Marketing budget and testing impact
How much to spend from a revenue goal and how to check that a channel gives incremental sales, not reported ones.
About the topic
A marketing budget of “last year plus ten percent” says nothing about return: it repeats past decisions, good and bad. A sensible approach goes from goal to money: how many orders are needed, where visits will come from, how much can be paid for a visit and still stay in profit. And then a check: does the channel really deliver sales that would not have happened without it. The topic joins budget planning with an honest test of effect.
The decisions here are these. What budget to set for the period and how to split it between channels. What share to give brand awareness and what to sales now. Whether to continue a channel, scale it or stop. How to read platform reports and when not to trust them.
The reasoning starts with the funnel. A revenue goal of RUB 30 million a month at an average order of RUB 6,000 means 5,000 orders. At a 2% conversion that is 250,000 visits, of which, say, 100,000 are paid. If the margin of an order after delivery and returns is RUB 2,400, the allowable cost of a visit is 0.02 × 2,400 = RUB 48: paying more is not possible. The budget gets a boundary instead of a habitual number. It is not fixed forever: after every test it is reviewed, so the plan keeps a free share of money, for example a tenth, for testing new channels and moving funds from weak channels to strong ones.
The test of effect rests on comparison. A last-click report from a platform credits advertising with every order that had its click, although some customers would have come anyway. A test with a control group (incrementality) measures what advertising added on top. Example: the region with advertising gave 1,200 orders, a similar region without it 1,000: 200 were added. The advertising cost RUB 600,000: by the platform report (800 orders) an order costs RUB 750 and pays back at a margin of RUB 2,400, by the test it costs RUB 3,000 and does not pay back. A fourfold difference.
Start with the goal and the allowable cost of a visit, then choose one channel for a test with a control group. Traps: planning from last year, comparing channels over different attribution windows, not testing by “switch off and see”, and reading the result of a small test as an established fact.
Key ideas
Budget from the goal
Sensitivity to conversion. A goal of RUB 18 million at an order of RUB 4,500 is 4,000 orders, at a 1.5% conversion about 266,700 visits. An order margin of RUB 1,800 gives an allowable cost per visit of RUB 27. If conversion falls to 1.2%, the allowable cost becomes RUB 21.6, and a click at RUB 25 that was profitable turns into a loss.
Incrementality
The incremental effect is how many sales a channel added beyond what would have come without it. It is measured by comparing a group with advertising and a similar group without it. Without such a check, a last-click report credits advertising with other people’s orders, and the order price comes out several times lower than the real one.
Sample size and test length
A small test cannot tell effect from chance. To notice a conversion rise from 2.0 to 2.4% (a fifth) with usual confidence you need roughly 21,000 visitors in each group. If weekly traffic is 5,000 per group, the test runs about four weeks, not three days.
Regression: contribution of factors
Sales depend on price, discount, traffic and season at once. Regression separates the contribution of each factor with the others held constant: for example, by how many units sales grow with 10% more traffic. But a model is only as good as the variety of its data, and it shows association, not cause.
Small sample and Bayesian estimate
If a new model was bought out 6 times of 10 against a plan of 50%, you cannot say the buyout is 60%: the sample is small. A Bayesian estimate gives the probability that the share is above plan: with a flat prior it is about 73%. This is the language of probabilities instead of false precision.
Segment size: TAM, SAM and SOM by two paths, and the batch is checked against demand
The market is counted in three sizes. TAM is the whole category: RUB 53.84 billion. SAM, the market one and not the sewing one (see the term “SAM: serviceable available market”), keeps the buyers of your price segment and available channels at your price: RUB 1.62 billion, 3% of TAM. SOM is the smaller of two paths: top-down RUB 8.09 million (SAM × the capture share 0.5%), bottom-up RUB 7.27 million (clients from the budget RUB 1.2 million at a first-order cost of RUB 1,450). The calculated SOM of RUB 7.27 million is 0.45% of SAM and gives a demand of about 993 units, while the batch of 1,200 units is 121% of this demand.
Formula: TAM = people × buyers × units × market price; SAM = buyers × price-segment share × channel share × units × your price; SOM = the smaller of (SAM × capture share; clients from the budget × order × (1 − returns))
What to do: Count both paths of SOM and check the batch against demand: if the batch is above a year’s demand, cut it or revisit the assumptions, not the budget.
Think first: The acquisition budget is raised from 1.2m to RUB 1.45 million to cover a batch of 1,200 units. Is it enough?
Answer: No: the bottom-up path will grow, but SOM is the smaller of the two, and the top-down ceiling of RUB 8.09 million gives a demand of about 1,105 units. The batch of 1,200 is still larger.
An annual media plan spreads the budget over the demand season and the order price in a channel
The budget of RUB 1.2 million is divided among six channels and twelve months. The orders of a channel equal its budget divided by the order cost: 829 orders in total at RUB 1,448 on average, which is close to the first-order cost of RUB 1,450 in the segment size calculation. Autumn, September to November, gets 43% of the budget: RUB 516,000. Mailings and repeat orders give 240 orders at 10% of the budget because an order costs RUB 500, but the ceiling of the channel is one’s own client base. Revenue excluding VAT after returns is RUB 5.97 million, part of December orders arrive in January: RUB 119,324 is left beyond the year border.
Formula: Channel orders = channel budget ÷ order cost; monthly budget = annual budget × month weight ÷ sum of weights
What to do: Set the budget by the demand season and check the ceiling of cheap channels: their orders are limited by the base, not by money.
Think first: Why is the budget not split equally across months?
Answer: Demand is uneven: in autumn 43% of the budget, that is RUB 516,000, while an equal split would give RUB 300,000. The shortfall at the peak costs about 149 orders.
A content format: reach or an order
The margin per kept order is RUB 2,700 (an average order of RUB 7,320 with 22% VAT, margin 45%). Short video gives large reach, but a kept order from it costs RUB 3,936 and is more expensive than the margin; reviews and story series give an order for 557 and RUB 1,224. A reach format is assessed by reach with a separate goal, a selling one by the price of an order.
Formula: Price of a kept order = costs ÷ (clicks × conversion × (1 − returns))
What to do: Redistribute posts to formats with a cheap order, keeping reach where it is needed for the collection launch.
Attribution divides credit but does not prove causality
In the example by the last click Email gets 180, Search 60; by the linear one Search 93.3, Social 73.3, Email 73.3. The spread of models shows where an experiment is needed.
Formula: Position-based: 40% to the first and last, 20% among the middle ones
What to do: Do not cut a channel’s budget by one model; check with a test against a control.
Think first: A channel wins in all five models. What does it mean?
Answer: The assessment is reliable across rules but causality is still checked by an experiment.
The budget optimum equalizes the marginal return
Budget steps go to the channel with the highest marginal return while it is above 1: in the example 6 of 20 m is spent, profit 1.06 m RUB against −2.11 m at an equal split.
Formula: Marginal return = margin × (a ÷ b) × e^(−spend ÷ b)
What to do: Update the channel parameters every quarter.
A price test is decided by profit per visitor
In the example conversion falls from 3.20% to 2.80% while profit per visitor rises from 47.06 to RUB 52.65 (z = 1.96): at the significance bound.
Formula: Profit per visitor = CR × (price ÷ (1 + VAT) − cost)
What to do: Extend the test to significance and decide by profit.
How to work through it
- Set the revenue goal and turn it into orders, visits and the allowable cost per visit.
- Split the budget between channels using the allowable cost, not spending history.
- Pick one channel and set up a test with a control group: a region, a week or an audience without advertising.
- Calculate the sample size and duration in advance so as to notice the needed effect.
- Compare the test result with the platform’s last-click report and find the difference.
- For price, discount and traffic estimate the contribution of factors by regression and check it on new data.
- For new models with a small sample use a probabilistic estimate instead of a “percentage”.
- Set the target advertising cost ratio (DRR) and the payback time of acquisition; decide: scale, change or stop.
Where to start
First the budget from a revenue goal: how many orders and visits are needed and the allowable cost per visit. Then test the channel against a control group (incrementality), after that the target ad share, promo codes and a first-order discount. Calculations of the topic in order: “Marketing budget from a revenue goal”, “Channel incrementality: test against control”, “A/B test on proportions: sample size, duration and reading the result”, “Multiple regression: the contribution of price, discount and traffic to sales”, “Bayesian estimate of a rate: buyout or conversion of a new style from a small sample”, “Segment size: TAM, SAM and SOM in two ways (top-down by shares and bottom-up from the budget)”, “Path attribution models: last click, first, linear, position-based and time-decay”, “Advertising adstock and half-life”, “Optimal marketing budget allocation across channels”, “Price test: profit per visitor and significance”. See also (their primary topic is another one): “Promo code: payback with self-selection”, “First-order discount: CAC and LTV”, “Target ad-cost ratio and acquisition payback”, “Annual media plan: budget by channel and month, orders, revenue, margin after marketing, ad-cost share and payback”, “Content and SMM: reach, engagement, the price of a contact and an order by format, content payback”.
Calculations in this topic
Marketing budget from a revenue goal
Works backwards from a revenue goal: needed orders and visits, the allowable cost of a visit, the acquisition and brand budget and its share of revenue. Shows whether your cost of a visit pays back at your margin.
How much traffic is needed and how much a visit may cost to reach the revenue goal and stay in the black.
Channel incrementality: test against control
Compares the revenue of a group with ads on against a similar group without them. Shows incremental revenue, incremental return and profit after spend, and compares with the ad platform’s last-click report.
What the channel added to sales beyond what would have come without it and whether it pays.
A/B test on proportions: sample size, duration and reading the result
In the “Plan” mode counts how many visitors each group needs and how many days the test must run to notice a given conversion lift with the required power; in the “Result” mode tests the significance of the difference, shows the confidence interval of the difference and the smallest noticeable difference.
How many visitors and days are needed to notice a conversion lift, and whether to believe the difference the test showed.
Multiple regression: the contribution of price, discount and traffic to sales
From a table of periods estimates how sales depend on up to three factors at once (for example price, discount depth, traffic): coefficients with confidence intervals and p-values, R², a check of the factors’ interdependence (VIF), the effect of a one-standard-deviation shift of a factor and a forecast at given values with a band.
By how many units sales change when one factor changes with the others unchanged, and how far the model can be trusted.
Bayesian estimate of a rate: buyout or conversion of a new style from a small sample
Combines your prior estimate of a rate (by category or past styles) with a small number of observations: shows the posterior rate, the interval, the probability of exceeding a target, the weight of the data against the prior, a comparison with the ordinary Wilson interval and the expected range of successes in the next trials.
What the buyout or conversion rate is when data are scarce and with what probability it exceeds the plan.
Segment size: TAM, SAM and SOM in two ways (top-down by shares and bottom-up from the budget)
From a table of segments (how many people, what share buys the category, what share is in your price segment and reachable by your channels, how many units a year) it counts TAM (the whole category), SAM (your segment) and SOM in two ways: by a share of SAM and by the number of clients the budget brings. The calculated SOM is the smaller of the two; the calculation shows the gap between the estimates, the demand in units and checks it against the first batch. This is an order-of-magnitude estimate, not a forecast: every share is set by you.
How large the whole category is, what part of it is available to your price and channels, how much of it can really be taken in a year and whether such demand is enough for the first batch.
Path attribution models: last click, first, linear, position-based and time-decay
Splits the value of orders between channels in five ways from a table of customer paths: last click, first click, linear, position-based 40/20/40 and time-decay by the age of a touch. Shows each channel’s value in each model and the gap between them: where it is large, the assessment of the channel depends on the choice of model and cannot be decided by it alone. Attribution divides credit but does not prove causality; the proof of causality is a test against a control.
What value each channel gets in five models and how much the assessment depends on the choice of model.
Advertising adstock and half-life
Advertising works not only in the week of the show: the rest of the effect carries over to the following weeks. The calculation accumulates spend by the rule “accumulated effect = the week’s spend + carry-over share × the past accumulated effect”, gives the half-life and the full effect of a rouble, and, if there are sales, picks the carry-over share by search so that the regression of sales on the accumulated effect explains most. Use it together with the response curve of the channel.
How long advertising keeps working after the show and how many times its full effect exceeds the effect of the week of the show.
Optimal marketing budget allocation across channels
Splits the budget between channels so that profit is highest: each channel’s response grows ever more slowly with spend (a saturation curve with two parameters from the response curve fit), and each next step of budget goes to the channel with the highest marginal return. It stops when a rouble no longer pays back: the remaining money is not spent. Shows the budget by channel, the profit, the gain over an equal split and the marginal return.
How to allocate the budget across channels so that profit is highest and how much money is better not spent.
Price test: profit per visitor and significance
Compares two price levels by profit per visitor and not by the number of orders: the expensive variant almost always gets fewer orders but may earn more. Counts the conversion, profit per visitor and the significance of the difference both by conversion and by profit, and suggests the decision. The test should be run on your own site: platforms restrict A/B tests of price.
Which price brings more profit per visitor and whether the difference is significant and not just whether there are more orders.
Promo code: payback with self-selection
Splits orders with a code into “would have bought anyway” and truly additional ones, calculates the code’s profit, the discount lost on “own” buyers and the threshold share of “would have bought anyway” above which the code loses money.
Does a code pay back if part of the buyers would have bought without it.
First-order discount: CAC and LTV
Calculates the margin of a discounted first order, repeat purchases over 90 and 180 days, the acquisition cost and the customer payback period; compares with spending the same money without a discount if you have a control group.
Does a customer who came on a first-order discount pay back and is that better than acquisition without a discount.
Target ad-cost ratio and acquisition payback
Computes the target ad-cost ratio (DRR) from the margin before advertising, the share of fixed costs and the desired profit and compares it with the actual one. Separately estimates the CAC payback and the 12-month LTV; the traffic light works only on your thresholds.
What ad-cost ratio we can afford at our margin and how soon an acquired customer pays back.
Annual media plan: budget by channel and month, orders, revenue, margin after marketing, ad-cost share and payback
From the annual budget, channel shares and the cost of an order in each channel, month weights, the average order, returns, margin and the delay of the effect it spreads the budget by channels and months, counts orders, revenue ex-VAT after returns, margin after marketing, the ad-cost share, ROMI and the month when the cumulative margin covers the cumulative budget. All numbers follow from your assumptions: this is a teaching frame, not a forecast of advertising return.
How to split the annual budget by channels and months, how many orders and how much revenue it gives under your assumptions, when the cumulative margin will cover the cumulative budget and what ad-cost share comes out.
Content and SMM: reach, engagement, the price of a contact and an order by format, content payback
From the number of posts per month, their production and promotion, reach, engagements and clicks counts for each format and for all content: engagement rate (ER), the price of a thousand reach, of an engagement, a click and a kept order, the margin of orders and the payback. Shows which format gives reach and which orders, and at what number of orders the content pays back.
Which content format gives cheap reach and which gives cheap orders, what a kept order from content costs and at what number of orders the content pays back.
Self-check questions
- How many orders and visits are needed for the goal and what is the allowable cost per visit?
- Is the budget set from the goal, not from last year?
- Has the channel been tested with a control group, not only by the platform report?
- Is the control group similar to the test group in season, price and traffic?
- Are the sample and time enough to notice the needed effect?
- What are the order cost by the report and by the test and why do they differ?
- Were the same attribution windows used when comparing channels?
- What target DRR does our margin allow and in what time does the customer pay back?
- Is the conclusion “scale, change, stop” chosen with a criterion in advance?
Knowledge check
Multiple-choice questions. The correct answer and the explanation follow each question.
Top-down SOM is RUB 8.09 million, bottom-up RUB 7.27 million. Which SOM is taken into the calculation?
- A. 8.09 m ₽
- B. The average: 7.68 m ₽
- C. 7.27 m ₽
- D. 1.62 bn ₽
Correct answer: C. 7.27 m ₽. The smaller of the two is taken: 7.27 m ₽. The gap of the estimates of −10.2% is small; if it is above a half, one of the estimates is unreal. Covered in: Segment size: TAM, SAM and SOM by two paths, and the batch is checked against demand.
Demand at the calculated SOM is about 993 units, and the first batch is 1,200 units. What is visible?
- A. The batch exceeds a year’s demand by 207 units: 121% of demand
- B. The batch is half the demand: more can be ordered
- C. The batch equals the demand
- D. Demand and the batch cannot be compared: SOM is measured in roubles
Correct answer: A. The batch exceeds a year’s demand by 207 units: 121% of demand. Demand is SOM divided by the shelf price: RUB 7.27 million ÷ RUB 7,320 = 993 units. The batch of 1,200 is 121%: an excess stock. Covered in: Segment size: TAM, SAM and SOM by two paths, and the batch is checked against demand.
The budget grew to RUB 1.45 million. What demand results at the calculated SOM?
- A. About 1,105 units: the top-down ceiling does not let it grow
- B. About 1,200 units: the batch sells out at once
- C. About 993 units: the budget does not affect SOM
- D. About 1,200 units: the bottom-up path is taken
Correct answer: A. About 1,105 units: the top-down ceiling does not let it grow. The bottom-up path grew to RUB 8.78 million, but the smaller one is taken: top-down RUB 8.09 million, a demand of 1,105 units. Money does not replace the capture share. Covered in: Segment size: TAM, SAM and SOM by two paths, and the batch is checked against demand.
What share of TAM is SAM?
- A. 0.4%
- B. 0.5%
- C. 100%
- D. 3%
Correct answer: D. 3%. RUB 1.62 billion ÷ RUB 53.84 billion = 3%: of the whole category only a small part is available to your price and channels. TAM is a ceiling, not a target. Covered in: Segment size: TAM, SAM and SOM by two paths, and the batch is checked against demand.
In autumn, September to November, by the month weights 43% of the annual budget RUB 1.2 million goes. How much is it in roubles?
- A. 300,000 ₽
- B. 600,000 ₽
- C. 516,000 ₽
- D. 168,000 ₽
Correct answer: C. 516,000 ₽. The weights of September, October and November are 14, 14 and 15 out of 100: 43% × RUB 1.2 million = RUB 516,000. An equal split would give RUB 300,000. Covered in: An annual media plan spreads the budget over the demand season and the order price in a channel.
The margin per kept order is RUB 2,700, the price of an order from short video is RUB 3,936. What does it mean?
- A. The format must be closed at once
- B. The format is profitable because reach is large
- C. The price of an order is not compared with the margin
- D. By direct orders the format is loss-making; it needs a separate goal for reach
Correct answer: D. By direct orders the format is loss-making; it needs a separate goal for reach. An order more expensive than the margin means the direct effect does not pay for the format. But content also works for awareness, so the format is not closed but assessed by reach and a term of checking. Covered in: A content format: reach or an order.
How is the engagement rate ER counted?
- A. Orders ÷ reach
- B. Engagements ÷ reach × 100%, with the base labeled
- C. Costs ÷ engagements
- D. Clicks ÷ orders
Correct answer: B. Engagements ÷ reach × 100%, with the base labeled. ER shows the share of people who responded; the base (reach or followers) is labeled. Costs per engagement is the price of an engagement, and orders to reach is not ER. Covered in: A content format: reach or an order.
What does the spread between attribution models show?
- A. The exact cause of an order
- B. The size of the budget
- C. The number of channels
- D. How much the assessment of a channel depends on the choice of rule
Correct answer: D. How much the assessment of a channel depends on the choice of rule. A large spread means an experiment is needed. Covered in: Attribution divides credit but does not prove causality.
When does the budget optimum calculation stop allocating?
- A. When the week is over
- B. When the marginal return of all channels is below 1
- C. When all channels are equal
- D. Never
Correct answer: B. When the marginal return of all channels is below 1. A rouble does not pay back: the money stays in the company. Covered in: The budget optimum equalizes the marginal return.
What is a price test decided by?
- A. By the number of orders
- B. By the lowest price
- C. By profit per visitor and its significance
- D. By the number of clicks
Correct answer: C. By profit per visitor and its significance. The expensive variant loses orders but may earn more. Covered in: A price test is decided by profit per visitor.
Common mistakes
Mistake: Planning the budget from last year’s turnover
What it costs: Past mistakes repeat and the limit is not visible
How to avoid: Count from the goal and the allowable cost per visit.
Mistake: Trusting the last-click report
What it costs: Advertising gets other people’s orders and the order price is understated several times
How to avoid: Check the uplift with a control test.
Mistake: Comparing channels over different attribution windows
What it costs: One channel looks better because of the window
How to avoid: Set a common window.
Mistake: Stopping a test after three days
What it costs: Random noise is taken for effect
How to avoid: Calculate sample size and duration in advance.
Mistake: Reading a regression as a cause
What it costs: The association of two factors is taken for influence
How to avoid: Confirm findings with a test.
Mistake: SOM is taken as a share of TAM
What it costs: The batch and budget are built on a number dozens of times larger than the real one Demand would come out at about 36,774 units instead of 993: 37 times larger, and the batch would turn out unrealistically large.
How to avoid: Count the share of SAM and check it against the path from the budget Take the share of SAM, not of TAM, and check it bottom-up through the budget and the first-order cost.
What it looks like (illustrative figures): A share of 0.5% of TAM gives RUB 269.2 million, while the calculated SOM is RUB 7.27 million.
Mistake: A format is assessed by reach and likes
What it costs: Orders stay invisible The format eats RUB 185,143 a month and returns RUB 127,008 of the margin of direct orders.
How to avoid: Count the price of a kept order and ROMI. For a reach format set a separate goal and a term of checking, for the others compare the order price with the margin.
What it looks like (illustrative figures): Short video: reach of 18,000 per post, an order price of RUB 3,936 at a margin of RUB 2,700.
Taking attribution for causality
What it looks like (illustrative figures): A channel may not have influenced the order.
Using the parameters of one period
What it looks like (illustrative figures): Response curves change.
Russia and the EAEU
In Russia advertising is regulated by the advertising law (Federal Law 38-FZ): information must be truthful, and online advertising is marked and reported to the Unified Register of Internet Advertising (ERIR) through an ad system or operator; responsibility lies with the advertiser, so check the contract with the agency. Mailings are allowed only with the recipient’s consent (Article 18 of the law), and personal data fall under 152-FZ.
To measure effect in Russian channels, account for specifics: search advertising, ad networks, social networks with ad accounts and in-marketplace advertising give different reports and attribution windows, so compare them by a test, not by reports. Advertising on marketplaces (inside the platform) is measured by the advertising cost ratio (DRR) and grows with competition: check auction bids monthly. A brand books advertising expenses under tax accounting rules; normed expenses and VAT on services should be checked with an accountant. Currency risk applies to advertising in foreign systems: payments follow currency control rules.
Advertising marking and personal data requirements differ by country: Belarus, Kazakhstan, Armenia and Kyrgyzstan have their own advertising and data laws; a guideline, check with a local lawyer. Channels also differ: in Kazakhstan Kaspi and local platforms play the key role, Belarus and Armenia have other search and social networks, and Wildberries and Ozon are present in several countries. The calculation of the allowable cost per visit and the test with a control carry over unchanged; currency and VAT (VAT rates by country are on the page “EAEU and CIS countries”, a guideline as of 14.11.2026, verify) change the order margin.
Categories and audiences
- Apparel
- Apparel has high returns, so the order margin for the allowable cost per visit is counted after returns, not before. Demand is seasonal and the budget must run ahead of the peak. A control test is best run in a mid-load season so as not to confuse advertising effect with the effect of weather.
- Footwear
- Footwear has a higher order value and returns more often by size, so the allowable cost per visit is higher but returns and logistics eat the margin. The purchase decision takes longer: views, comparison, return to the basket. Remarketing gives a noticeable share, but its effect is overstated by last click, and a control check is especially important.
- Accessories
- Accessories are bought on impulse and as gifts: the order is small, the allowable cost per visit is low and gift seasons influence demand more. Advertising works from the date of the holiday. The contribution of add-on sales to a main purchase on the site is often higher than that of separate advertising.
- Kids
- The buyer is an adult and demand is tied to seasons: school, winter, holidays. Advertising is aimed at the child’s age, and the audience changes every year. For a children’s audience there are restrictions on advertising and on working with children’s data: check the rules and do not use data of minors.
- Men
- The men’s audience buys less often and more purposefully, and decides faster; search and direct visits give a larger share. Broad brand advertising makes less sense, while precise targeting and service work better. A control test shows that part of the advertised sales would have come anyway.
- Women
- The women’s audience is more active in social networks and content, chooses longer and compares more often, so the contribution of brand and content is higher and the chain of touches is longer. A seven-day attribution window undervalues advertising, while last click overvalues remarketing. Check channels by test and account for returns.
Questions this topic answers
- How much traffic does the goal need and how much can a visit cost?
- What share of the budget goes to brand and what to sales now?
- What did the channel really add and what would have come without ads?
- What part of sales did advertising create and what part did it take from organic?
Key measures and terms
Study first
What next
Related: “Customers and loyalty”, “Channels: wholesale, retail, platforms”, “Promotions: planning and evaluation”.
Documents on this topic
- CRM analytics: RFM, cohorts, retention, LTV/CAC, loyalty
- PR, seeding and influencer marketing: budget and evaluation
- Unit economics with buyout: CM1-CM3, LTV/CAC, customer survival, channel payback
- A/B tests for promotions and promos: hypothesis, sample, power, p-value, peeking
- Multiple regression of sales: demand drivers, advertising and promo contribution, multicollinearity, lags and model checks
- Bayesian estimates of conversion and buyout: small samples, prior distributions, shrinkage and comparison with the frequentist approach
- 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
- Brand platform: a one-page working template and a readiness checklist
- Target audience segmentation and segment size: TAM, SAM and SOM with honest assumptions
- Marketing launch calendar and a media plan with a budget: by channel and month, payback as a calculation
- Label launch: from an idea to the first sales — stages, documents, calculations in order and readiness criteria
- Customer behavior and research methods: from a hypothesis to a decision (interviews, surveys, A/B, JTBD, sample size)
- Content plan and SMM economics: reach, engagement, the price of a contact and an order, content payback
- Measuring marketing: path attribution, advertising carry-over, the budget optimum, the price test, NPS and the referral reward
- Production and content: shoot budget and calendar