Micro-Differences, Macro-Results: What Marketplace Algorithms Hide and Why Shoppers Never See Your Product

Which product are you more likely to pick — the one at eye level, or the one lying down by your feet?

The one that catches your eye first, of course. The highest-converting, highest-priority shelves go either to the products that bring in the most revenue or to the brands that paid for shelf space at "eye level".

Visibility is the share of shoppers who saw your card out of everyone who searched for the product using a given keyword.

Put simply: users typed the keyword "apple" 100 times. In the results for "apple" you sit in position 60, 85 shoppers clicked through to another product before reaching you, and only 15 users scrolled far enough to see yours.

Congratulations, you are the owner of 15% "visibility".

The goal for any product is to reach 100% visibility for a keyword, so that every shopper searching on that keyword sees your card.

The more shoppers see your product, the more often it gets bought. A shawarma stand at a station in central Moscow will get more orders than an outlet in a village of 300 people.

Behavioral signals are any shopper actions that marketplace algorithms read as quality signals for a card: purchases, conversions, adding to Favorites, carts, questions, and time spent on the product page.

This data is used to forecast revenue for the product — it gets analyzed and compared against market benchmarks. Who is better? More in demand? Who will bring in more money? That is who gets shown more often: their "visibility" is higher.

Which factors affect your "visibility"?

• Internal advertising — want to be shown? Pay up! If you have a unique product in a niche nobody else is in, you can sit pretty on the "shelf" for free.
• The card's current revenue. The best predictor of future behavior is past behavior. Big sales in the past mean high sales in the future. The algorithms remember that.
• Reviews for a batch of goods. Rating dropped? Lots of defects? Bad reviews? The marketplace would rather not show that batch at all.
• Repeat purchases — people liked the product so much that they are willing to buy it again and again.
• Conversions: to click (CTR), to cart, to order, to buyout — multiply them together and you get the chance that a person buys the product.
• And others: brand likes, favorites, time on the card, defect rate, reviews-to-purchases ratio, product questions.

Thanks to neural networks and machine learning, algorithms evaluate hundreds of factors at once and use them to decide where and how often to show your card.

A marketplace has two main goals:

• Show the best product at the best price — so the shopper pays as much as possible but still buys.
• Not lose the customer — so they do not buy somewhere else.

Because of this, the whole position calculation system rests on two parameters:

• How relevant the product is to the shopper's search query.
• How much revenue it can bring the marketplace.

What affects a product's "visibility" the most

1. 1Sales volume for the keyword — the single most important factor.
2. 2The size of your ad bid — the higher it is, the better the chance you get shown in the right spot.
3. 3The sum of micro-differences in behavioral signals. One product has a 2.5% CTR, another 2.75%; add-to-cart is 10% for one and 11% for the other. A human cannot see the difference, but for the algorithm it adds up with everything else into a sizeable gap in the ranking.

Potential is determined by the marketplace's forecast: will they buy or not. A customer who did not buy is a bad customer. And the marketplace likes the good ones.

Even if you crank your ad bid to the maximum, you will not get 100% "visibility" — the algorithms will cap the number of impressions and the placement.

Want to stand higher? Show good sales. No good sales? Stay where you were and do not get in the way of the big players selling.

Your behavioral signals for a keyword depend directly on your "visibility", and your "visibility" depends directly on the number of sales for that keyword.

That means products in distant positions for their target keywords cannot climb unless the sellers with higher keyword sales make mistakes: keeping stock topped up on time, staying in internal advertising constantly and remaining in shoppers' field of view all the time.

Or else it will be very expensive and very slow.

1. 1Raise CTR for the target keyword — so more people click through to the card from the general results.
2. 2Raise the intermediate conversions (click to cart to order to buyout) — so more of the people who already clicked through end up buying.
3. 3Simulate purchases on the marketplace (buyouts / cashback giveaways / barter) on the keyword — so that a person makes a purchase from their own account using the right query, not via the article number or a direct link.

Your product is NOT unique. Before buying, the shopper will see at least 50 other products. You are competing for the attention of a live human being who has 3 seconds to decide: click your card or scroll past.

If you did not give the shopper what they were looking for within a couple of seconds, you lost them.

How to fix the CTR of the first slide

Collect the full list of keywords shoppers use to search for your product. Include slang and the phrasings people actually type into search, not the "correct" ones.

The more often users type a specific query, the more important it is to reflect it in the creative.

• The user typed SPF 50 or another keyword? Reflect it on the creative: SPF 50, the protection level, whether the filter is physical or chemical, "protection from aging and UVA/UVB rays", safe during pregnancy.
• Look at the card through the eyes of someone seeing your product for the first time. Ask yourself: "Would I have understood what this product is within 3 seconds?" If not, redo it.
• Test the cover image, checking conversions by keyword.

Use as few words as possible on the title slide. Do not overload the cover. Only the words that actually lift conversions.

Right now there are 4 main ways to simulate a purchase:

• Bloggers — you ask them to buy the product using a keyword, and along the way they shoot content.
• Cashback giveaways — people buy the product at the current price using a keyword and leave a review, and you refund part of the money (cashback).
• Buyout services / contractors — purchases are made from disposable accounts shared across all sellers.
• Purchases made by aged buyer accounts — these people do not exist offline, but they leave a virtual trail online: they read Dzen, log into VK, watch YouTube, make purchases.

Do not wait for the algorithm to work out on its own that your product is special. Help it.

Not every buyout is read by marketplaces as organic

The algorithms have gotten smarter and have long been able to tell real purchases from artificial ones. The technical details are critical — they decide whether the buyout counts as a sale on the keyword or whether you simply waste money.

What marketplaces look at:

• Payment method.
• Buyer data: IP, device, WIN, fingerprint, cookies from the marketplace and other sites.
• Purchase geography.
• Account age.
• The chosen pickup point (PVZ) and any change of pickup point.
• The buyer account's type and history.
• How frequently reviews are left.
• How many orders the account has placed before and how "alive" it is.
• Suspicious purchase patterns.

• Search for and buy the product only through search, and where possible without filters (for growth on search queries).
• If the goal is to grow on a specific keyword, make the purchases using that specific keyword.
• Start from your current ad bid and position in order to calculate the required number of buyouts correctly.
• Search and buy through the category if the goal is to grow within a category.
• Control every action of bloggers, helpers and buyers. Require a screencast of the product being ordered via the target query.
• Use services / contractors that can prove the purchase worked on the keyword, provide an ID or a screenshot of the review, and measure positions.
• Calculate the effective cost of a review a month after it is posted, not the price you paid for the buyout plus the review.

MYTH #1: external advertising / purchases by SKU / direct links raise keyword visibility

External advertising drives clicks and purchases, but the algorithm does not tie them to a search query and does not credit the sale to a specific keyword. Is it useless? No, you make money. But for promotion on a specific keyword it does not help.

MYTH #2: you can use any filters when ordering

Filters add specific words to the original keyword, which changes the search results and credits the purchase to different keywords.

• A brand filter (say your brand is called LABUBO). Before: "moisturizing serum" — after: "moisturizing serum LABUBO". Positions on the main keyword will not budge.
• A color filter. Before: "polka dot dress" — after: "red polka dot dress". The growth will happen on a different query.

Price filters, for example, can be used. Carefully.

MYTH #3: buyers in cashback giveaways and bloggers follow the brief and find the product from page 15

If your friend buys the product following the instructions, it works. If it is a blogger or a freebie hunter too lazy to hunt for the product on page 30, they will simply tap the link, order by article number — and you get no keyword growth.

1. 1You have to break the cycle of no sales because of no visibility. As long as nobody sees you, nobody buys; as long as nobody buys, nobody sees you.
2. 2The only way to break it is a combined approach: a well-crafted creative with high CTR, flawless content inside the card, and carefully planned simulated buyouts with aged buyer accounts.
3. 3The algorithms are your allies if you speak to them in the language of behavioral signals. Do not fight them — work with them.
4. 4Promote products through behavioral signals systematically, relying on analytics rather than guesswork.

Do not count on a miracle from advertising, promotions and price cuts. Work with behavioral signals systematically and consistently — and you will take the shelf at shoppers' "eye level" on the marketplaces.