Facebook & InstagramCollect the Facebook and Instagram ads your competitors are running. Look for repeated offers and messages, then use those examples to plan your own creative tests.
Try these examples for marketing, growth and competitor research.
If you run a subscription brand, track five competitors every Monday. Group new ads by the problem they address, the promise they make, the proof they use and the offer. Pick three angles for your next round of tests.
You'll have a Google Sheet with creative links, hook categories, your notes and a proposed test.
Collect ads daily around a seasonal sale. Compare new ad IDs and copy changes to see when competitors introduce discounts, bundles or delivery promises.
Keep a launch timeline and an offer library with your annotations.
Search a category such as running shoes to find advertiser pages beyond your current shortlist. Review the results before adding them to your regular watchlist.
Build a broader competitor list and decide which advertisers to research first.
FROM FIRST RUN TO REPEATABLE WORKFLOW
{
"pageIDs": ["ZARA"],
"country": "GB",
"activeStatus": "active",
"maxAds": 50,
"proxyConfiguration": {
"useApifyProxy": true,
"apifyProxyGroups": ["RESIDENTIAL"]
}
}Paste this into the Actor’s JSON input editor. Replace the example targets with yours before running.
Check the current input form on ApifyChoose a page when you know the advertiser, or a keyword when you want to discover ads around a topic. These are separate searches: a keyword does not automatically narrow every page you entered.
| Search method | When to use it | What changes |
|---|---|---|
| Advertiser or brand name | When to use itYou have a competitor watchlist. | What changesProvide numeric page IDs, names, usernames or Facebook page links. Names are resolved first: verified matches win; multiple unverified exact matches can expand into several pages. |
| Keyword discovery | When to use itYou want to find relevant results before choosing specific targets. | What changesSeveral topic searches sharing the same country, dates and match type. They share one maxAds cap, so overlapping results are deduplicated rather than counted twice. |
| Ad Library URLs | When to use itYou already configured the search on the source website. | What changesPaste page, keyword-search or individual-ad URLs. A search URL’s country, dates and status override global settings; a single-ad URL fetches that ad without those filters. |
Start of a delivery window, using YYYY-MM-DD. Ads started earlier can still match if they ran during the window. Equal start and end dates are widened by one day.
Compare the related scraperUse the exact field names below in JSON. In Apify’s form, enter list items separately, choose filters, and keep numbers and booleans in their proper types.
Default and prefill are different. A default applies when you omit a setting; a prefill is an example already entered in Apify’s form. Review prefilled targets and limits before every run. Some settings have no schema default. You still need to supply at least one supported target.
adLibraryUrlsPaste page, keyword-search or individual-ad URLs. A search URL’s country, dates and status override global settings; a single-ad URL fetches that ad without those filters.
adLibraryUrlThe single-URL version of adLibraryUrls. Use one field or the other for clarity. Ordinary Facebook page links use your global filters.
pageIDsProvide numeric page IDs, names, usernames or Facebook page links. Names are resolved first: verified matches win; multiple unverified exact matches can expand into several pages.
pageIDOne page, using the same forms as pageIDs. Clear the form’s prefilled page when you only want a keyword search, otherwise both targets are collected.
keywordFind ads mentioning one topic across advertisers. This is separate from page lookup and runs alongside any page targets you left in the input.
keywordsSeveral topic searches sharing the same country, dates and match type. They share one maxAds cap, so overlapping results are deduplicated rather than counted twice.
searchTypekeyword_unordered matches words in any order. keyword_exact_phrase requires them together in that order. It only affects keyword searches; a pasted URL’s search_type takes precedence.
keyword_unordered keyword_exact_phraseminDateStart of a delivery window, using YYYY-MM-DD. Ads started earlier can still match if they ran during the window. Equal start and end dates are widened by one day.
maxDateInclusive end of the delivery window. Together with minDate, it selects ads running during the period, not only ads first created during it.
activeStatusactive collects currently running ads; inactive collects stopped ads; all includes archive history. The form starts with active, but omitting this field uses all.
all active inactivecountryOne country code or name, or ALL for every market. Unrecognized countries fall back to ALL with a warning, so check the log if your market filter seems too broad.
proxyConfigurationUse an Apify residential proxy. Datacenter exits are often challenged. Keep useApifyProxy true and apifyProxyGroups set to RESIDENTIAL for a normal run.
maxConcurrency1-6 parallel sessions. Several targets and large date ranges can run together. Use 1 for strict ranking order or when captchas appear; the default is 4.
maxAdsA hard cap on unique ads across the whole run, including all targets and markets. Start with 50. Leaving it empty removes this cap; a broad search can then become much larger.
maxPagesLimits pages in each date shard, not the whole run. Useful for testing; clear it for fuller coverage. Zero or empty means unlimited.
delayMsPause between page requests, in milliseconds. Increase it when the source throttles you; reducing it does not remove network delays or the source’s rate limits.
pageSizeThe requested batch size, not a guaranteed result count. Meta actually paginates about 10 ads at a time; raising this value will not remove that limit.
rawReturns the source structure rather than the normalized presentation. Keep false for consistent analysis columns; use true when you need the source payload for debugging or your own transformations.
proxyRotationsRetries refused work with a new proxy session. More retries can recover temporary blocks but add time and traffic. Keep the default until the log shows a reason to change it.
resumeSaves progress about every 30 seconds so an Apify restart or migration can continue the current run. Leave it on for normal use.
continueFromLastRunContinues unfinished work from the previous run with matching input. Earlier results remain in that run’s dataset. Keep false for a fresh collection or a recurring snapshot.
shardSplitThresholdLarge date ranges are split when their reported ad count exceeds this threshold. It matters only with maxConcurrency above 1. Keep 1500 unless tuning large archive runs; values below 100 become 100.
docIdAn internal request-document override. Leave empty: the Actor detects and refreshes stale IDs automatically. It is not an advertiser or ad ID.
proxySessionIdPins a proxy session for debugging. Leave empty for routine collection. Parallel jobs need their own sessions, so this setting is ignored when parallelism prevents a single fixed session.
impersonateThe browser identity used for requests. Keep the Actor’s default unless you are diagnosing immediate blocks. It changes the request fingerprint, not the data you ask for.
This reference follows the Actor’s published input fields. Check the live form before changing a production workflow. Check the current input form on Apify.
Each example is a separate run. Start small, inspect the results, then increase coverage. Update the targets, countries and dates to match your question.
Collect active UK ads from two brands, with 100 unique ads across the whole run. Schedule a fresh run each week, append a collection date in Sheets and compare new creative IDs with last week’s snapshot.
{
"pageIDs": ["ZARA", "H&M"],
"country": "GB",
"activeStatus": "active",
"maxAds": 100,
"proxyConfiguration": {
"useApifyProxy": true,
"apifyProxyGroups": ["RESIDENTIAL"]
}
}Find US ads using the phrase “free shipping”. This is topic discovery across advertisers. Replace the phrase with your offer and clear any prefilled page target before running.
{
"keywords": ["free shipping"],
"searchType": "keyword_exact_phrase",
"country": "US",
"activeStatus": "active",
"maxAds": 50,
"proxyConfiguration": {
"useApifyProxy": true,
"apifyProxyGroups": ["RESIDENTIAL"]
}
}Use this September 2026 example to review ads delivering during a window. Change both dates for your research. These are delivery-overlap filters, so an ad may have been created before the first date.
{
"pageIDs": ["ZARA"],
"country": "GB",
"activeStatus": "all",
"minDate": "2026-09-01",
"maxDate": "2026-09-30",
"maxAds": 100,
"maxConcurrency": 4,
"proxyConfiguration": {
"useApifyProxy": true,
"apifyProxyGroups": ["RESIDENTIAL"]
}
}Use page and creative IDs as your joins, and retain source links, ad text, media and delivery dates. For commercial ads, missing spend or reach is normal; the library does not supply conversion results. Store weekly snapshots in BigQuery, flatten media arrays with dbt and build a Looker Studio view of newly observed creatives.
resume protects the current run if Apify restarts it. continueFromLastRun continues an earlier run with the same configuration; earlier records stay in the earlier dataset. Combine both datasets for the complete collection, and raise a previously reached result cap when continuing. Start fresh when you want to see what changed today.
Save one configuration above as input.json. Set APIFY_TOKEN to your Apify API token in your terminal, then send the file as the request body.
curl --fail-with-body --request POST \
--url "https://api.apify.com/v2/actors/jmlp~meta-ad-library-scraper/runs" \
--header "Authorization: Bearer $APIFY_TOKEN" \
--header "Content-Type: application/json" \
--data-binary @input.jsonThe response contains a run ID and defaultDatasetId, not finished results. Wait for the run to succeed, set DATASET_ID to that dataset ID, then fetch its items. For large datasets, use limit and offset to page through the export.
curl --fail-with-body \
--url "https://api.apify.com/v2/datasets/$DATASET_ID/items?format=json" \
--header "Authorization: Bearer $APIFY_TOKEN"Change one setting at a time, keep a small cap, and check the run summary before scaling up.
Provide numeric page IDs, names, usernames or Facebook page links. Names are resolved first: verified matches win; multiple unverified exact matches can expand into several pages.
Use an Apify residential proxy. Datacenter exits are often challenged. Keep useApifyProxy true and apifyProxyGroups set to RESIDENTIAL for a normal run.
An internal request-document override. Leave empty: the Actor detects and refreshes stale IDs automatically. It is not an advertiser or ad ID.
Success means the Actor finished handling the request, not that the source returned data. Check SUMMARY.inputProblem, SUMMARY.problem and the log for missing targets, unsupported filters or refused requests. Test one known target with fewer filters.
Check the global limit, per-search or per-page limits, platform coverage and deduplication. Several searches can find the same record. A source’s headline count can include records the public endpoint does not return. Review unfinished jobs before treating the dataset as complete.
Use Make or n8n to append records after each successful run. Remove duplicates using ad_archive_id and add collected_at. Leave columns for your own hook labels, offer notes and next test.
Read the setupAsk Claude to collect a capped number of ads for one advertiser page, group the messages by theme and cite ad_archive_id. Check the original creatives before using the brief.
Read the setupPut the fields you need in a sheet or BigQuery table. Chart ad counts by advertiser, start date and publisher platform. Label the counts as observations from the library.
Read the setupSave the raw JSON for each run with its collection date. Use dbt to unpack snapshot and keep a separate creative table keyed by ad_archive_id.
Read the setup
Use jmlp/meta-ad-library-scraper to collect up to 50 active ads for ZARA in GB. First check the Actor input schema. Group the creative messages by hook and offer, cite the original ad IDs, and propose three testable campaign hypotheses. Do not infer spend or conversion performance.Keep the collection time and original IDs with your records. You’ll need them to check where a result came from or compare it with a later run.
Commercial ads usually don't disclose spend or reach. Meta mainly publishes those figures for political and social-issue ads. If an ad has been running for a long time, it's worth reviewing, but its age doesn't tell you its conversion rate, profitability or ROAS.
Yes. The scraper reads the public Ad Library, so it does not require a Facebook account. You still need an Apify account to run it.
You can investigate recurring messaging and long-running creatives. The library does not give you the conversion data needed to call an ad a winner.
Source and current product details: JMLP’s Meta Ad Library Actor on Apify.
Search, YouTube & Display01See a brand's ads across Google.
Collect advertiser creatives from Search, YouTube and Display. Compare formats and markets to plan where your own campaign might fit.
Commercial Content Library02Find video ideas worth testing.
Collect TikTok ads from 32 markets across the EU/EEA, UK and Switzerland. Save video links and delivery dates, then review the creatives you want to learn from.
Tell me what you need to collect or understand. I can help with a custom scraper, a pipeline or the analysis.
Start a project