Advertising return and reading market figures
Where a channel budget stops paying back and how to compare market estimates that are counted differently.
About the topic
Advertising is like watering a plant: the first buckets help growth, then extra water just runs into the tray. Returns on ad budgets nearly always diminish: the first RUB 100,000 finds the warmest buyers, the next reaches a colder audience and costs more per person. This topic teaches you to see that effect in your own numbers and answers the question of where to stop: at what budget the last rouble invested stops coming back.
The reasoning goes like this. You have points of budget and revenue from different periods of one channel. Example: at RUB 100,000 revenue is RUB 1.0 million, at RUB 200,000 it is 1.5 million, at RUB 300,000 it is 1.8 million. The first RUB 100,000 brought 1.0 million, the second added 0.5 million, the third only 0.3 million. That is marginal return: what the next rouble adds. To decide whether to pay for it, multiply by gross margin. At 50% margin the second hundred thousand gives 5 x 0.5 = RUB 2.5 per rouble, the third 3 x 0.5 = RUB 1.5, and if the fourth adds 0.15 million, then 1.5 x 0.5 = RUB 0.75 per rouble, and it is not worth paying for. The budget should stop where marginal return after margin approaches one.
It is easy to go wrong here. Average return (all revenue divided by all budget) always looks better than marginal and misleads: 1.8 million on RUB 300,000 is RUB 6 per rouble, while the last rouble brings 3. Decisions are made on the marginal figure. A second trap is attribution. Some of the people who buy after an ad would have bought without it (brand, repeat purchases), so the true contribution is lower than last-click shows.
The second half of the topic is about reading market figures. One report says the market is RUB 800 billion, another 450 billion, and both can be right: one counts retail prices with VAT, the other purchase prices without VAT; one counts orders, the other completed purchases; one the whole market, the other only online. Before quoting a number, find out the unit (money or pieces), the stage of the chain, taxes, period and scope. Bringing estimates to one base gives comparable numbers, not the right one.
Start with your own channel data for 3 to 6 periods and the question of which decision depends on the figure. Enter your own market figures, without borrowed benchmarks.
Key ideas
The limit depends on margin
The stopping point depends on margin, not only on the curve. Steps of RUB 50,000 give revenue of 0.6, 1.0 and RUB 1.2 million: the second step added 0.4 million, the third 0.2 million, which is 8 and RUB 4 per ruble. At a 50% margin the third step returns RUB 2 per ruble and is worth it; at a 25% margin only RUB 1, which is break-even. For a low-margin product, stop the budget earlier.
The break-even bar depends on margin
Advertising pays when a rouble brings more than a rouble of gross profit. At 50% margin marginal revenue must exceed RUB 2 per rouble, at 25% margin above RUB 4. High revenue per rouble with low margin can still be a loss.
Last click credits what it did not cause
If 40% of those who bought after an ad would have come anyway, true return is 40% lower than stated. It is checked by a test: part of the audience does not see the ad and purchases are compared. Without a test any return is provisional.
A market can be measured ten ways
Retail with VAT of RUB 122 billion equals 100 billion without tax, and at purchase prices with a 100% markup it is 50 billion. Units and money, orders and completed purchases, the whole market and one channel give different numbers. Always write the base next to the figure.
Data must be comparable in age
Market figures from different years or sources are comparable only if the base is the same. Growth of 10% in one report and 6% in another may result from different scope, not different processes.
How to work through it
- Write down the channel budget and revenue for 3 to 6 periods; periods should be comparable by season.
- Calculate the revenue gain per budget step: that is marginal return.
- Multiply marginal return by gross margin and compare with one.
- Find the budget where marginal return with margin approaches one: above it spending is wasteful.
- Check attribution: is there a test without ads, or at least a comparison with a control region.
- For a market figure write down the unit, stage of the chain, taxes, period and scope.
- Bring several estimates to one base and only then compare and quote.
Where to start
Marketer: “Response curve” over 3–6 periods of one channel. Anyone reading market reports: “How to read market figures” before quoting a number. Calculations of the topic in order: “Channel response curve: where advertising stops paying back”, “How to read market figures: bringing estimates to one basis”.
Calculations in this topic
Channel response curve: where advertising stops paying back
From several pairs of “budget and channel revenue” fits a saturating curve, shows how close you are to the ceiling, how much the last unit of money brings and at which budget the marginal return equals the cost.
Where a channel budget stops paying back and whether it should change.
How to read market figures: bringing estimates to one basis
Different sources give different market sizes because they count differently: with or without VAT, by orders or by what was bought, not the whole market. The calculation brings your estimates to one basis (excl. VAT, bought-out, the whole market) and shows how much of the spread the method explains.
Why the market estimates differ and what remains once they are brought to one basis.
Self-check questions
- What decision are we making on this figure?
- Can we see where the channel’s return slows down?
- What is the gross margin of the product being advertised?
- What is the marginal return of the last budget step?
- How many purchases would have come without the ad?
- Are the periods comparable by season and promotions?
- How is the market measured: money or units, retail or purchase, with or without VAT?
- Is the scope the same in the compared reports?
Knowledge check
Multiple-choice questions. The correct answer and the explanation follow each question.
How does the marginal return of an ad channel differ from the average, and which one decides whether to raise the budget?
- A. Marginal return is the revenue from the last ruble of budget: it is multiplied by margin and compared with one ruble of cost
- B. Marginal return is revenue on the whole budget; it always falls, so the budget should be cut
- C. Average return is more accurate: it covers all periods, so decisions are made by it
- D. There is no difference: both give the same number if margin is constant
Correct answer: A. Marginal return is the revenue from the last ruble of budget: it is multiplied by margin and compared with one ruble of cost. Average return is the whole channel revenue on the whole budget; because of the first “warm” buyers it almost always looks better. Marginal return shows what the next ruble brings: at a 50% margin and RUB 3 of revenue per ruble that is RUB 1.5, worth paying; at RUB 1.5 of revenue only RUB 0.75, not worth it. The budget is moved until marginal return × margin falls to one. This estimate also does not separate sales that would have happened without the ads.
A channel: at a budget of RUB 400 thousand revenue is RUB 2.6 million, at RUB 500 thousand revenue is RUB 2.95 million. Gross margin is 40%. Do the additional RUB 100 thousand pay back, and how much margin per ruble above the ruble of cost do they bring?
- A. They do not pay back: average return fell from RUB 6.50 to RUB 5.90 per ruble
- B. They pay back: each ruble brings RUB 5.90 of revenue, that is RUB 2.36 of margin
- C. They pay back: RUB 3.5 of revenue per ruble gives RUB 1.40 of margin, that is RUB 0.40 above the ruble
- D. They do not pay back: RUB 3.5 of revenue per ruble is below the average of RUB 5.90
Correct answer: C. They pay back: RUB 3.5 of revenue per ruble gives RUB 1.40 of margin, that is RUB 0.40 above the ruble. The revenue gain is 2.95 − 2.6 = RUB 0.35 million on RUB 100 thousand of budget: RUB 3.5 per ruble, at a 40% margin that is RUB 1.40, that is RUB 0.40 more than the ruble spent. Average return falls (6.50 → 5.90), but that is no reason to stop: decisions are made by marginal return. 5.90 × 0.4 = RUB 2.36 overstates the benefit because the average contains the first, cheapest rubles.
A channel curve R = a × (1 − e^(−b × x)) with a revenue ceiling a = RUB 4 million and b = 0.000002 per ruble of budget. The budget is x = RUB 500 thousand, margin 45%. How much margin above the ruble of cost does the last ruble of budget bring?
- A. RUB 0.63
- B. RUB 2.94
- C. RUB 1.32
- D. RUB 0.32
Correct answer: D. RUB 0.32. Marginal revenue R′(x) = a × b × e^(−b × x) = 4,000,000 × 0.000002 × e^(−1) = 8 × 0.3679 = RUB 2.94 per ruble. Margin 2.94 × 0.45 = RUB 1.32, above the ruble of cost RUB 0.32: the budget can still be raised. 2.94 is revenue, not margin, 1.32 forgets to subtract the ruble itself, 0.63 is the share of the ceiling reached, 1 − e^(−1) = 63%, not a return. The budget where the margin from the last ruble equals the ruble: x = ln(a × b × 0.45) / b = ln(3.6) / 0.000002 ≈ RUB 640 thousand. The curve is fitted to 3–6 points, it is an estimate, not a guarantee.
A source puts the market at RUB 3,660 billion including 22% VAT (check the rate with an accountant), 20% of orders in the estimate are not bought out, and it covers only half of the market (coverage 50%). What value results from reducing it to the basis “excluding VAT, bought out, whole market”?
- A. RUB 2,400 billion
- B. RUB 4,800 billion
- C. RUB 3,000 billion
- D. RUB 5,856 billion
Correct answer: B. RUB 4,800 billion. Reduction: value ÷ (1 + VAT) × (1 − share not bought out) ÷ coverage. 3,660 / 1.22 = 3,000; × 0.8 = 2,400; / 0.5 = RUB 4,800 billion. 2,400 forgets coverage (the estimate relates to half of the market), 3,000 removes only VAT, 5,856 forgets VAT (3,660 × 0.8 / 0.5). The numbers are illustrative, not market data.
Two estimates of one market: RUB 5,500 billion including 22% VAT and RUB 4,000 billion excluding VAT; otherwise they count the same thing. The spread is (maximum − minimum) / minimum. What is it as is and after reduction to one basis?
- A. 37.5% as is and 12.7% after reduction
- B. 37.5% as is and 37.5% after reduction: VAT does not affect the spread
- C. 27.3% as is and 12.7% after reduction
- D. 12.7% as is and 37.5% after reduction
Correct answer: A. 37.5% as is and 12.7% after reduction. As is: (5,500 − 4,000) / 4,000 = 37.5%. After reduction 5,500 / 1.22 = 4,508, the spread is (4,508 − 4,000) / 4,000 = 12.7%: VAT explains about two thirds of the gap (1 − 12.7 / 37.5 = 66%). 27.3% comes from dividing by the maximum instead of the minimum. Equal 37.5% would come from assuming VAT does not affect the estimate; the pair “12.7 and 37.5” is the same numbers reversed, but reduction cannot increase a spread caused by VAT.
Common mistakes
- Looking at average return. Consequence: the budget grows while the last rouble is already losing money. Avoid: calculate marginal return with margin.
- Comparing channels by revenue per rouble with different margins. Consequence: choosing a channel that gives less profit. Avoid: compare gross profit per rouble.
- Trusting last click. Consequence: paying for purchases that would happen anyway. Avoid: run a test with a control group.
- Quoting a market without a base. Consequence: two numbers look contradictory while they measure different things. Avoid: write unit, taxes and scope next to the figure.
- Building a curve from two points or different seasons. Consequence: a random shape. Avoid: use 3 to 6 comparable periods.
Russia and the EAEU
Advertising in Russia is regulated by Federal Law 38-FZ on Advertising: mailings and calls require the recipient’s consent, and online advertising is labeled. Online ad materials must carry a token and be reported to an advertising data operator; this rule has applied since 2023 (Law 260-FZ), and fines are significant. So ad spend should include not only the platform budget but also labeling and reporting work; check details with a lawyer.
Advertising platforms in Russia include Yandex search and contextual ads, VK and Telegram ads, and in-platform marketplace ads (Wildberries, Ozon, Yandex Market). Their tariffs and commissions change, so take current ones from the dashboard, not from articles. Ad spend carries VAT at 22% from platforms registered in Russia; when calculating return, distinguish amounts with and without VAT: if you are a VAT payer, input tax to be credited is not part of your cost.
Market figures in Russia are published by analytics agencies, associations, marketplaces and the statistics service. They differ in scope: retail turnover, online turnover, wholesale supply, units. Before comparing, find out whether revenue was counted at retail with VAT, by orders or by completed purchases, and whether the category or the whole direction is included. Enter your own figures yourself.
The response curve works everywhere. Advertising rules and platforms differ. Guide, please verify: Belarus, Kazakhstan, Armenia and Kyrgyzstan each have their own advertising laws and VAT rates (20%, 16%, 20%, 12%). Russian-style labeling of online advertising in other countries: indicative as of 14.11.2026, check the text of the act; treat it as a Russian feature. In Kazakhstan strong local platforms (the Kaspi marketplace) run advertising under their own rules. Take market data for each country from your own sources.
Categories and audiences
- Apparel
- For apparel: wide choice and seasonality, so the response curve shifts from month to month. Compare comparable seasons. Advertising works well on new arrivals and the season start, while on remaining stock returns fall faster than on the whole range.
- Footwear
- If it is footwear: purchases are more deliberate and rarer, higher check, many size returns. Revenue from ads is larger than gross profit after returns, so count return after returns, not by orders.
- Accessories
- On accessories: low check and high margin, quick decisions, gift demand before holidays. The curve saturates early, so split the budget by occasion and channel. For the market, check whether the category is included in the report.
- Kids
- In children’s wear: the buyer is the parent, demand is tied to school, birthdays and seasons. The peak is at the end of summer, so August ads pay back better than July. Sizes change fast: repeat purchase is higher and the customer base more valuable.
- Men
- Men respond to a specific offer and price comparison, with fewer impulse purchases. Search and marketplaces work better than emotional advertising. Watch the share of repeat purchases and the gift peak before 23 February.
- Women
- In the women’s line part of demand is gifts for 8 March, birthdays and New Year, and they are often bought by someone other than the wearer. In such weeks revenue grows because of the occasion, not the advertising: if you put them into the budget-and-revenue curve, advertising gets credit it did not earn. Mark holiday weeks separately and build the curve from ordinary ones.
Questions this topic answers
- Where does a channel budget stop paying back?
- Why do market estimates differ between sources?
- Which market estimate can be quoted and how?
- Why do market estimates differ between sources and to what basis should they be brought?
Key measures and terms
Study first
What next
Related: “Marketing budget and testing impact”, “Platforms and marketplaces”, “Customers and loyalty”.
Documents on this topic
- Diffusion lines, private label, multibrand balance, collaborations, capsules and drops
- Production and content: shoot budget and calendar
- PR, seeding and influencer marketing: budget and evaluation
- Marketing channels in Russia: Telegram, VK, MAX, Dzen, Rutube, Yandex Direct, bloggers and placements, content plan and metrics
- CRM chains and the loyalty program: triggers, frequency, consents, tiers, points lifetime and payback by cohort
- Launching an item in 30 days: the e-commerce route from a niche to the decision on scale