Search for "Facebook ad copy examples" and you get lists of a few well-written ads. They are useful, but they share a problem: they come from another category, another audience and often another year. They do not tell you what works in your category.
The Meta Ad Library is much closer to that answer. Every piece of copy your competitors keep live today is in there. The hard part is reading it one ad at a time. After thirty ads you cannot remember how often a promise appeared or which button dominated.
This article describes a way to read ad copy in bulk: first measure the language of the category, then separate the lines within it that live the longest.
Why a single ad tells you little
The copy of a single ad is one decision a brand made on one day. You cannot tell whether it was a good decision; spend and results are not published for commercial ads. But if the same word appears across ten brands, or a brand has kept the same promise live for months, that is no longer a decision. It is a pattern.
So the goal of copy research is not "find the best ad". It is to find the category's recurring language. Not to copy it, but to decide where in that language you want to stand.
Step 1: search in the category's language
In the Meta search in the side panel, search the product the way stores sell it, not the way you would describe it. That is what the Ready Terms chips are for: Free Shipping, BOGO, Shop Now, Money Back and similar. Searching with an offer phrase brings back everyone who uses that offer.
If you want ads that contain a promise word for word, set Search Type to Exact Phrase; that setting is in the Starter plan. In a keyword search the words can appear in any order; in an exact phrase they appear together, in the same order.
Keep the country fixed and set the status to active. The language of stopped ads belongs to the past; you are looking for today's.
Step 2: load the whole list
The library loads as you scroll. Copy analysis gets more meaningful with more cards; the thirty cards on the first screen do not represent a category. Auto Scan scrolls the page for you and never counts the same ad twice. You set the card limit and speed in the panel.
Step 3: read the category's language in the Analytics tab
When the scan finishes, switch to the Analytics tab. Three boxes summarise the ad copy in bulk. One important note: the Analytics tab summarises every card loaded in the scan; card filters do not change these numbers. What you read here is the picture of the whole search.
Top Keywords
The words used most often in the ads' own text fields (primary text, headline, description). Very short words, numbers and filler words such as conjunctions are not counted, so the list keeps the words that carry meaning.
Click a word and the cards are filtered to the ones that contain it. You immediately see which sentences, brands and visuals that word appears with.
Top CTAs
The five most used buttons and each one's share of the scan. If Shop Now dominates a category, the ads sell directly; if Learn More dominates, they explain first. Click a button and the cards are filtered by it.
Offer language
The share of ads whose copy mentions a sales promise: free shipping, a percentage discount, cash on delivery, limited stock, a money-back promise and buy-one-get-one. The patterns are deliberately narrow. A false match would show a promise that is not there, so the extension prefers a rate that may be slightly low to one that is inflated.
Step 4: separate the lines that last
The language of a category is one thing; which lines hold on within it is another. For that, go back to the cards.
- Set Sort to Longest Running. The copy on the top cards is copy the advertisers are not ready to switch off.
- Look at the stage badge on the cards. Winner means the ad has been running 45 days or more and is still active. An ad of the same age that has been stopped is marked Ended.
- Do the opposite as well: Newest Published shows which lines competitors are testing right now.
Reading the two lists side by side is the most productive part of copy research. A promise that appears often in new tests but never among the long-runners may be one the category tried and dropped.
Step 5: ask your own question with text search
If you are curious about a word the Analytics tab does not show, use the Search card text box. It filters cards by text in the ad copy, headline, description, brand name or domain. That is how you count how many ads make category-specific claims such as "warranty", "handmade" or "made locally": just look at how many cards stay visible after the filter.
If you want to limit the list to ads with a specific button, the Button Filter is in the Starter plan.
Step 6: write down what you find
Copy research that is not written down is forgotten within a week. Save the ads you like with a double-click on the creative and use the Saved state filter to look only at them. Note the copy together with its pattern: "question in the first line", "social proof in the second sentence", "offer at the very end". We describe the same method for the creative side in our article on building a competitor swipe file.
What a copy research session should produce
You should end up with three things:
- The category's default language. The five to ten words everyone uses and the dominant button. Using that language makes you ordinary; ignoring it makes you hard to understand.
- The promises that last. Two or three promises that repeat among ads with the Winner badge. These are claims that have passed the test of time in your category.
- The gaps. Something nobody says that your product can truthfully say. This is usually the most valuable finding.
Three mistakes to avoid
- Treating word frequency as success. A word that appears often is common, not necessarily effective. The closest signal to success is how long an ad stays live, and even that is not certain.
- Mistaking one brand's language for the category's. A large advertiser can fill the list. Turn on Group Same Advertisers, or check one brand's share in the Top Advertisers list.
- Copying the copy word for word. Take the pattern, not the sentence. We explain at more length why copying is not analysis in our article on competitor ad analysis.
Cost
The Ad Library is free and public. The extension's free plan scans 2,000 ads a day, the limit renews in 6-hour windows, and no credit card is required. You can see the plans for paid features such as exact phrase search and the button filter on the pricing page.
To do this with the extension: Competitor creative research · Facebook ads market research