Why people say Ozon ads stopped working
A common claim right now is that advertising on Ozon no longer delivers. In practice the problem is usually setup and measurement, not the channel. Below is the approach we use for advertising and promotion on the marketplace, so you can build campaigns that produce a predictable result.
Update. The separate "push to top" format no longer exists in the Ozon ad cabinet. What remains is pay per click (auto strategy and manual strategy) and pay per order.
Ads affect the marketplace-funded discount
Advertising is not only about attracting customers more cheaply. It also affects the marketplace-funded discount you receive on Ozon - the discount the platform gives on your products at its own expense. You set one price, Ozon shows the buyer another, usually lower one.
The more of this discount you get, the cheaper your product looks, and the more competitive you become. Your products get cheaper for buyers while you spend nothing extra and do not cut your own price. That drives every conversion and the buyer's choice between you and a competitor.
So advertising today is not just about correct settings. It is also about maximising the marketplace-funded discount, which in turn depends heavily on how you run your ads.
What Ozon analytics gives you out of the box
The promotion analytics section has a dedicated area for ad campaigns. You can view data by promotion type, by category, by product category and by each individual campaign. Metrics are also available per product, so you can drill down fairly deep.
Period comparison is there too: pick a period, pick the metrics to display on the chart - sales, promotion, ad cost share (ACOS), cost per click, add to cart, conversion. Everything can be broken down by day or by week.
This is the baseline. For our own stores we use deeper analytics, because the cabinet does not show ACOS against revenue, let alone ACOS against buyouts.
Ads lift organic rankings too
It is well understood that advertising gives you visibility beyond the ad slot. Running ads improves your positions in Ozon's organic search results as well.
How our view of ACOS evolved
At some point we realised that counting plain ACOS was not quite right. It is more correct to count ACOS against revenue. But Ozon analytics does not provide that data, not even close.
As we kept working with the platform and gaining experience, we concluded that ACOS against revenue is still not quite right either. So we started calculating ACOS against buyouts. The reason: whenever you have orders, you always have returns, cancellations and so on.
If you count plain ACOS - spend this much on ads, earn this much, this many sales - that is ACOS against revenue. It is correct, but it ignores returns and cancellations, which very few sellers account for. That is why we analyse ACOS against buyouts as well.
Net of buyouts and cancellations you get a clean figure for promotion spend, and that figure gives a far more accurate picture.
Why third-party analytics are necessary
Ozon's cabinet does not offer anything close to this, so for promotion we rely heavily on third-party analytics services. The in-cabinet analytics are basic and only suitable for a quick, surface-level check. Deeper analysis needs a dedicated service.
Pay per order: two variants
The second format is pay per order. It splits into two options.
- Pay per order on selected products: you choose which products join the campaign.
- Pay per order on all products: the whole assortment is included.
What matters: pay per order on selected products is expensive, with a high ad percentage - 23%, or 10% with a discount. In the all-products format you can promote at 5%.
The downside of the all-products format is that it only runs across the entire assortment. With selected products you tick specific items and add them to the campaign. In the all-products format there is only a launch button, and ads immediately run on everything you sell. The only thing you can change is the rate: 5%, 7% or 9%.
When pay per order makes sense
In some categories 5% is already enough to get a volume of traffic and a volume of sales. That is why, when choosing pay per order, I recommend launching it in a number of cases - especially when the store is new and you are only building momentum. Pay per order often works as a basic, foundational tool.
Still, I recommend launching on all products and then experimenting with the rate: 5%, 7% or 9%.
The overall strategy for pay per order on all products is to test every rate in sequence. Start at 5%, run it for about a week, and look at sales, sales volume, customer volume and the money that came in from that campaign over the week.
Sequential bid testing
If traffic is not enough and you want more, move to 7% and watch how it performs and how big the difference from 5% is. Sellers often find 5% does not bring enough traffic while 7% does. They keep 7%, and it works as the base promotion tool.
Third step: if after the second week of testing 7% still brings little traffic, launch 9%. At 9%, most categories already get some volume of traffic.
That is why the strategy is usually built on testing all rates in sequence and finding the one that gives you the best cost per acquisition while keeping the customer volume you need.
Pay per click: campaign setup and product selection
Now the second format, pay per click. Here everything is simple: there is a "Create campaign" button. At the start you enter the campaign name, then select the products you want to run.
You can run a separate campaign per product, which is what I recommend. In our tests, campaigns grouped by shared characteristics performed best.
Say you sell mats - promoting them in one campaign is acceptable. But those mats should be identical. They can be identical in different colours - black, grey, white - and still be promoted in one campaign. But the most optimal and effective approach is a separate campaign for each product.
Budgeting and campaign efficiency
In all our tests, such campaigns get a much lower ACOS. Overspend is possible at the start: when you launch, more spend goes out. We set a budget from $33 per product, spend starts, some products may bring no conversions or sales and will leak money.
But you run the tests and learn what works and what does not. Turn off what does not, keep and scale what does. The optimal strategy in pay per click is a separate campaign per product.
On budget, you can set it unlimited or weekly. I recommend limiting it: an unlimited budget can burn tens or hundreds of thousands overnight. We had cases like that when we were still growing the agency - we simply forgot to set a budget, did not tick the limit, and tens of thousands went into promotion. It is not a safe setup, so do not leave the budget unlimited.
A weekly budget can be set at a level you are comfortable with and raised later. There is even a strategy: once you find a product with a good, low cost per acquisition, raise the weekly budget.
Reading campaign results
Suppose you have a campaign with a low cost per acquisition. Look at the products: ad data by SKU, stock, order value, cancellations, returns and ACOS against buyouts.
Here you analyse the results you get. Some products show ACOS against buyouts of 7.8%, some 7.1%, some 17.2%, and so on.
You can then expand and see which campaigns promote that product. You see the campaign for a black belt is active, with its ID. Copy it and go to the Ozon cabinet to check sales, spend and budget.
The weekly budget trick
There is a small trick we often recommend, tied to raising the weekly budget. Look at your ACOS: this store's unit economics allow around 10% including returns, and 7.1% suits us more than well.
So this campaign is worth scaling up. Even with a large budget it will spend, and nothing bad will happen, because the result suits us.
The trick is not to cap the weekly budget. Set not $117 but $330, $500, $830, $1,700.
The larger your weekly budget, the more traffic Ozon gives you, the more positions you get and the better the traffic. We see this often.
When to use the trick
A higher ad budget makes sense, but only on products where ACOS against buyouts suits you more than well. If a product does not deliver an optimal cost per acquisition, do not use this trick. Set a high weekly budget only where you are happy with the ACOS.
Strategy 1: target spend
Back to strategies. On Ozon there are four pay-per-click strategies.
The first is target spend. You set a target spend from 5% to 100% - the ad percentage you built into your unit economics. Say you allowed 10% for advertising, so you enter 10% and you are satisfied. The algorithm will aim to attract customers at that target spend. There may be fluctuations at the start, but over distance the figure levels out and trends toward 10%.
This format makes sense when you want reasonable confidence and a guarantee on cost per acquisition and ACOS.
Strategy 2: push to top and top-4
The second format is push to top. It is not entirely clear-cut. On one hand it is very powerful: it includes a top-4 placement.
Top-4 means your products land in the top 4 of a category. That is almost a guarantee of high conversions, good behavioural metrics on product cards and sales growth.
But top-4 has an important feature: it is a very expensive ad format. Everyone wants top-4, because everyone wants the first positions in their category and more traffic, customers and sales.
So top-4 is worth launching, but you must watch closely that campaign ACOS does not fly into space. Third-party analytics help keep ACOS under control here.
Other top strategies and auto strategy
The remaining strategies - top-12, top-20, top-30 - have little global value, because they do not deliver wow results. Top-12 still gives some sales volume. Top-20 and top-30 are roughly on par with any other format.
You can launch and test them, but only as a test - to check whether your specific case is an exception and they deliver. Overall, top-4 gives great results for almost everyone, genuinely great, but the cost per acquisition can be high. So watch it very carefully.
The auto strategy is simple. You set either search or recommendations, launch, and the system decides at what bid to run and grow your campaign.
What to understand about auto strategy on Ozon: it is a format for the lazy. For those who do not want to dig in or track anything and just want to set it up and get some sales volume. But it will always be more expensive than formats where you work manually. I do not really recommend auto strategy - only if you do not want to understand other formats or test anything and just want something running.
In such cases it is better to launch pay per order. It will already bring sales without sky-high campaign costs. Then you can set an average cost per click that suits you.
This format is also interesting. As a rule we use either average cost per click, push to top and top-4, or target spend. It always comes down to testing which format gives you the lower cost per acquisition.
Comparing formats through testing
There are plenty of formats and plenty of strategies. The approach in pay per click is very similar to pay per order. You test one format for a week, then a second, then a third, compare and keep what worked best. Same principle.
As a rule we launch all formats and see what performed best. In the moment this causes a spike in ACOS.
Even in our own stores you can see periods when ad spend rises abnormally compared with normal times. Normally spend is around $83 a day, and some days bring abnormally high spend. Those days usually coincide with heavy testing. It raises costs, but over distance it lowers ACOS and cost per acquisition.
Record results, then scale
The strategy is sequential testing with mandatory recording of every metric. This includes tracking which campaigns run for each product, which strategies and placements they use, spend, orders and results. The logic is simple: test in sequence, see what worked, then scale it.
Scaling means you have found a campaign with a low cost per acquisition. Then you identify which campaigns give the optimal cost and increase either bids or the weekly budget.
Raising budgets and tracking the effect
As a rule we start with the weekly budget: raise it, raise it, raise it, and watch how much growth it allows. Traffic volume grows, sales volume grows, and that is the scaling. Inside a campaign you can also raise bids on products in pay-per-click campaigns. That also lifts traffic and sales. But keep tracking how bid changes drive sales growth and how they affect final ACOS.
Should you trust competitive bids?
An important point: competitive bids. These are the recommended bids Ozon supposedly considers competitive and suggests you set to get the traffic and customers you need.
What to understand: we have a specific product with a competitive bid of $0.35. Yet at a bid of $0.20, even $0.18, it worked perfectly. At $0.20 it works fine: we get the sales volume we need, order value close to $8,300 net of spend and returns, and low ACOS - at least low enough for these stores. And that is without reaching the recommended values Ozon suggests.
Do not pay much attention to Ozon's recommended competitive bids. Ozon's goal is to earn from us on ad sales. Treat Ozon as a large ad platform, a large ad system whose goal is to sell us more advertising and profit from it.
So always find for yourself the bid that gives your store a lower, more optimal cost per acquisition. Do not look at what Ozon recommends, because your goals and Ozon's are different. Your goal is to earn with minimal investment. Ozon's goal is to earn more from your ad spend.
Summary
Advertising on Ozon still works when setup and analytics are done properly. Test different formats and strategies, track ACOS against buyouts and scale the campaigns that succeed. Do not blindly trust Ozon's bid recommendations - find what is optimal for your store.
