weekly competitor research process turns this routine into a 30-minute habit. If you are starting from zero, read how to use the Meta Ad Library first.

Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.