Notes

Brand virtual try-on limitations: why it only shows that brand

Last updated: 22 August 2026

A brand's virtual try-on is a sales tool built on one catalogue, so it can only ever point you back into its own range.

A brand's virtual try-on only ever shows you that brand's own products because the tool is a sales feature sitting on top of one catalogue, not an independent stylist. It cannot recommend a product it does not sell, cannot combine two brands into a single look, and any shade quiz built on top of it can only ever point back into that same limited range.

What the tool is actually for

Nobody hides this, and nobody needs to. A cosmetics brand pays to put a try-on widget on its own product page for the same reason it pays for the product photography above it: to move the items listed on that page. The widget was commissioned by the brand, built to the brand's specification, and stocked entirely with the brand's own line. It is doing its job correctly every time it shows you only what it was built to show you.

That is worth saying plainly before anything else here, because it changes what follows from a complaint into an observation. The tool is not failing to be neutral. It was never asked to be.

What a brand's try-on can and cannot do

Read the last row of each column together. A brand's try-on can find you something. It cannot tell you that the something is wrong.

Why can't the tool just point somewhere else, if the answer really is somewhere else?

Structurally, it cannot. A prestige counter's shade quiz has no path that ends in "the drugstore blush two aisles over is closer to your undertone than anything we sell." Recommending that would mean recommending a sale to a different company, on a page the first company paid to build. The same holds in reverse: a drugstore brand's tool will never point you toward a prestige product it does not stock either. Neither failure is about the technology being unable to compare two products. It is about who commissioned the comparison.

The shade quiz that cannot say no

Why does a brand's shade quiz always hand back an answer, even when nothing fits?

This is the part that goes unnoticed, because a returned answer looks like a successful one. A shade-matching tool built on top of a fixed range has exactly as many possible answers as that range has items, and not one more. Fenty Beauty's Pro Filt'r Soft Matte Longwear Foundation ships in 50 shades, according to Fenty Beauty's own product page, checked on 22 August 2026. A quiz sitting on top of that range has 50 possible outputs. It does not have a fifty-first, and it has no output that means "none of these fifty are right."

That is the design gap. A nearest-match tool, run on any catalogue of any size, always has a nearest item. It has no threshold below which it refuses to answer, because a "we don't carry your shade" screen ends the visit instead of moving it toward a basket. So the quiz returns its closest of 50, or its closest of 12 on a smaller range, with exactly the same confidence whether that closest item sits a hair's breadth away or well past the edge of what the range was built to cover. The person on the far end of a narrow range gets a shade name and a percentage next to it, and nothing on screen tells them the tool ran out of options before it ran out of confidence.

One engine, six brands, six catalogues that never touch

The clearest way to see the boundary is to watch the same underlying technology drawn around six different shelves. Perfect Corp licenses its try-on technology to beauty brands rather than building a separate one for each, and its own business page names the brands running it. Checked on 22 August 2026, perfectcorp.com/business lists KOSÉ America's Decorté, Benefit Cosmetics, M·A·C Cosmetics, Aveda, RegimenPro and Clinique among the companies using its technology.

The tracking code and the matching logic are the same six times over. What each install is permitted to show you is not. Benefit's version will never surface a Clinique concealer. Clinique's will never surface a Benefit blush. The same company wrote both, and technically could merge the two catalogues into a single recommendation. It has no reason to, because each installation exists to sell the shelf it sits on, not the shelf next door.

Does shopping at a retailer instead of one brand fix the catalogue problem?

It widens the boundary without removing it. A multi-brand retailer's try-on can move between brands the retailer stocks, so it clears the narrowest version of this problem: it will not be stuck inside one company's twelve shades. It is still stuck inside one warehouse. A retailer's tool cannot recommend a product a competing retailer sells and it does not, for the same commercial reason a single brand's tool cannot cross brands. The shelf got bigger. The rule that the recommendation has to resolve to a shelf did not change.

Fair, not cynical

Is a brand's try-on a bad tool, then?

No, and saying otherwise would be the more comfortable and less honest answer. For the one job it exists to do, which is choosing between two shades or two finishes of a product you already intend to buy from that brand, it remains the fastest and cheapest way to answer the question. It is free, it is instant, it is tied to a real item with a real price and a basket one tap away, and none of that changes because the tool cannot also solve a problem it was never asked to solve. Walking in expecting an independent stylist is the mistake, not the tool.

Where Looksmith sits differently

Looksmith is an iOS app from Mirable Labs, Inc., coming soon to iOS and not released yet. Mirable Labs does not sell cosmetics, so a Looksmith look is not built to move one company's shelf. You save a Likeness once, a set of selfies that becomes your reusable face, then get to a look one of three ways: Style Me puts you in several looks at once, Create Look takes a description in words, and Mirror a Look transfers the makeup from an inspiration photograph. Looksmith renders the complete look across twelve regions onto your own photograph and lists the products it used beside the look, and that list is not drawn from one locked catalogue the way a brand's own try-on is.

Looksmith's own term for a product alternative is a Similar Match, never a dupe, and that word choice carries a real design consequence. Because Looksmith was not built to protect one company's shelf, a Similar Match is not required to stay on that shelf. That is the design difference worth stating, and it is a difference in what the tool was built to do, not a claim that Looksmith's recommendations are more accurate than anyone else's. Looksmith has not measured or published an accuracy figure for how close a Similar Match comes to the product it replaces, here or anywhere else.

Looksmith grants the first six looks free at onboarding, then runs on credits bought in packs, with no subscription. Pricing is set at launch, and Looksmith has not launched, so there is no dollar figure to print here.

Where does Looksmith not solve this either?

Looksmith does not run a shade-matching quiz, and does not hand back a shade code or an undertone label from your photograph. What comes back from Looksmith is a rendered look, not a measurement of your skin, and Looksmith carries the same honest limit as everything else on this page: a plausible picture, not a promise that one specific tube produces that exact pixel.

The short version

A brand's try-on is not withholding a better answer out of carelessness. It structurally cannot give one, because the tool was built to sell a shelf, and a shelf has an edge. The part worth remembering is not that the tool favours its own products. It is that the tool has no way to tell you when none of its own products are actually right, so it hands you its closest one anyway, with the same confidence either way.