Video or image? What e-commerce shops' own same-day launches show
Published 13 Nov 2026. Data read Meta's EU reach read 2026-11-11 to 2026-11-13; fact computed 2026-11-13 05:28 UTC. Source: Meta's Ad Library (EU transparency data), read by Poseidon, unless the text names another source.
When an e-commerce shop launches a video and an image on Meta the same day, the video reached more people in about two shops out of three, as read on 11 to 13 November 2026. The exact counts, the likely range and the chance of luck are in the table below, recomputed every morning; the lean toward video is real across shops, not luck.
This is Poseidon's first measured fact. It's about e-commerce shops in general, not about one niche: in weight loss and in menopause alone, there aren't enough shops yet to say.
The fact
| Measured | |
|---|---|
| Shops that launched a video and an image the same day, with reach read for both | 124 |
| Shops where one format won most of their pairs | 112 |
| Shops where the video reached more people | 72 |
| Share | 64% |
| Likely range (95%) | 55% to 73% |
| Chance this is luck | 0.3% (p = 0.003) |
| Reach read on | 11, 12 and 13 November 2026 |
How it was measured, in plain words
- Same shop, same day. We take a shop that started a video ad and an image ad on the same day. Same brand, same week, same budget decisions: the main difference is the format.
- Same reading day. We compare how many people in the EU each ad had reached, as Meta reports it, on a day both were read.
- One vote per shop. A shop with many pairs votes once, for the format that won most of its pairs. A shop that split evenly doesn't vote. A shop running several Meta pages counts once.
- Then we check it's not chance. If video and image were equal, about half the shops would pick video. The count in the table is far enough from half that luck would rarely produce it; the table gives the exact chance.
How solid it is
We re-ran the count leaving out one country at a time, then one niche at a time. Video stays ahead every time.
One caveat. About half these shops have no checked niche yet. On the shops that do, video still leads, but they alone aren't enough to pass the test (see the caveats below). That's why this page says "e-commerce shops", not a niche.
The caveats served with the fact
The caveats Poseidon's API serves with this fact:
- About 50% of the shops have no checked niche yet; the checked shops alone do not pass (34 of 56, p = 0.14).
- Based on 3 days of EU reach readings (11 Nov 2026, 12 Nov 2026, 13 Nov 2026); it can change as more days are read.
- Reach is the number of people Meta showed the ad to in the EU, not sales; Meta decides delivery.
What this doesn't say
- Not for your niche. In weight loss and in menopause alone, there aren't enough shops to say anything yet. The fact is shown again per niche when one passes.
- Not sales. Reach is how many people saw the ad, not who bought.
- Not that every video beats every image. In about one shop in three, the image won.
- Not why. Meta may simply show videos to more people. The fact describes what happened to these shops' own ads.
What counts as passing
A fact is shown only while it passes all three: 30 shops or more, less than a 5% chance of luck, and a likely range that doesn't include 50%. It's recomputed every morning. If it stops passing, this page comes down by itself.
FAQ
Should I make videos instead of images?
Across the shops in the table, the video reached more people more often. Test both on the same day, as these shops did, and compare reach.
Why compare on the same day?
Reach grows with time. Two ads started the same day have had the same time, so the gap comes from the ad, not the calendar.
Where can I get this from my AI assistant?
Through Poseidon's API and MCP server, which serve the same facts with the same figures.
Ask Poseidon what wins in your market. askposeidon.app