Open the Meta Ad Library, search a competitor, and you will see the same ad more than once. Sometimes twice. Sometimes eight times. Before you write down how many ads that brand is running, it is worth knowing how much of the list is genuinely different advertising and how much is one ad wearing several name tags.
So we counted properly. On 7 September 2026 we took up to 50 library entries for each of nine brands, downloaded every creative image, and hashed them. 370 entries resolved to 285 unique images, and to 221 unique combinations of ad copy and destination. Nearly a quarter of what the library listed was the identical image file, published again under a new ad ID.
The short answer
The Meta Ad Library shows the same ad multiple times because it lists ad IDs, not creatives. One creative can hold many ad IDs, for two different reasons:
- The advertiser split one ad across placements or audiences. Each split gets its own ID and its own library entry, all with the same launch date.
- The advertiser relaunched the ad later. Media buyers duplicate a fatiguing ad to reset delivery. The new ID carries a fresh start date, and the library treats it as a brand new ad.
Of the 69 byte-identical clusters in our snapshot, 18 were same-day splits and 51 were relaunches spread over time. Relaunching is by far the more common cause, and it is the one that quietly distorts what you think you are looking at.
What 370 library entries actually contain
Counting is not as simple as "deduplicate the list". There are three defensible numbers, and they say different things:
| Brand | Entries listed | Unique images | Unique copy + destination |
|---|---|---|---|
| Dunkin' | 49 | 37 | 19 |
| Gymshark | 40 | 27 | 16 |
| HelloFresh | 49 | 31 | 27 |
| Dior | 38 | 23 | 22 |
| Liquid Death | 37 | 27 | 22 |
| Oakley | 29 | 23 | 18 |
| True Protein | 50 | 45 | 37 |
| instant.one | 50 | 49 | 38 |
| Red Bull | 28 | 23 | 22 |
| Total | 370 | 285 | 221 |
23% of entries were byte-identical repeats. Not similar, not a variation. The same file, served again. That is the floor, and it is the number you can defend without argument.
40% of entries repeated a copy and destination pairing already in the list. That is the number to use when you are asking how many different ideas a brand is testing, because a fresh photo behind an unchanged headline and landing page is a variation, not a new angle.
Every brand duplicated. The lightest was instant.one, with 49 unique images across 50 entries. The heaviest at image level was Dior, listing 38 entries that held 23 images. On copy, Dunkin' was the most repetitive of all: 49 entries collapsed to 19 distinct copy and destination pairings.
We repeated the exercise on 15 separate capture days going back to 19 June, covering 639 entries that were confirmed live at the same moment. Every single snapshot showed duplication, ranging from 1.22x to 2.65x at copy level. This is not a quirk of one day.
The start date is not the age of the ad
This is the part that costs people money, because the library's start date is the most common shortcut for judging whether an ad is working.
Take Oakley. On 7 September its library carried three entries for the same Transitions lens ad, all running simultaneously, all pointing at the same landing page, all carrying the same copy: "One pair, all day: ultra-dynamic lenses that rapidly adapt to all light conditions."
Two of those three entries were the identical image file. One started on 22 January 2026. The other started on 22 July 2026. Same picture, same words, same destination, relaunched 181 days apart and both live on the day we measured.
Read the July entry and the ad looks 48 days old, which is unremarkable. Read the January entry and the same ad has been in market 229 days, which puts it in the top few percent of everything we track for longevity. The ad did not change. Only the ID did.
Across all staggered clusters in the snapshot, the median gap between relaunches was 4 days. That is the everyday case: an ad quietly restarting every few days, resetting its visible age each time, while the library presents each restart as something new.
How to count a competitor's ads properly
Four habits fix most of the damage.
Count creatives, not entries. If you are reporting "they are running 49 ads", check whether that is 49 pictures or 19 ideas. In our sample the honest figure was usually somewhere between half and three quarters of the headline count.
Treat repeat count as a signal rather than noise. Nobody splits an ad six ways or relaunches it four times for fun. When Dunkin' runs one recruitment ad across six entries, all launched on 27 July, all pointing at the same careers page, that concentration is telling you where the budget went. Deduplicate for your count, but keep the tally.
Look for the oldest copy, not the newest. When the same creative appears several times, the earliest start date is the one that tells you how long the idea has survived. The newest entry tells you only when they last hit restart. We wrote about why longevity is the most reliable public proof signal in how long Facebook ads actually run.
Do not read a fresh start date as a fresh test. Roughly three quarters of the byte-identical repeats we found were relaunches rather than same-day splits. A new date usually means an ad was restarted, not that anything was learned. For the wider pattern, see how often brands refresh their Facebook ads.
What this data cannot tell you
We would rather be useful than impressive, so here are the limits.
We cannot see spend or impressions. Meta publishes those only for political and social-issue advertising. Across every commercial record we hold, both fields are empty. Any tool quoting you a competitor's Meta spend is modelling it from assumptions, not reading it. We covered that in how much competitors spend on Facebook ads.
We cannot see why a duplicate exists. A same-day split could be placement testing, audience segmentation or geography. We can tell you the shape of the pattern, not the intent behind it.
We measured up to 50 entries per brand, not entire libraries. We excluded 28 entries whose ad copy was blank, because an empty text field cannot evidence that two ads are the same, and one further brand that fell below our minimum of 25 usable entries. The brands here are large advertisers running international campaigns, so a small local business will look different.
Frequently asked questions
Why does the Meta Ad Library show the same ad twice?
Because it lists ad IDs rather than creatives. An advertiser who splits one ad across placements or audiences, or who relaunches it to reset delivery, generates a new ID each time. Each ID becomes its own library entry even when the image, copy and landing page are unchanged.
How many of a competitor's listed ads are actually duplicates?
In 370 entries across nine brands measured on 7 September 2026, 23% were byte-identical repeats of another entry and 40% repeated a copy and destination pairing already in the list. Every brand in the sample showed some duplication.
Does a duplicate ad mean the advertiser made a mistake?
Usually not. Duplicating a fatiguing ad is a deliberate tactic to reset delivery, and splitting one creative across placements is normal campaign structure. The duplication is a side effect of how the library reports, not a sign of sloppiness.
Why do two identical ads show different start dates?
Because the start date belongs to the ad ID, not the creative. When an advertiser relaunches an ad, the new ID starts its clock at zero. We found one Oakley ad running as two identical entries with start dates 181 days apart.
Should I deduplicate before counting a competitor's ads?
Yes, but keep the original tally. The deduplicated number tells you how many ideas they are running. The repeat count tells you which of those ideas they are backing hardest, and that is often the more useful of the two.
Can I tell which duplicate is the original?
The entry with the earliest start date is the earliest one still running. It is not necessarily the first ever launched, because a version that has since been switched off disappears from the live library entirely.
See this for your own market
Every figure here came from ads anybody can look up for free. The work was in collecting them, hashing the creatives, grouping the repeats and reading the start dates as a set instead of one at a time.
That is what SOCIALFUEL does. Search any brand, see their live ads across Meta, Instagram, Google and TikTok with the long-runners surfaced first, then decode why each one works. It is free to start, with no credit card and no ad-account connection.
We will re-run this count each quarter. The number worth watching is whether relaunching keeps climbing as a share of what the library shows.
About the author
Joel Helou is the founder of SOCIALFUEL, a growth agency that manages paid social and search campaigns for ecommerce and DTC brands, and the creator of socialfuel.io. He has spent the past decade buying media for brands across apparel, supplements and consumer goods, and now builds the tooling his own team uses to decode why competitors' ads work. This study was run on data captured through that platform, measured on 7 September 2026.