Every ad platform shows its own cost-of-sale ratio or ROAS, and each one uses the revenue it has credited to itself. Add them up and the total often exceeds what the store actually sold in the same period. That is not a fault in one account; it is two different questions. The store’s blended cost of sale tells you whether you can afford your marketing. Platform figures hint at where to look for improvements inside each channel. Deciding the budget on just one of them means seeing only half the picture.
Two numbers, two questions
Platform cost of sale is the ratio of spend to revenue as a particular ad system sees it. Czech e-commerce calls it PNO (podíl nákladů na obratu). Sklik defines it as the cost of all clicks divided by revenue from goods sold, that is, the total value of the conversions it measured. ROAS in Google Ads is the same relationship inverted: conversion value divided by ad spend. A 12.5% cost of sale equals a ROAS of 800%. Both rest on revenue the platform has credited to itself.
Blended cost of sale – its inverse is often called blended ROAS or MER – sets all marketing costs across all channels against all store revenue for the same period. Take revenue from the system where orders are created. Shoptet can show turnover with or without VAT and filter by order status, so cancelled orders can be excluded. A VAT-registered store usually reads costs without tax, so calculate revenue without VAT too. At the Czech standard rate of 21%, a blended ratio calculated on gross turnover would come out roughly a sixth lower than it really is.
| The question it answers | What it cannot tell you | |
|---|---|---|
| Blended cost of sale | Can the store sustain its current marketing spend? | Which channel or campaign drives the result |
| Platform cost of sale / ROAS | Which campaign, ad group or ad performs better in this channel? | How much the store would sell without advertising |
Why credited revenue does not add up
Each platform credits only its own ads, by its own rules. By default, Google Ads reports give credit only to Google’s paid channels, whereas Google Analytics credits both paid and organic channels. Meta can count a purchase up to 7 days after a click on an ad, but also up to one day after a mere view. Sklik recognises a visitor for up to 30 days after a visit from its ads.
A customer who saw an Instagram ad during the week, clicked a Sklik ad and finally bought via Google can therefore appear in all three accounts. Each platform counts correctly by its own rules. The error only appears when you add their figures together.
An illustrative month with invented numbers:
| Spend | Credited revenue | Platform cost of sale | |
|---|---|---|---|
| Google Ads | CZK 90,000 | CZK 700,000 | 12.9% |
| Meta | CZK 60,000 | CZK 450,000 | 13.3% |
| Sklik | CZK 30,000 | CZK 250,000 | 12.0% |
| Sum of platforms | CZK 180,000 | CZK 1,400,000 | 12.9% |
| Whole store, excl. VAT | CZK 180,000 | CZK 1,200,000 | 15.0% |
Together the platforms credited themselves with CZK 200,000 more than the store sold in total – including orders from organic search, email and loyal customers. The figure that matters is the blended 15%, not 12.9%.
Credited revenue is not caused revenue
Even with duplicates removed, a second problem remains. A platform credits purchases to people who came into contact with an ad, but does not establish whether they would have bought anyway. In an eBay field experiment, brand keyword ads had no measurable short-term benefit, because customers searching for the brand arrived without the paid link. An analysis of 663 experiments at Facebook found that even advanced statistical methods applied to commonly available data could not reliably replace a randomised test. Both studies come from large US advertisers, so the size of the error cannot be transferred to a Czech store. The direction holds more generally: brand and remarketing campaigns often collect purchases that would have happened anyway, and that is why they look best in platform reports.
That does not make platform figures worthless. Within one channel the bias affects campaigns in a similar way, so comparing two ad groups or ads in the same account is usually more useful than the absolute number. Platform figures are weakest when you compare channels with each other, because each measures differently.
Revenue, margin and advertising in one place
Korzaro connects your store’s revenue with spend from your ad accounts, so you can see the blended cost of sale and platform figures for the same period side by side.
The blended cost of sale as the business constraint
The blended figure only makes sense next to gross margin. After paying for goods, the store keeps a contribution from which it must pay for marketing, other variable costs and, finally, fixed costs. If the store in the illustration has a 30% gross margin, CZK 360,000 remains after goods. Marketing at CZK 180,000 takes half of it. A blended cost of sale above gross margin means marketing consumed more than was left after paying for the goods.
The blended figure has two weaknesses worth knowing.
It does not separate advertising from the rest of the business. When demand rises in December or loyal customers return, the blended ratio improves even if advertising works no better. Compare like periods, ideally year on year.
It is an average, not the price of the next koruna. Advertising has diminishing returns: the first part of a budget captures the cheapest demand, further spending increasingly expensive sales. Marketing mix modelling guidance therefore recommends judging the past by average return, but deciding further budget by the return on the next unit spent. Illustratively: if the store spent CZK 150,000 and sold CZK 1,120,000 one month, then spent CZK 180,000 and sold CZK 1,200,000 the next, the last CZK 80,000 of revenue cost CZK 30,000 – 37.5%. That is more than the 30% margin, even though the 15% average looks comfortable. A comparison of two months is not proof, though: seasonality or a promotion could explain the difference.
Reading both numbers together
| What you see | Likely explanation | What to do |
|---|---|---|
| Platforms improve, blended ratio is flat or worse | Platforms overlap more in their credit, or campaigns collect purchases that would have happened anyway | Check the share of brand and remarketing campaigns; do not add budget on platform figures alone |
| Blended ratio worsens, platforms are stable | Fewer orders outside advertising (season, loyal customers, organic search), or spend is rising without extra sales | Separate the trend in non-advertising revenue; compare year on year |
| After cutting a channel’s budget, total revenue barely changed | Part of the spend probably was not bringing new sales | Keep the cut and watch for several weeks |
| Both improve and gross margin holds | Healthy position | Add budget gradually and check that total margin after marketing grows too |
When to run a test
Only a test with a control group that does not see the ads can tell you how many sales advertising actually brought. Google Ads offers Conversion Lift for this, but it is not available to every account. For its own test, Meta indicates as a guide a campaign started within the past year with at least USD 5,000 of spend and 500 optimised conversions. A smaller Czech store will often fall short of these thresholds.
A simpler option is a controlled change: for a set number of weeks, sharply reduce or pause one part of the budget – brand campaigns, for example – and watch the store’s total revenue, not the platform’s figures. Treat the result as indicative. Research across 25 large advertising experiments showed that individual customers’ sales are so volatile that even large tests often produce very imprecise estimates of return. A small difference in revenue may therefore mean nothing. A large, repeated difference is a signal you can work with.
What to do now
- Calculate the blended cost of sale for the last full month. All marketing costs (ads, agency fees, comparison sites) against store revenue excluding VAT and cancelled orders.
- Set it next to gross margin. The gap between them is what remains for other costs and profit.
- Alongside it, record each platform’s cost of sale and the sum of their credited revenue. If the sum exceeds store revenue, you know how large the overlap is.
- Decide the total budget on the blended ratio and margin, and shifts within a channel on platform figures.
- For the largest disputed item, plan a controlled change with a fixed end date and watch the store’s total revenue.