Online cosmetics store: brands, composition, shades

Let's see how the buyer is looking. The shade is chosen inside the card - we make options, one address, canonical on it. A specific shade is searched for by a separate request, and it has its article number — a separate card, but then a separate title and a separate text must be written under it. How the opposite decision without rules ends is clearly visible in the measurement of a retail underwear store: there each size and color has a separate address, the catalog has grown to 115,456 items, and out of a sample of 150 addresses on the site map, 50 have been redirected to another item - another size or another color. In cosmetics, the shade replaces the size, the mechanism is the same. Our recommendation: variations inside the card by default, separate pages - only for requests that you see in your own statistics.

Separate pages that are located in the site map and are linked from the product cards. In the wholesale beauty catalog from our perspective, the canonical card holds 3,947 products, 166 brands, 47 categories and 6 information pages — a total of 4,167 addresses. The brand there works as a second full catalog entry, not a list of logos in the footer. Fair margin: the presence of a page in the sitemap does not mean either indexing or traffic. Google writes it directly, and a brand page without its own text and without demand may well be indexed and not receive a single visit.

Speed ​​depends on the setting, not the number of cards. In a 5,561-item professional cosmetics store, the main markup weighs 153.8KB — that's easy — and the first server response is 801.2ms. Nearby, in a natural cosmetics store, the page is heavier (196.3 KB), and the response is 126.3 ms. Slows down the server part and requests to the database, not the layout, and it is treated without redesign. What speed doesn't do: It doesn't make the offer more attractive. If the price is higher than the market or the item is not available, a fast site will simply give the buyer a reason to leave faster.

Filtering by skin type, purpose, and volume generates a combinatorial number of addresses: each pair of ticks is technically a separate page. Therefore, the work includes a list of combinations that we open consciously - usually it is "brand + category" and "category + destination" - and the rest is closed by a canonical for the category and a rule in robots. Without it, you'll end up with thousands of nearly identical pages in the index, and that's a degradation, not an improvement. Separately: on a catalog with up to several hundred items, complex filtering gives little, there a clear category tree works better.

Yes, through groups of buyers: after entering, the salon sees its prices and multiples, the retail buyer - an ordinary shop window. The border here is not technical. An account on the website does not make you a supplier for a legal entity: the purchase is also based on a contract, payment deferment, availability of a personal manager according to the volume - the website closes only the first step. If your wholesale is the main volume, the frame should be built around it, and not completed on the side; then see the wholesale shop page.

This is a commercial setting and should be fixed before launch, not after the first unclaimed parcels. In the measured store of professional cosmetics, post-payment is allowed only with a deposit, and free delivery begins with a certain amount of the order. We display such rules both in the checkout and on the payment page, so that the manager does not have to repeat them in every call. There is also a typical hole: in the same store, not a single carrier is named on the delivery page - neither Nova Poshta, nor Ukrposhta. At this stage, the buyer simply does not understand how the parcel will arrive.

28–48 working days from the approval of the TK. Layout: Price and TOR audit 3-5 days, catalog structure 4-6, design 5-9, assembly 8-13, import with variation layout 5-9, integration with testing and transfer 3-6. The longest stage is not assembly, but import. If the shades in the price list are in one text field, such as "Tone 03 beige, 50 ml", they have to be sorted into attributes semi-automatically, and it takes weeks for several thousand items. That is why the work begins with the analysis of the price, and not with layouts.

Because traffic in cosmetics is almost always paid, and without cart and purchase events, there is nothing to optimize the campaign. Of the three measured beauty stores, one has no meter at all, the other has Universal Analytics, which has not collected data since July 2023, and GA4 is not in the code. What the meter doesn't do: It doesn't improve anything by itself, it just gives you an opportunity to see. Plus blockers, consent mode, and browser restrictions on cookies — GA4 shows a trend, not an accounting truth, and you need to consciously compare its numbers with the cash register.