Collagen Is on 113 Indian Skincare Brands and Sells Almost Nothing

Launched is not sold

Brand count measures how many companies shipped an ingredient. Amazon velocity measures whether anyone bought it. These are usually treated as the same signal. They are not, and in the two most-adopted cases here they invert.

Collagen sits on 113 Indian brands and returns ₹0.013 crore per Amazon SKU. Rice sits on 7 SKUs and returns ₹0.461, the highest we measured. A founder choosing a hero ingredient off a list of what is popular would pick the first and avoid the second.

Highlighted rows are ingredients adopted by 50 or more brands that return under ₹0.05 crore per SKU: widely launched, barely bought.

A ₹0.0 here means unmeasured, not zero. Amazon publishes a “bought in past month” figure on some listings and not others. Fermented, panthenol and glutathione show no revenue because no SKU we saw carried a published figure — not because nothing sold. The Amazon side rests on 735 SKUs across 19 brands, so any row under about ten SKUs is indicative and no more.

India takes about four years to go from one brand to ten

We dated every Indian launch from the brand’s own catalogue, which publishes the date each product went live. Across the 30 ingredients where enough brands are dated to measure it, the median gap from the first Indian brand to the tenth is 50 months.

The range around that median is what makes it useful. Hyaluronic acid took 21 months. Glycolic took 85. An ingredient does not diffuse at a rate you can assume; it diffuses at a rate you have to look up.

This is the part of the study that needs no Korean data at all, and it is the part we would trust most.

What is on Korean shelves and thin on Indian ones

PDRN is on all ten Korean brands and about 3% of Indian ones. It came first under two methods that share no logic: a hand-written ingredient list, and a bottom-up derivation that was given no vocabulary at all. Two independent instruments agreeing is the strongest single result in this study.

It is still not a prediction. Ten Korean catalogues is a small shelf, and we have no validated hit rate — no record of how often something heavy in Korea and thin in India goes on to arrive. Read this as a list of things worth watching, not a list of things that will happen.

Formats outrank ingredients, and a hand-written list could never have found that

Our first ingredient list was written from memory, and it produced a control group of nothing: every term in it appeared on both shelves. That was not a finding about the market. It was survivorship, with the author as the filter — we could only write down ingredients we already knew had crossed.

So the second pass supplied no vocabulary. It compared the full text of both shelves and let the differences surface on their own. What came back was not mostly actives.

Alongside the actives sit ampoule, toner pad, sun stick, hydrogel, double cleansing, low-pH — formats and concepts rather than molecules. An ingredient-only list could never have surfaced these, however many ingredients we had added to it. The thing India has imported least from Korean skincare is not a molecule. It is a product format.

Method, and what this is not

This is not a forecast. It has a control group and a watchlist and no validated hit rate. We have not shown that anything on the watchlist arrives, and we will not publish it as a prediction until the Korean sample is wider than ten brands.

The Korean shelf is ten catalogues. That is enough to say an ingredient is near-universal there and almost absent here. It is not enough to rank two ingredients against each other.

Source is each brand’s own public catalogue, read directly rather than through a marketplace. That is where the launch date comes from, and it is the cleanest source we have found: 224 of the 1,671 Indian brands we tried publish one openly.

Ingredient detection is from product text, not from a certified ingredient list. We tried reading published INCI and abandoned it: coverage died at 4–6% on both Amazon listings and brand sites. A product that mentions an ingredient in its title or description is counted; one that carries it without saying so is not.

The Amazon side is a different instrument from the catalogue side and is far thinner: 735 SKUs across 19 brands. It is there to answer whether an adopted ingredient sells, and it can carry no more weight than that.

One read — 12 August 2026. An arrival curve changes slowly, but a revenue figure does not.

Related measurements

We later measured UV filters across six South-East Asian markets from published ingredient lists, which is the same question asked where the data is better: that report is in the report library. For what Indian buyers actually spend on a category, the shelf measurements are more direct — sunscreen and face serums.

If you are choosing a hero ingredient or a format for a launch, the useful question is not what is popular. It is how many brands already ship it, how long the last one took to diffuse, and whether anything at all is selling. That is what we do for enterprise clients.