AI Skincare Recommendations: A Guide for Beauty E-Shops

A beauty shopper rarely searches for a product name. They describe their skin: "dry and a bit sensitive", "oily with blackheads", "tight after cleansing". In a shop, a good consultant asks two or three questions and walks them to the right shelf. Online, the same shopper usually lands on a category page with dozens of serums and a filter sidebar.

This article explains how AskMerra handles skincare conversations for beauty e-shops: what it asks, what it needs to know about your products, how it puts together a simple routine, and where it deliberately holds back.

Why skincare is hard to sell with filters alone

Filters work when the shopper already knows the vocabulary. Skincare questions are about combinations. Someone with oily skin and blackheads, a budget, a sensitive patch around the nose and a dislike of fragrance is not one filter. They are five, and they probably do not know which of your attributes match which of their worries.

Shoppers also use their own words. One writes "breakouts", another "spots", a third writes in Romanian or Italian. A conversation handles this better than a checkbox list. We compare the two approaches in more detail in conversational product search vs keyword search.

The assistant asks before it recommends

When a request is too vague to recommend well, such as "Which moisturizer should I buy?", AskMerra asks one short clarifying question, usually about skin type or budget, and still suggests something useful where it can. It does not run the shopper through a quiz.

From then on it respects what the shopper said: budget, skin type, preferences and allergies. It never recommends a product above a stated budget.

Context helps too. On a product page, AskMerra knows which product the visitor is looking at, so "Is this good for oily skin?" gets an answer about that product. And the assistant replies in English, Romanian, Italian, French, German or Spanish, whichever the shopper writes in, even if your catalog is in one language only.

What the assistant knows about each product

Enrichment does the reading

When you import your catalog, AskMerra analyses each new or changed product and extracts what shoppers ask about: use cases, skin and hair types, concerns, key ingredients, routine steps and safety notes. The result is indexed for semantic search.

When a shopper asks a question, candidates come from three sources combined: semantic search over the enriched products, keyword search on names, brands and SKUs, and the concern and attribute tags from enrichment. Then hard filters apply, such as a price range like "under €30", and the results are ranked by stock and relevance.

Your own attributes count too

If your export already has columns such as skin type, ingredients or "vegan", map them as product attributes during import (for example attribute:skin_type). Ingredient-style columns are suggested as attributes automatically.

You stay in control. In the dashboard you can inspect each product's enrichment and edit the tags. Your manual edits are kept when the product is enriched again after a content change.

From single products to a simple routine

Skincare is rarely one product. Because enrichment records the routine step of each product (cleanser, serum, moisturizer and so on), AskMerra can link products that pair well: products from a different routine step that address the same concern. A hydrating serum and a barrier cream for dry skin are a typical pair.

That lets the assistant answer "What routine do you suggest?" with a short, practical sequence built from products you actually sell, instead of a list of five unrelated best sellers.

An example conversation

Here is an illustrative exchange on a fictional beauty shop. The product names are made up.

Shopper: My skin gets oily during the day and I have blackheads on my nose. What should I use?

Assistant: Happy to help. Is your skin also sensitive or easily irritated? And do you have a budget per product?

Shopper: A bit sensitive. Under €30 per product, please.

Assistant: Then keep it simple. In the evening, use the Clear Pore Gel Cleanser to remove excess oil, then the BHA Pore Serum for blackheads, as described on its product page. Finish with the Light Balance Gel Cream, which is oil-free. All three are under €30 and in stock. Since your skin is a bit sensitive, do a patch test with the serum first.

(Three product cards appear under the answer, each with the live price, a stock badge and a link to the product page.)

Shopper: I also have some painful spots on my chin that won't go away.

Assistant: For persistent or painful breakouts, it's best to see a dermatologist or pharmacist, who can look at your skin properly. The routine above is meant for everyday oiliness and blackheads, not as a treatment.

Behind that answer, a few things happened. The assistant asked one question instead of guessing. The budget became a hard filter. The serum's usage details came from the shop's own product description, not from the model's memory. And the painful spots were handled as a health question, not a sales opportunity.

Stock, prices and only the products you sell

A skincare recommendation is only useful if the shopper can buy it. AskMerra recommends products from your catalog only. Product ids in the AI's answer are checked against the products it was given, and anything else is dropped. Prices, sale prices and stock on the product cards are read live from your catalog, never from the model.

Out-of-stock products are never recommended. Cached answers to common questions are skipped if any product they mention has gone out of stock. Price and stock changes from your feed or the Push API apply immediately, so a sale price shows up in the chat as soon as it reaches your catalog.

Careful by design: no medical claims

Beauty sits close to health, and that calls for restraint. AskMerra gives careful, non-diagnostic answers to health questions and recommends consulting a professional, for example for pregnancy or allergies. Ask it "Is the retinol serum safe during pregnancy?" and it will point the shopper to a doctor or pharmacist rather than give a verdict. Persistent or severe skin problems get a suggestion to see a dermatologist.

The same care applies to your product data. Enrichment is instructed not to invent facts or make medical claims, and the assistant does not promise that a product cures or eliminates anything. That protects your shoppers, and it protects your brand from claims you never made.

Setting up a beauty catalog for good recommendations

The quality of the answers follows the quality of your data. Before you go live:

• Write useful descriptions. Mention skin types, the concerns a product addresses, key ingredients and how to use it.
• Map the attributes you already have, such as skin type or vegan, during import.
• Keep stock flags accurate, because the assistant never recommends unavailable products.
• Check the enrichment of your best sellers and correct tags where needed.
• Add your policies for shipping and returns, plus FAQ entries such as "Are your products cruelty-free?"
• Set suggested questions in the widget, up to six per language.
• Test in the Playground with the questions your customers really ask.

For more on how AskMerra keeps answers tied to your data, read accurate AI answers grounded in your catalog.

Try it on your own catalog

Create your account, import your products, ask "I have dry, sensitive skin, what should I use in the evening?" in the Playground, and see what your shoppers would get. Prefer a guided tour? Book a 20-minute demo through our contact page and we will load a sample of your products. The catalog docs explain the import and enrichment in detail.