What won brands' own A/B tests on Meta in Europe this week

Published 13 Nov 2026. Data read 2026-11-11 and 2026-11-12 (frozen 2026-11-12 22:48 UTC). Source: Meta's Ad Library (EU transparency data), read by Poseidon, unless the text names another source.

In short, as read on 12 November 2026. Brands test their own ads all the time. They launch the same text with different videos, or the same video with different words, often on the same day. Meta then publishes how many people each version reached. Across 81 tests where the versions went live the same day, with the same text and a different video, the best version reached a median of 6.5 times more people than the weakest, and in 80% of them at least 3 times more. Same brand, same day, same words. Only the video changed.

Every number is Meta's own EU reach (the total of people in the EU who saw the ad), read by Poseidon on 11 and 12 November 2026. Links go to both versions in the Meta Ad Library, so you can watch them side by side.

How many tests did we find?

What the brand changed	Test groups found	With a measured winner
The video (same text)	2,498	193
The text (same video)	114	3
The first line only (same body text)	1	1

A group is ads from the same brand, launched within 14 days, that share one element and differ in another. A winner is measured when one version is still running while the others stopped, or when its EU reach is at least twice another's. Below, we keep only tests where the best version reached 100,000 people or more.

Same text, different video, same day, the biggest gaps

One row per brand. Both versions launched the same day and were still running on 12 November.

Shop	Main country	Launched	Versions	Best video, EU reach	Weakest video, EU reach	Gap	Both ads
lanuvi.com	Germany	2026-07-23	8	800,212	3,196	250.4×	ad · ad
kijimea.it	Italy	2026-05-13	4	5,528,270	32,131	172.1×	ad · ad
hairfiller.se	Sweden	2026-09-01	2	303,513	3,405	89.1×	ad · ad
fezi.pl	Poland	2026-06-03	5	638,201	7,162	89.1×	ad · ad
dahliabeauty.nl	the Netherlands	2026-10-08	7	364,470	4,676	77.9×	ad · ad
kijimea.de	Germany	2026-04-22	5	1,565,401	37,320	41.9×	ad · ad
aldo-masolo.it	Italy	2026-09-11	6	1,304,854	39,200	33.3×	ad · ad
kijimea.pl	Poland	2026-04-08	3	1,094,834	34,208	32.0×	ad · ad
adibody.com	Italy	2025-11-17	6	1,424,459	67,288	21.2×	ad · ad

What this shows, without reading into it.

The video can be most of the result. Kijimea in Italy ran the same text, « Diarrea, dolori addominali, gonfiore? », on four videos launched on 13 May 2026. One reached 5,528,270 people. Another reached 32,131.
It happens in every big market. The 81 same-day video tests come from 46 shops, mainly showing in Germany (20), Italy (18), Poland (14), France (12), the Netherlands (7) and Spain (6).
Reach is not sales. Meta pushed one version much further. That's the clearest public signal of what the brand's own numbers favoured, but it isn't revenue.

When the brand changed the words

Only four tests where the words changed have a measured winner. They're worth reading line by line.

Shop	What changed	Launched	Best, EU reach	Weakest, EU reach	Gap
byheartjewelry.com	7 texts on the same video	9 to 19 Apr 2026	3,590,391	290,483	12.4×
byheartjewelry.com	2 texts on the same video	same day, 29 May 2026	263,551	37,920	7.0×
hansegruen.de	2 texts on the same video	same day, 14 Mar 2026	2,123,088	380,533	5.6×
nootri.pl	the first line only, same image and body	same day, 9 Mar 2026	2,841,003	552,802	5.1×

The words, verbatim, with a translation.

byheartjewelry.com, 7 texts. Best « A, M, L, S — die Initialen meiner 4 Kinder. » ("A, M, L, S, the initials of my 4 kids.") ad. Weakest « Noch kein Geschenk zum Muttertag? » ("No gift yet for Mother's Day?") ad. The weakest launched 10 days later, so it had less time.
byheartjewelry.com, same day. Best « Meine Mama hatte Tränen in den Augen, als sie ihre Initiale auf der Kette sah. » ("My mom had tears in her eyes when she saw her initial on the necklace.") ad. Weakest « Es gibt einen Buchstaben, den du hundertmal am Tag sagst. » ("There's a letter you say a hundred times a day.") ad.
hansegruen.de, same day. Best opens on the product, « 60 Bio Ingwer-Kurkuma-Shots » ("60 organic ginger-turmeric shots") ad. Weakest opens on a question, « Du willst dich ideal vor Erkältungen schützen? » ("Want the best protection against colds?") ad.
nootri.pl, same day. Best « Prawdziwy powód, dlaczego coraz więcej osób nie może spać spokojnie. » ("The real reason more and more people can't sleep peacefully.") ad. Weakest « Nowe badanie odkryło prawdziwy powód… » ("A new study found the real reason…") ad.

Four tests is too few for a rule. Check them yourself on Meta.

How this was measured

Source: Meta's EU transparency data on each ad in the Ad Library. Read by Poseidon on 11 and 12 November 2026, snapshot frozen at 22:48 UTC on 12 November. Groups found by Poseidon's test finder (same brand, launched within 14 days; texts 90% alike count as the same; the same preview image counts as the same video).
Reach is Meta's count of people in the EU who saw the ad at least once. It is the only number of people shown here.
Main country: the country with the largest part of the best ad's breakdown.
Shops are shown by the domain of the ad's landing link, never by a person's page name.
What this can't tell you: sales, spend, or why Meta pushed one version. Videos are not described or copied, only linked.

FAQ

Why compare ads launched the same day?

Reach adds up over time. Same-day versions have had exactly the same time, so the gap comes from the creative, not the calendar.

Can I get these tests for my niche?

Yes. Poseidon finds them for any brand it follows. Ask "which videos won brands' own tests in weight loss in France?"

"All of these spy tools just regurgitate the Ad Library, while we have deep analysis and rank for each ad." Poseidon finds the tests brands run against themselves and shows which version won, with Meta's own numbers. Ask Poseidon what's winning →

Ask Poseidon what's winning. askposeidon.app