Up to 60% off your first month or first year with code BFCM60. Ends 1 December, 23:59 US Pacific.60% off · BFCM60 · ends 1 Dec

10d 16h 10m 31s10d 16h

See the offer
Face Age

How Automated Beauty Recommendations Work: From Skin Scan to Product List

Automated beauty recommendations work in three steps: collect inputs about the shopper, match those inputs against rules attached to each product, then filter and rank what's left into a short list. In a skincare store, the inputs are usually a skin scan and a few quiz answers. The matching is mostly rules you control, not a black box. That's good news, because it means a weak recommendation is almost always fixable.

People ask "how do automated beauty recommendations work?" because the result can look a bit magic from the outside. It isn't. Here's the plumbing, explained the way I'd explain it to a store owner who has to set it up.

The three inputs behind automated beauty recommendations

1. What the shopper shows you. A skin scan reads a photo. Face Age analyzes 13 visible concerns (wrinkles, pores, pigment, redness, oiliness, acne, dark circles and others) across 250+ attributes, and gives each concern a severity: low, medium or high. Not a medical diagnosis; does not replace a dermatologist.

2. What the shopper tells you. Quiz answers cover what a photo can't: self-reported sensitivity, preferred textures, how many steps someone will actually do, budget, fragrance preferences.

3. What the shopper has done before. Previous scans, if history is on, and past orders. This input matters more for returning shoppers than for first-timers.

A lot of tools stop at input 2, which is just a product finder quiz. Others stop at input 1 and show products for whatever the camera saw. The better results come from combining the first two, then using the third to adjust over time. Our comparison of a skin quiz vs an AI skin analyzer goes deeper into which input does what.

The matching step: rules, not mind reading

This is the part most people picture as "the algorithm." In practice it's a set of conditions on each product.

In Face Age, every product you connect gets:

  • Concerns and severity levels it suits. A pore-refining toner might be set to pores, high only. A gentle daily moisturizer might cover several concerns at every level.
  • A routine slot, morning or night, or none if it doesn't belong in a routine.
  • Extra conditions from quiz answers, which can include a product ("show this SPF to anyone who says they're outdoors a lot") or exclude it ("hide this exfoliant if they say their skin is sensitive").

The logic is simple once you see it written down. Rules in the main Condition box use AND, so every one must match. Extra Conditions use OR, so any one is enough. And there's a fixed order of priority: exclusions first, then quiz-based includes, then the scan matches. An exclusion always wins. Our pillar guide on how AI skin analysis picks products has worked examples of each.

So is there machine learning involved? Yes, in reading the photo. The product side is mostly rules, and that's deliberate. A store needs to know why a product appeared, and "the model felt like it" isn't an answer your support team can give.

How the shortlist gets built

Matching usually leaves more candidates than anyone wants to scroll through. Three settings narrow it:

Display limits. In Customization > Basic you can cap the total number of products and the number per concern, so a shopper with five medium concerns doesn't get twenty products.

Product priority. Decide which products show first when several qualify. Hero products or ones with strong reviews are the usual picks.

Concern switches. If you don't sell anything for eye bags, turn that concern off. Otherwise the result points at an empty shelf, which looks worse than not mentioning it.

Then the result screen ties it together: Skin Report, Routine, and Products tabs, with the product recommendations labelled by the concern they're mapped to. That label does a lot of trust work. "Recommended for medium pores" explains itself.

Automating the mapping itself

Mapping a big catalog by hand is the slow part. Face Age has an Automate AI option on the Connect Products page that reads product names and descriptions and links products to concerns for you.

How I'd use it:

  1. Clean up product descriptions first. If a serum's description never mentions what it's for, the automation has nothing to go on.
  2. Run Automate AI with high strictness on a first pass. High makes no guesses and links each product to fewer concerns. Medium and low link more broadly.
  3. Choose whether new links replace or add to existing ones. On a fresh setup, it doesn't matter.
  4. Open the product list and review every product by hand. Fix severity levels, add routine slots, add exclusions.
  5. Rescan with a few test photos and read the results like a shopper would.

The full catalog walkthrough is in how to connect AI skin analysis to your product catalog.

Why some automated recommendations feel generic

When a store tells me the results "all look the same," it's nearly always one of these:

Everything is tagged for everything. If every moisturizer is mapped to all 13 concerns at all levels, every shopper sees the same moisturizers. Narrow the tags.

No severity split. Your strongest retinoid shouldn't appear for a low fine-lines reading. Use the high level for strong products and low or medium for gentler ones.

No exclusions. Without them, a shopper who said "sensitive" can still get a strong acid. That's the fastest way to lose trust.

Too few products in a concern. If you only sell one product for pigment, everyone with pigment gets that product. That might be fine, but know it's happening.

Testing helps more than theory here. Our guide to validating AI skincare recommendations across ages, skin tones and concerns has a test plan you can run in an afternoon.

Where people still beat the automation

Automation is good at consistency. It shows the same logic to the thousandth shopper as to the first, at 3 a.m., in the store's language. It's less good at judgement calls: a shopper who's pregnant and unsure about an ingredient, someone with a skin condition that needs a professional, a gift buyer who isn't scanning their own face.

That's why I'd keep a human route visible: a chat link or email on the result screen. We compared the two in AI beauty consultant vs human beauty consultant, and the honest answer is that each covers the other's gaps.

Wording that keeps it honest

A few lines I'd keep out of your result copy and ads:

  • "Our AI knows your skin better than you do." It doesn't. It read a photo.
  • Any percentage about results or accuracy. You can't support it, and you don't need it.
  • "Prescribed for you." It's a recommendation from your catalog.

And lines that work: "Picked from our range for what your scan showed," "Why this product: matched to medium redness," and "Not sure? Ask our team."

What it costs to run

Product recommendations are on every Face Age Shopify plan, from Start at $59/month. Plus ($129), Growth ($249) and Scale ($499) add scan history, comparison and PDF reports, which is what you'd want once you start adjusting recommendations for returning shoppers. There's a 7-day free trial, and you can install it from the Face Age Shopify page.

FAQ

How do automated beauty recommendations work? Inputs (scan, quiz answers, history) are matched against rules on each product, then filtered and ranked into a short list.

Is it machine learning or rules? Both. AI reads the photo; the product step is mostly rules the store controls.

Can a store control what gets recommended? Yes: concerns and severity per product, quiz-based include and exclude rules, display limits and priority. Exclusions always win.

Why do some recommendations feel generic? Loose mapping, no severity split, missing exclusions or thin coverage in a concern.

Are automated recommendations medical advice? No. Not a medical diagnosis; does not replace a dermatologist.

Share
Ask AI for a summary
Article byJJ SEOadmin

Put a skin scan on your store today

Add Face Age to your site with the SDK, a plugin or the Shopify app — and start with a 7-day free trial.

Choose which cookies Face Age can use on this site. You can change this any time from “Cookie preferences” in the footer. Privacy Policy