Advertising on platforms, card SEO, bid auctions and promotions: mechanics and the metrics of ad cost share, ROAS and ROMI
How to promote goods on Wildberries, Ozon, Yandex Market, Lamoda and other platforms: how the search optimization of a card works, how the bid auction and payment per impression, click or order are built, how platform promotions differ from seller discounts and how to calculate ad cost share, ROAS and ROMI after returns. It decides how much to pay for impressions and when to stop advertising before it eats the contribution.
Why this document
Advertising on a platform looks like a handy growth lever: set a bid and get impressions. But by itself it does not create profit: it buys a place in the results, and a sale depends on price, reviews, size and buyout. For fashion the most hidden risk is that the buyer clicks, orders and returns while the advertising is already paid. The marketplace manager, the marketer and the finance person read this document. It explains the auction mechanics in plain words, shows how bid, conversion and ad cost share are tied and gives a threshold of allowed ad cost share above which a sale is loss-making. The names of advertising modes differ by platform: check them with the account.
At a glance
- What is this document and why is it needed?
- How to promote goods on Wildberries, Ozon, Yandex Market, Lamoda and other platforms: how the search optimization of a card works, how the bid auction and payment per impression, click or order are built, how platform promotions differ from seller discounts and how to calculate ad cost share, ROAS and ROMI after returns. It decides how much to pay for impressions and when to stop advertising before it eats the contribution.
- Who owns the document and how often is it rebuilt?
- The marketplace manager or marketer with the finance person; bids are revised daily or weekly, the budget monthly; the advertising report is part of the weekly channel report.
- What mistakes are the most common?
- The ad cost share is counted from paid orders and not from kept ones → the measure is understated → count it from revenue net of returns. Joining a promotion for impressions → the promotion price kills the contribution → calculate contribution at the promotion price. Paying for impressions of a weak card → a high cost per click and low conversion → the card and size table first, then bids.
For whom
- Brand
- The brand uses advertising to launch novelties and support hits and aims to grow the share of organic through reviews and brand search; it matters not to devalue the price with promotions.
- Retail
- The retailer advertises remainders and seasonal items across many items; it needs rules for the automatic bid and for switching advertising off by stock and contribution.
What it consists of
Card SEO: how a product enters the results
- Title and category: the name is built from item type, cut, material and color in the order of the query; an exact category matters more than a clever name because filters and commission depend on it
- Keywords: words are taken from real buyer queries (cut, length, size, season, occasion); add them to the name, description and characteristics without needless repetition
- Characteristics: fill in all fields because filters by color, size, material and season determine which results you take part in
- Behavioral measures: platforms consider click share, adds to cart, buyout and rating; a strong card gets more organic and makes advertising cheaper
- Size query: fashion buyers search by size and fit; a size table and model parameters in the description work both for search and for returns
Bid auction and payment modes
- Auction: the ad position depends on your bid and on the card score (clicks, conversion, relevance); a poor card has to pay more for the same position
- Payment per impression, click or order: you pay for impressions (CPM), clicks (CPC) or a result (order, sale); with payment for result the risk is lower, but the platform may take more
- Automatic campaigns: the platform picks queries and places itself; convenient for a start and tests but less predictable in spend
- Manual campaigns: you choose queries and bids; needed for hits with known conversion and to protect brand search
- Budget and limits: daily and total limits protect from a spike in spend; set the limit from the allowed ad cost share and not from the wish to collect impressions
Promotions, reviews and metrics
- Platform promotions: the platform invites to sales and auto-promotions; the promotion price reduces contribution and participation may be a condition of impressions; decide by the contribution calculation and not by the invitation
- Seller discount and co-funding: the discount is set by the seller and a part of promotions is co-funded by the platform; the report shows separately who pays
- Reviews and rating: rating affects position and conversion; size reviews are read as a guide; paid and faked reviews violate the rules and the law on advertising and competition
- Ad cost share: advertising spend ÷ revenue; the allowed ad cost share equals contribution before advertising ÷ revenue; anything higher makes a sale loss-making
- ROAS and ROMI: ROAS is revenue ÷ advertising spend, ROMI is (contribution before advertising − advertising) ÷ advertising; count after returns or the measures are overstated
Differences for the brand and the retailer
- Brand: Brand search: the brand protects queries with its own name because they bring the warmest buyers and are cheap by bid.
- Brand: Novelties: the brand pays for impressions of a new model while there are no reviews and counts payback by kept orders.
- Brand: Price and promotions: the brand sets a lower bound for promotion prices so that platform retail does not devalue wholesale.
- Brand: Feedback: the brand uses reviews and questions as a source of changes in the product and size grid.
- Retailer: Many items: the retailer sets rules by groups: hits, tail, remainders; the bid depends on turnover and stock.
- Retailer: Stock: advertising is switched off when the stock of a size or a run falls below a threshold, otherwise you pay for impressions with no sales.
- Retailer: Other brands: a brand may restrict advertising and promotions of its goods; observe the brand price rules.
- Retailer: Common budget: the retailer distributes the advertising budget between platforms and its own site by contribution and not equally.
- Brand: Organic: the brand measures the share of organic orders and lowers the paid share as reviews grow.
- Retailer: Clearance: the retailer spreads advertising of seasonal remainders over time so as not to pay at the end of the season for impressions of goods without sizes.
Where the data comes from
| Data source | Fields |
|---|---|
| The platform ad account | impressions, clicks, orders, spend, bids, positions |
| The platform report and brand records | revenue net of returns, contribution before advertising, buyout |
| Search queries and card positions | query frequency, position, CTR |
How to build it step by step
- Step 1: Calculate the allowed ad cost share for every item: contribution before advertising ÷ revenue net of returns; this is the ceiling of the bid.
- Step 2: Put the card in order: category, name with keys, all characteristics, a size table, photos; only after that pay for impressions.
- Brand: Step 3: Launch protection of the brand query and a test of the new model with a small budget; measure clicks, orders and buyout.
- Retailer: Step 3: Split items into groups (hits, tail, remainders) and set bid and switch-off rules by stock and contribution.
- Step 4: Set the bid at which the ad cost share does not exceed the allowed one, and budget limits per day and week.
- Step 5: After two weeks review the campaigns: impressions, click, conversion to order, buyout, ad cost share and contribution; keep what brings contribution, stop the rest.
- Brand: Step 6: Set the lower price bound for promotions and check contribution at the promotion price; items with positive contribution go into promotions.
- Retailer: Step 6: Switch on automatic switch-off of advertising at low stock of sizes and recalculation of bids after a price change.
- Step 7: Once a month calculate the total ad cost share on total revenue (the share of advertising in the whole turnover, including organic) and review the budget by contribution.
Formulas
| Measure | Formulas | Note |
|---|---|---|
| Ad cost share from bid and conversion | Ad cost share = CPC ÷ (CR × P × (1 − returns)), where CR is the conversion of click to order | Growth of returns raises the ad cost share at the same bid because revenue per click falls. |
| Allowed ad cost share | Ad_share_max = contribution_before_advertising ÷ revenue_net_of_returns | Above this value every kept unit is loss-making. |
| ROMI of advertising | ROMI = (contribution_before_advertising − advertising) ÷ advertising | Take contribution before advertising after returns and all platform deductions. |
Formulas in notation
- Ad cost share
δ = c / (u · P · k)
c cost per click, u click-to-order conversion, P price, k kept share. Example: 40 / (0.04 × 4,000 × 0.75) = 33.3%.
Worked example
An illustrative link of conversion and ad cost share at a cost per click of RUB 40 (price RUB 4,000, returns 25%; the cost per click and conversions are illustrative)
| Click-to-order conversion, % | Revenue per click, RUB | Ad cost share, % | Pays back |
|---|---|---|---|
| 1 | 30.00 | 133.3 | no |
| 2 | 60.00 | 66.7 | no |
| 3 | 90.00 | 44.4 | no |
| 4 | 120.00 | 33.3 | yes |
| 5 | 150.00 | 26.7 | yes |
The allowed ad cost share in the model (contribution before advertising ÷ revenue) is calculated by code and is about 36.1%; at a 1-3% conversion the ad cost share exceeds the allowed one and a sale is loss-making; at 4-5% advertising pays back. Figures are illustrative.
Figures
Data as a table
| Ad cost share, % | Profit per ordered |
|---|---|
| 0 | 1,083 |
| 8 | 843 |
| 16 | 603 |
| 24 | 363 |
| 32 | 123 |
| 40 | −117 |
| 48 | −357 |
Profit per ordered unit and ad cost share. Profit falls as the ad cost share grows and crosses zero at the allowed ad cost share.
Data as a table
| Measure | Count |
|---|---|
| Impressions | 10k |
| Clicks | 200 |
| Orders | 8 |
| Kept | 6 |
An advertising funnel per 10,000 impressions (illustrative). Every step of the funnel loses part of the buyers; returns remove part of the paid orders.
Specifics by category
- Apparel
- Apparel: the buyer searches by cut and size; seasonal peaks raise bids; put returns into the ad cost share and switch advertising off for empty sizes.
- Footwear
- Footwear: a high cost per click because of competition, but also a high average receipt; size returns are expensive, so the ad cost share is counted after them.
- Accessories
- Accessories: a low price does not bear a high bid; rely on organic, reviews and a set of several items.
- Kids
- Kids: peaks before school and holidays; parents read reviews; advertising works on narrow queries by age and height.
- Men
- Men: stable demand and a lower acquisition price; basics repay advertising faster.
- Women
- Women: the most competitive and seasonal segment; advertising needs strong photos and size information, otherwise there are many clicks and little buyout.
Common mistakes
- The ad cost share is counted from paid orders and not from kept ones → the measure is understated → count it from revenue net of returns.
- Joining a promotion for impressions → the promotion price kills the contribution → calculate contribution at the promotion price.
- Paying for impressions of a weak card → a high cost per click and low conversion → the card and size table first, then bids.
- Advertising an item without sizes → impressions without sales → switch-off by stock.
- Faking reviews → card blocking and liability for unfair competition → collect reviews honestly and work with negatives.
Russian specifics
- Advertising on external platforms (bloggers, social media) is subject to the advertising law of 13.03.2006 No. 38-FZ and the marking requirements: an identifier and the transfer of information to an operator; the platform internal advertising follows its own rules. Rule: Law 38-FZ of 13.03.2006: Federal Law of 13.03.2006 No. 38-FZ on advertising.
- Faked reviews and unfair competition: the competition protection law of 26.07.2006 No. 135-FZ; check the platform rules and the law. Rule: Law 135-FZ of 26.07.2006: Federal Law of 26.07.2006 No. 135-FZ on protection of competition.
- Brand: The platform rules on a ban on links to external resources, contacts in the card and comparison with competitors must be read before preparing content.
- Retailer: Before advertising another brand check its rules on prices and channels; a violation may end in an owner complaint.