How to research competitor ads in the Meta Ad Library: a step-by-step guide

To research competitor ads in the Meta Ad Library, open the library, choose the market in the country selector, keep the category on All ads, search for each competitor's Page, then open every ad and record it in one sheet with the same tags. For APAC markets, capture on a schedule: a commercial ad that does not reach the EU or UK shows only while it is active, with no spend, reach or targeting data.

Key takeaways

• The Ad Library is free, open to anyone and contains all active ads shown across Meta's technologies.
• Outside the EU and UK, a commercial ad shows its creative, Page and start date, but no spend, impressions, reach or targeting, and it drops out once it stops running.
• Search for the competitor's Page, not a keyword: a keyword search also returns other Pages' ads that use the term.
• Record and tag every ad with one codebook, then compare shares and distinct messages rather than raw ad counts.
• Deduplicate before you count. In SOMIN's read of ad-optimisation vendors, Madgicx's 187 Meta ads carried just 55 distinct primary texts.

What does the Meta Ad Library show for APAC markets?

Less than most marketers assume. What you see depends on where an ad was delivered and what it is about, and Meta's help article About the Meta Ad Library sets out the differences. An ad that runs only in Singapore, Indonesia or Australia falls in the first column.

Two consequences matter for an APAC team. First, the library is a record of what competitors say, not of what works: there is no spend or result for commercial ads. Second, it has no memory for your markets. If you look once a quarter, you see only what happens to be running that week.

The EU rules come from the Digital Services Act. When it took effect, Meta announced that ads targeting people in the EU would be stored in the public library for a year, with the dates they ran and the targeting parameters used. A regional competitor that also advertises in Europe therefore leaves a longer trail there than at home.

The library also holds active branded content with a paid partnership label, created on or after 17 August 2023. You search it by Page or Instagram profile, after choosing Facebook or Instagram and a date range, which helps when competitors work through creators.

How do you search the Meta Ad Library, step by step?

Eight steps cover a competitor in one market. Repeat them per market rather than mixing countries in one search.

1. Open the library. Go to the Meta Ad Library. Meta says anyone can view and search it. You need to log in only to download searches, which also means accepting the terms of service.
2. Choose the country. Meta's instruction is simple: to view ads from a specific country, use the country selector. Run Singapore, Malaysia, Indonesia and the rest as separate passes.
3. Keep the category on All ads. The other options cover ads about social issues, elections or politics, and the special ad categories for housing, employment and financial products and services.
4. Search for the advertiser and pick their Page. Meta notes that any ads that include the search term, or ads run by Pages that include the term, may appear. Choosing the Page itself keeps other advertisers out of your results.
5. Narrow the results. Meta's help centre lists filters including status, language, platform, media type and impressions by date. Use what the interface offers for your search to separate, say, Instagram video from Facebook images.
6. Open "See ad details". It shows when the ad started and whether it is still running, and for lead ads it can show more about the lead form. Ads built with dynamic creative combine uploaded images and headlines into variations, so review every version you can see.
7. Check the Page's transparency details. Creation date, name changes, connected accounts and the primary countries where the Page is managed help confirm you have the right advertiser, and show whether a "local" brand is run from another market.
8. Save the evidence. Copy each ad's link and ID into your sheet with the date you captured it, and take a screenshot, because the ad may be gone next week.

Two more checks. Ads with age-restricted content, such as alcohol or gambling, do not show their creative to people who are underage or not logged in, so log in for those categories. And Meta covers only Meta: search ads sit in Google's Ads Transparency Center, which shows the region an ad ran in and the last date it ran. In our worked example below, 331 of the 633 ads were on Google.

What should you record for each competitor ad?

Record what you will need to compare, and the evidence to check it later. One row per ad, with these fields:

How should you tag competitor ads?

With one written codebook, applied the same way to every competitor and every market. Without it, two analysts will tag the same ad differently and your comparisons will measure the analysts.

• Customer need: what the ad promises to solve, such as security, efficiency, growth or knowledge.
• Content type: product offer, feature, education, thought leadership, testimonial or event.
• Offer type: free trial, free plan, discount, report or guide, demo, event.
• Claim type: a specific number ("3.3x faster ad creation"), a comparison ("Try Smartly Instead"), a risk reversal (upgrade only after you see a return) or none.
• Tone and format: confident, humorous or informative; single image, video, carousel or dynamic creative.
• Funnel stage: awareness, consideration or conversion, judged from the offer and button.

Allow several tags per ad and write a one-line definition with an example for each. Then group near-duplicates: the same text with a different image is one message with several executions. Tag local-language ads in the original language with a native speaker's check, since translation flattens tone; our piece on content localisation in Asia explains why.

AI makes this practical at scale. SOMIN's models tag every captured ad on these dimensions, and the Strategy agent turns the tagged corpus into themes, with an analyst checking the clusters.

How do you compare competitors fairly?

Compare proportions within each competitor, count distinct messages as well as ads, and keep markets apart. Raw counts mislead.

• Shares, not totals. A brand with 300 ads and one with 30 can both be "mostly trial offers". Compare each brand's share of its own output.
• Messages, not executions. 187 ads with 55 texts is a different strategy from 187 ads with 187 texts.
• Timing. Madgicx started 110 of its 187 Meta ads in November 2025. A one-month sample would have read a seasonal push as the norm.
• Run length with care. An ad that runs for months is one the advertiser is content to keep, not proof that it works.
• Paid next to organic. In the same twelve months Smartly published 698 organic posts on the channels we track and Madgicx 11, so their ads play very different roles.

For turning a comparison into whitespace verdicts per country, see competitor ad analysis by market; for the measurement side, see share of voice by market.

What does a worked example look like?

Here is the method applied to SOMIN's own category. Our ads-optimisation workspace captured every Meta ad that three ad-optimisation vendors ran at any point between 18 October 2025 and 17 October 2026, then tagged them as above.

The table reads as two strategies. Madgicx runs many short-lived ads across every Meta surface and closes on a free trial; Smartly runs fewer messages for longer and offers reports and guides to download. Neither pattern shows up in a single snapshot: Madgicx's peak came in November, and 34 of its ads in the window had started before it opened. Google's data added the regional view Meta lacks for these markets: 56 of Madgicx's 84 Google ads were shown in at least one APAC market, in English.

The library alone could not tell us which ads worked. To fill that gap, SOMIN scored each creative for predicted clickability, explained in what creative clickability is. The full read of what these vendors promise is in what ad optimisation platforms promise in their own ads.

What mistakes should you avoid?

Most errors come from treating the library as an archive or a scoreboard. It is neither.

• Reading absence as history. Outside the EU and UK, an ad that stopped last month is simply gone. Capture weekly, or use a tracker that does.
• Keyword searches. They mix in other advertisers that use the same words. Search by Page.
• Counting duplicates. Dynamic creative and copy tests multiply ads without multiplying messages.
• Assuming every ad is the brand's own marketing. A Page can carry ads for other products, as two of Strike Social's did. Read the ad before you count it.
• Treating volume or run length as performance. There is no spend or result data for commercial ads.
• Ignoring search and language. Check Google's centre as well, and filter by language: only one of the 633 ads in our example was in Thai.

FAQ

Sources

• About the Meta Ad Library, Meta Business Help Centre
• About dynamic creative in Meta Ads Manager, Meta Business Help Centre
• New features and additional transparency measures as the Digital Services Act comes into effect, Meta, August 2023
• Announcing the launch of the new Ads Transparency Center, Google