What nutra actually runs on Facebook: 2.6 million creatives, counted

Where the numbers come from

Every figure here comes from one measurement of our database taken on 1 September 2026. The database is built from the public Facebook Ad Library, and the vertical is decided by a model reading the creative itself — video frames and the text on the image — rather than by whatever the advertiser called their page.

Each number below is stated with the population it was measured over. That matters more than it sounds: a share of all 2.6M ads and a share of the classified subset are different claims, and quietly swapping one for the other is how this kind of article usually lies.

More than half of nutra is video

• Video — 1,416,461 ads, 54%.
• Static image — 719,757, 27%.
• Carousel — 493,213, 19%.

Worth keeping in mind when planning production: the market you are entering is half moving footage, and a static creative competes for attention against video.

One creative in five runs in more than one version

Near-identical variants of one creative are collapsed into a stack, which means we can count decisions rather than ads: how often an advertiser, having found an approach, went looking for its best version. Of 1,268,024 uncollapsed nutra creatives:

• 1,007,118 (79%) exist in a single version;
• 190,073 run in two to four;
• 43,395 in five to nine;
• 21,151 in ten to twenty-nine;
• 6,287 in thirty versions or more.

So 260,906 creatives — 21% — are the ones where somebody decided to keep paying and to scale. A single ad can be a hundred-dollar test; a stack of thirty versions cannot. If you are looking for something to learn from, that 21% is a better starting point than the feed at large.

Which categories fill the vertical

The offer classifier reads the ad copy and has reached 403,713 ads so far — 15% of the nutra corpus. The shares below are of that classified subset, not of the whole database.

• Anti-aging and skin — 70,258 (17%).
• Joints and back — 61,223 (15%).
• Diabetes — 47,086 (12%).
• Weight loss — 46,740 (12%).
• Male potency — 24,652 (6%).
• Hypertension — 23,538 (6%).
• Hair loss — 18,612 (5%).

Below those come vision, parasites and detox, immunity and memory, skin conditions, prostate, varicose veins and haemorrhoids, each between nine and fifteen thousand ads. The first four categories account for more than half the classified volume: age, joints, diabetes and weight are what the vertical turns on.

The caveat is not optional: 15% classified is a sample, not a census, and it leans toward creatives with legible copy. The ordering holds on a sample that size; the exact percentages do not.

What we are not publishing, and why

The two most quotable numbers in this subject are missing above, deliberately.

• Average creative lifespan. In the Facebook export, an ad's end date is the date of the snapshot, not the day the ad was switched off. We checked: among stopped nutra ads, the most common 'end dates' are our own crawl days, roughly six thousand rows each. An average lifespan computed from that column would describe our crawler's schedule, not the market.
• Per-country ad counts. The row stores the geo basket an ad was found in — composites such as 'CZ,HU,SK' in a single field — not the ad's targeting. 'N ads are running in Poland' is not a sentence this data can support, however confident it looks in a table.

Both figures appear regularly in write-ups of this vertical. There is nowhere to get them except those same columns — so an average creative lifespan quoted without an explanation of how it was derived is best read as an eyeball estimate.

What follows from this

• Treat video as the baseline format rather than an extension: half the market is already there.
• Search by number of versions, not down the general list — the 21% with stacks is what separates scaling from one-off tests.
• Do not assume the vertical reduces to weight loss: age and joints both run ahead of it by volume.
• Ask any source, including this one, what population a percentage was measured over. If there is no answer, there is no number.