Site economics and the funnel
From visit to payment: conversion by stage, profit per visit, the effect of a rise in conversion and order value and acquisition payback with returns.
About the topic
A brand’s online store earns money not on visits and not even on orders, but on paid and kept orders after returns and after the cost of traffic. This topic teaches you to see the site as a chain of steps, each with its own share of people who get through and its own price of loss, and to use that chain to decide where money goes: advertising, basket design, delivery, the product page or the size chart.
The logic is simple but has three layers. The first layer is the funnel. Of all visits some put an item in the basket, some reach checkout, some pay. These shares multiply, so the final conversion is always small and any change at one step carries into orders proportionally. The second layer is money. From an order you receive the ticket, then subtract returns, cost of goods, card-acquiring fee, delivery and packaging; what remains is the order contribution. Multiply it by the number of orders, subtract the cost of visits, and you get profit; divide by visits and you get profit per visit. That is the key number of the topic: it says how many roubles you can pay for one visit without going into the red. The third layer is time. A won customer does not always return, and not at once, so acquisition pays back over several months, not in one order, and those months must be financed.
The decisions made here are practical. Should you raise the advertising bid when the price of a visit grows from 12 to 15 roubles. What is better: remove a field at checkout or add instalments. Is it worth lifting the average ticket by 10% through recommendations if returns will grow with it. Is there enough cash for acquisition if a customer pays back in six months.
Start not with a pretty dashboard but with honest inputs: visits, baskets, checkouts, payments, returns and average ticket for the same period and by the same definition of an order. Then compare the funnel by source and by new versus returning visitors, because the average funnel hides differences of several times. Only then count the effect of improvements.
Typical traps: counting payback without returns (in apparel they often decide everything), comparing channels by the average, judging a change without a control group, and mixing session conversion with visitor conversion. For footwear and women’s apparel, size-driven returns are usually the main line, so here the funnel ends not at payment but at keeping the item.
Key ideas
Funnel steps multiply, and a point on a small base weighs more
Say 6% of visits add to basket, 45% of baskets reach checkout, 55% of checkouts pay: 6 × 0.45 × 0.55 = 1.485% of visits, that is 2,970 orders from 200,000 visits. Add one point at each step: basket 7% gives +16.7% orders, checkout 46% gives +2.2%, payment 56% gives +1.8%. So compare improvements as a percentage of the current step level, not in points.
Profit per visit and the maximum price of a visit
Illustrative example: 100,000 visits, 1,485 orders at 6,500 roubles, returns 25%, margin 52%, variable costs 14% of revenue after returns. After returns revenue is 7.24 million, contribution (52 − 14 = 38%) is about 2.75 million, that is 27.5 roubles per visit. That is the maximum price of a visit: at 12 roubles you keep 15.5 roubles of profit, at 30 you lose money even though orders grow.
A higher ticket pays only while returns grow more slowly
A ticket of 6,500 roubles with 25% returns leaves 4,875. If recommendations lift the ticket by 10% to 7,150, at the same 25% you keep 5,362. At 30% returns it is 5,005 (still better), at 33% it is 4,790 (already worse). The boundary: 4,875 ÷ 7,150 = 68.2% kept, so returns no higher than 31.8%. Always check where returns went after the change.
Acquisition payback is counted in orders and months
Illustratively: acquiring a customer costs 2,500 roubles, the contribution of one kept order is 4,875 × 38% ≈ 1,850 roubles. One order does not cover the cost (you need 2,500 ÷ 1,850 ≈ 1.35 orders). With 1.2 repeat orders a year, the second order comes on average after several months, and that gap has to be financed. So payback is compared with your own target term, not with an abstract good.
The average funnel misleads: count by segment
New visitors convert at 1.0%, returning at 3.0%. With 60% new, the average conversion is 0.6 × 1 + 0.4 × 3 = 1.8%. Advertising brought more new people, the share became 70%, and the average fell to 1.6% (−11%), though no segment got worse. Without the split you will fix what is not broken.
The abandoned cart chain pays back through the response
In the example 14,000 carts are abandoned, 4,900 reachable, 392 orders returned, profit RUB 471,787; break-even response 0.90%.
Formula: Profit = incremental orders × check × margin × redemption − cost
What to do: Determine by a test how many would have returned on their own.
The delivery threshold is chosen by the distribution of checks
Moving the threshold 3,000 → 4,000 cuts the share of free orders from 60% to 42.5%, the change in profit is RUB 53.7 per order.
Formula: Δ = saving + catch-up margin − losses
What to do: Check the threshold with a test on part of the traffic.
The instalment pays back through conversion growth
The average fee over the mix is 2.03%, the break-even conversion growth with the instalment 3.3%.
Formula: ΔCR* = margin ÷ ((1 + ΔAOV) × (margin − fee)) − 1
What to do: Take rates from contracts on the date.
How to work through it
- Fix the definitions: what counts as a visit, basket, order and kept order, over which period and with or without VAT. One definition for the whole team.
- Build the funnel from four numbers: visits, baskets, checkouts, payments. Compute step shares and compare them with the previous period.
- Add money: ticket, returns, margin, variable costs. Get the order contribution and profit per visit.
- Compare profit per visit with the price of a visit by source: which channels pay more than they earn.
- Pick the step to improve by the effect as a percentage of its current level, not by feel. Estimate the monthly cost of the change.
- Check the side effect: a higher ticket can raise returns, a basket discount can cut margin. Compare scenarios.
- Check acquisition payback with returns and compare it with your target term. Launch the change with a control group.
Where to start
Start with “The site funnel”, then “What a rise in conversion and order value gives” and “Acquisition payback with returns”. Calculations of the topic in order: “The site funnel: from visit to payment and money per visit”, “What a rise in conversion and order value gives: effect on profit and the price of the change”, “Acquisition payback with returns”, “Abandoned cart recovery: payback of the chain”, “Free shipping threshold”, “Payment fees and instalment payback”. See also (their primary topic is another one): “CAC and LTV: quick acquisition check”.
Calculations in this topic
The site funnel: from visit to payment and money per visit
From visits and the shares of carts, checkouts and payments calculates orders, conversion, contribution and profit per visit, the maximum price of a visit and which stage gives the most orders for one point of improvement.
How many orders and how much profit the funnel gives, how much one can pay for a visit and which stage to improve.
What a rise in conversion and order value gives: effect on profit and the price of the change
Compares scenarios: a rise in conversion, a rise in order value and the possible rise of returns with it. Shows the profit of each scenario against the base and by how many percent conversion or order value must rise to pay for the monthly price of the change.
What a rise in conversion or order value gives allowing for returns and whether it pays for the change.
Acquisition payback with returns
Counts in how many months the customer’s contribution (the first order and repeat purchases) covers the acquisition cost, once returns and variable costs are included. Shows the gap against a “naive” calculation without returns and the maximum acquisition cost for your payback term.
In how many months an acquired customer pays back the cost allowing for returns and how much one may pay to acquire.
Abandoned cart recovery: payback of the chain
Counts the economics of one abandoned-cart reminder chain: how many carts are abandoned, how many can be reached (there is consent and a contact), how many return, how many of them would have ordered anyway, and what profit remains after the margin, redemption and the cost of the chain. Shows the payback, the cost of a recovered order and the minimum response at which the chain pays back.
Whether the reminder chain pays back and at what response it breaks even.
Free shipping threshold
Compares profit at two free shipping thresholds by the distribution of checks and not by the average check: how many orders stop being free and how much money that saves, what share of customers from the band under the new threshold will add items to reach it (catch-up) and what margin this gives, how many orders are lost because of the new delivery fee. Runs through the thresholds at band boundaries and shows the best.
Which free shipping threshold gives the highest profit per order at your distribution of checks.
Payment fees and instalment payback
Adds the fees of payment methods into an average rate for your payment mix and shows what it costs per order and per month. Separately assesses the instalment: the fee is charged on the whole order while the check and conversion grow. Finds the growth in conversion at which the instalment pays back and compares it with the expected one. Take the rates from contracts: they change.
What the fee is on average over the payment mix and whether the instalment pays back.
CAC and LTV: quick acquisition check
Calculates what a new customer costs to acquire, how much profit they bring over their lifetime and how fast marketing spend pays back. A quick estimate on averages; for the exact payback use Customer cohort.
What it costs to acquire a customer, how much profit they bring and how fast the investment returns.
Self-check questions
- Does the definition of an order match in site analytics, in accounting and in the management report?
- At which funnel step do most people drop out and how large is it as a percentage of its base?
- What is profit per visit and is it above the price of a visit in every source?
- Are returns and their causes (size, quality, expectation) included in the order contribution?
- Which segment (new, returning, mobile, app) drags the average funnel down?
- Will returns grow if you lift the average ticket, and up to what level is the change still worthwhile?
- After how many months will a customer repay acquisition and can you finance the wait?
- Is there a control group or period to test the change, and does it hold enough orders?
- Do you know what share of revenue comes from one advertising platform and what happens if it is switched off?
Knowledge check
Multiple-choice questions. The correct answer and the explanation follow each question.
What does the “order contribution” show in the site economics calculation?
- A. Order revenue excl. VAT: order value × number of orders
- B. Order margin before returns: order value × (margin − variable costs)
- C. Money from one order after returns and variable costs, less the cost of handling returns
- D. Site profit after traffic, divided by the number of visits
Correct answer: C. Money from one order after returns and variable costs, less the cost of handling returns. Contribution = order value × (1 − returns) × (margin − variable costs) − share of returns × handling. With an order value of RUB 5,000, returns of 20%, a margin of 50%, variable costs of 10% and handling of RUB 250: 5,000 × 0.8 × 0.4 − 0.2 × 250 = RUB 1,550. If returns are ignored, you get 5,000 × 0.4 = RUB 2,000: the contribution is overstated by RUB 450 (this is the “margin before returns” option). Order revenue shows neither cost nor returns, and profit per visit is contribution × orders ÷ visits minus the cost of a visit: it is calculated later.
The site had 80,000 visits in a month. 5% of visits add to the cart, 40% of carts reach checkout, 60% of checkouts are paid. How many orders will there be?
- A. 960 orders
- B. 1,600 orders
- C. 2,400 orders
- D. 4,000 orders
Correct answer: A. 960 orders. 80,000 × 0.05 × 0.4 × 0.6 = 960 orders, a conversion of 1.2%. 1,600 comes from ignoring the payment stage (80,000 × 0.05 × 0.4), 2,400 from skipping checkout (80,000 × 0.05 × 0.6), and 4,000 is merely the number of carts: 80,000 × 0.05.
The funnel: 80,000 visits, cart 5%, checkout 40%, payment 60%. The checkout share was raised from 40% to 41%. How many orders a month will be added?
- A. 9.6 orders
- B. 16 orders
- C. 192 orders
- D. 24 orders
Correct answer: D. 24 orders. One point at a stage is 0.01 of visits × the shares of the other stages: 80,000 × 0.05 × 0.01 × 0.6 = 24 orders. The same point at the cart stage would give 192 orders (80,000 × 0.01 × 0.4 × 0.6), at the payment stage 16. 9.6 comes from taking 1% of 960 orders: that is a relative growth of 1%, not a point.
Acquiring a customer costs RUB 2,300. Order value RUB 5,000, returns 20%, margin 50%, variable costs 10%, return handling RUB 250, the customer makes 1.8 repeat orders a year. In how many months will the customer pay back the acquisition?
- A. 1.0 months
- B. 3.2 months
- C. 9.9 months
- D. 1.5 months
Correct answer: B. 3.2 months. Order contribution with returns: 5,000 × 0.8 × 0.4 − 0.2 × 250 = RUB 1,550. After the first order 2,300 − 1,550 = RUB 750 remains, repeats cover it: 1,550 × 1.8 / 12 = RUB 232.5 a month, 750 / 232.5 ≈ 3.2 months. 1.0 month comes out when returns are ignored (contribution RUB 2,000: a remainder of RUB 300 at RUB 300 a month). 9.9 comes from not counting the first order (2,300 / 232.5), and 1.5 from dividing the acquisition cost by the order contribution (1.48) and calling it months.
Base: 100,000 visits a month, conversion 1.2%, a visit costs RUB 8, order value RUB 5,000, returns 20%, margin 50%, variable costs 10%, return handling RUB 250. What gives more profit: conversion up 5% (relative) or order value up 5% together with returns up 2 pp?
- A. Conversion: +RUB 93,000 versus +RUB 39,600 for order value
- B. Order value: +RUB 96,000 versus +RUB 93,000 for conversion
- C. Order value: +RUB 300,000 versus +RUB 93,000 for conversion
- D. The same: both scenarios give +RUB 53,000 (5% of profit)
Correct answer: A. Conversion: +RUB 93,000 versus +RUB 39,600 for order value. Base: order contribution RUB 1,550, profit after traffic 100,000 × 0.012 × 1,550 − 100,000 × 8 = RUB 1,060,000. Conversion +5%: 1,260 orders, +RUB 93,000. Order value RUB 5,250 at returns of 22%: contribution 5,250 × 0.78 × 0.4 − 0.22 × 250 = RUB 1,583, a gain of +RUB 39,600. Order value would win (+RUB 96,000) only with unchanged returns: that is how the scenario with forgotten returns looks “better”. RUB 300,000 is revenue growth (1,200 × 5,000 × 5%), not contribution, and RUB 53,000 is 5% of profit, but each idea changes profit in its own way.
What must be subtracted when counting an abandoned cart chain?
- A. All abandoned carts
- B. Orders the customer would have made on their own
- C. The margin
- D. Redemption
Correct answer: B. Orders the customer would have made on their own. Otherwise the effect of the chain is overstated. Covered in: The abandoned cart chain pays back through the response.
What is the choice of the free shipping threshold based on?
- A. On the average check
- B. On a competitor’s delivery price
- C. On the distribution of checks by band
- D. On the number of orders
Correct answer: C. On the distribution of checks by band. The average hides the distribution. Covered in: The delivery threshold is chosen by the distribution of checks.
When does the instalment pay back?
- A. When the conversion growth is not below the break-even one
- B. Always
- C. When the fee is above the margin
- D. When there are no returns
Correct answer: A. When the conversion growth is not below the break-even one. The fee is charged on the whole order. Covered in: The instalment pays back through conversion growth.
Common mistakes
Mistake: Counting acquisition payback without returns
What it costs: The customer repays on paper but not in the till
How to avoid: Use the contribution of a kept order.
Mistake: Watching a single average funnel across all devices and sources
What it costs: You fix what is not broken
How to avoid: Count by segment.
Mistake: Judging a change without a control group
What it costs: Season or promotion pass as the effect
How to avoid: Keep a control or compare with a similar period.
Mistake: Comparing channels by average CAC
What it costs: The cheap customer turns out to be one-off
How to avoid: Compare by contribution over time, not by acquisition price.
Mistake: Measuring conversion by sessions and calling it customer conversion
What it costs: The figure depends on the number of visits
How to avoid: Agree what to count and label the base.
Not subtracting orders the customer would have made anyway
What it looks like (illustrative figures): Some would have returned without a reminder.
Relying on the average check when choosing a threshold
What it looks like (illustrative figures): The distribution of checks is not symmetric.
Russia and the EAEU
On a Russian site, the consumer price is shown including VAT (the standard rate is 22% from 01.01.2026, 10% for children’s goods on the list when conditions are met; confirm with your accountant), while funnel calculations run excluding VAT, otherwise contribution and margin are overstated. Payment on the site goes with an online cash register: a receipt is mandatory, and for goods under mandatory labeling the code enters the receipt. Return terms for distance sales are set by the Consumer Protection Law: refusal of a proper-quality item is possible within seven days after receipt, and the refund is due within ten days of the claim; ask a lawyer about specifics for apparel and footwear, since some goods fall under exceptions.
Collecting data on visitors and customers is governed by Federal Law 152-FZ: you need consent, a clear policy and storage of Russian citizens' data in databases in Russia. Advertising that brings traffic must be labeled with a token (erid), so include the effort of ad reporting in the price of a visit. Payment methods affect conversion at the payment step: Mir cards, the Faster Payments System (SBP), instalments and pay-in-parts; fees and terms differ by provider, so enter your own values and check the contract. If part of sales goes through marketplaces, their own funnel and fees are not equal to your site’s economics: compare both after all deductions.
The funnel and contribution principle is the same in every country; rates and rules change. Reference VAT as of the check date 14.11.2026, confirm with your accountant: Belarus 20% (10% on a list), Kazakhstan 16% from 2026, Armenia 20%, Kyrgyzstan 12% plus a sales tax at the rate for the goods. When selling to a buyer in another union country, taxes follow the destination-country rules, the site shows prices in local currency, and the calculation gets the exchange rate as a separate risk. Cross-border delivery and returns are noticeably more expensive and slower, so put them into the order contribution separately. Local payment methods and platforms differ by country; check exact fees and return rules locally.
Categories and audiences
- Apparel
- The main question is returns by size and expectation: customers order two sizes and send one back. Watch the keep rate by size and by product page; a size recommendation often gives more than advertising. The ticket grows with multi-item baskets, but returns grow with it.
- Footwear
- Footwear is returned more often than almost any category: last, width, stiffness. The funnel must end at keeping the pair, so watch the return share by model and size run, use in-store fitting and fitting delivery where available. The cost of reverse delivery cuts contribution noticeably.
- Accessories
- Few size variations mean fewer returns, but higher seasonal gift peaks and more dependence on price and photos. Conversion is usually lower and the ticket smaller, so the price of a visit is critical. Test cross-sell on the main purchase and the minimum ticket for free delivery.
- Kids
- A parent buys, but a growing child wears: size errors are more frequent, and peaks are sharply tied to the back-to-school season in late summer and New Year. Repeat purchases exist and the cycle is short, so acquisition payback may be faster.
- Men
- Men more often buy with purpose: higher conversion, shorter path, fewer views and returns, but a smaller ticket. Quick reorders of basics and size subscriptions win. Measure the funnel separately from women’s, because the average hides different behaviors.
- Women
- Women choose longer, add to favorites and basket more often, and return more often. Conversion is lower, but ticket and repeat purchases are higher, so judge by contribution over time, not by the first order. Photos, a fit description and size reviews influence the keep rate more than a discount.
Questions this topic answers
- Which funnel stage loses most orders?
- Does a rise in order value pay for a rise in returns?
- In how many months does a customer pay back acquisition?
- What does acquisition cost at your conversion and how does it differ by channel?
- What adds more profit: more traffic, higher conversion or a higher average order?
Key measures and terms
Study first
What next
Related: “Marketing budget and testing impact”, “Customers and loyalty”, “Channels: wholesale, retail, platforms”.
Documents on this topic
- Production and content: shoot budget and calendar
- Business continuity plan (BCP) and crisis measures
- Scale and barriers: economies of scale, experience curve, CAC, tipping point
- Site, CRM and receipts: abandoned carts, the delivery threshold, payments, message frequency, the lapsed threshold, day indices and pairs of items