Scale and barriers: economies of scale, experience curve, CAC, tipping point
The document shows where growth still makes a unit cheaper and where it starts costing more: economies of scale in buying and logistics, the experience curve, entry and scaling barriers, customer acquisition cost (CAC) in the fight for traffic, and the tipping point after which growth spend stops paying back.
Why this document
The owner and the strategy lead read this when deciding whether to keep growing, enter a new channel or country, and how much to pay for a customer. It matters because growth is not always profitable: while purchasing and logistics get cheaper with volume, scale helps, but once CAC climbs in the fight for traffic and management gets heavier, each unit costs more. If the tipping point is not found early, revenue grows while profit shrinks, and the company learns it from a cash gap.
At a glance
- What is this document and why is it needed?
- The document shows where growth still makes a unit cheaper and where it starts costing more: economies of scale in buying and logistics, the experience curve, entry and scaling barriers, customer acquisition cost (CAC) in the fight for traffic, and the tipping point after which growth spend stops paying back.
- Who owns the document and how often is it rebuilt?
- Maintained by the strategy or finance director with the commercial and marketing directors; shown to the owner at strategy and growth budget reviews, recalculated twice a year.
- What mistakes are the most common?
- Deciding on average CAC -> budget grows past the tipping point -> compute marginal CAC and customer contribution. Customer contribution measured by revenue -> losses on buyout and returns -> 12-month contribution after cost, returns and delivery. A ready-made learning rate instead of own data -> a false saving -> estimate from batches.
For whom
- Brand
- For a brand, scale works through order volume at the factory (price and minimums), shared material base, product platform and brand awareness; barriers: access to factories and materials, distribution, brand, inventory capital. CAC rises when going beyond the core audience.
- Retail
- For a retailer, scale works through buying terms, logistics, bundled leasing, data and loyalty; barriers: locations, suppliers, network capital, traffic. Scale benefits quickly hit store cannibalization and the cost of traffic.
What it consists of
1. Sources of scale effects
- Buying: unit price falls with order volume (tier discounts, smaller changeover share); counted by MOQ and price tiers.
- Logistics and warehouse: fixed warehouse costs spread over more units; full-truck transport; limit: warehouse throughput and expansion steps.
- Management and data: head office fixed costs are spread over growing revenue; data improves forecasting and markdown.
2. Experience curve
- Brand: Each doubling of cumulative volume lowers unit cost by a fixed share (learning rate, for example 85%); applies to sewing and repetitive operations, weakly to materials.
- Check: the learning rate is estimated from own data (operation cost by batch), not taken as a given number.
3. Entry and scaling barriers
- Brand: Entry: inventory capital, access to factories and materials, certification and labeling, brand, platforms.
- Scaling: quality at higher volume, management load, people, IT, cannibalization, traffic cost; for each barrier an estimate of the cost and time to overcome it.
4. CAC and the fight for traffic
- Response model: customer count depends on budget with diminishing returns; average CAC is budget divided by customers; marginal CAC exceeds the average.
- Decisions use marginal CAC and customer contribution over 12 months after returns, not average CAC.
5. Tipping point
- Growth tipping point: the budget at which marginal CAC equals customer contribution; above it every extra rouble loses money.
- It moves up with customer contribution (repeat purchases, ticket, buyout rate), organic traffic and conversion; it moves down with higher auction bids and competition.
6. Brand and retailer: where scale shows up
- Brand: Production scale: moving from a micro-batch to a run, fabric and sewing price steps; the criterion is saving per unit against the cost of extra stock and markdown probability.
- Retailer: Network scale: head-office and DC fixed costs are spread over the number of stores; a new store dilutes the center but adds rent and payroll; evaluate the marginal store, not the average.
- Brand: Channel cannibalization: D2C and marketplace growth against wholesale; price and discount rules for partners so that the own web shop does not take sales from wholesale clients without compensation.
- Retailer: Cannibalization between stores and between a store and online in the overlap zone; include online orders collected in store and compute the zone contribution, not each point separately.
- Brand: Brand barriers: factory capacity and quality as volume grows, AQL control, certification and marking of new groups, materials stock.
- Retailer: Retailer barriers: access to malls and sites, people (store directors), IT and warehouse, the contract base with brands and exclusives.
- Brand: Brand marginal CAC: computed on D2C customer contribution, taking into account the wholesale revenue share that needs no end-customer acquisition spend.
- Retailer: Retailer marginal CAC: customer contribution net of marketplace commission, delivery and returns; loyalty lowers CAC through repeat visits.
How to read the result and which decisions follow
- The optimal budget of RUB 13.1 m per month gives 5,461 clients per month, the marginal CAC of RUB 3,200 equals the client contribution and profit is RUB 4.37 m over 12 months. Decision: investing beyond S* means paying for a client more than the client brings.
- Brand: the experience curve gives RUB 614 per unit after three doublings and RUB 377 after six (from RUB 1,000). Decision: scale pays where volume really doubles, not where only turnover grows; count doublings by the cumulative output of the style.
- Retailer: scale in a chain works through buying, logistics and rent; but each new store reduces average revenue per square meter if the market is limited. Decision: look at the marginal, not the average, payback of a new store.
- Brand: the sensitivity table shows that at b = 0.85 the optimum rises to RUB 106 m against 13.1 at b = 0.75. Decision: estimate b from real data at several budget levels; an error in b changes the recommendation most.
- Retailer: entry barriers (rent in the best places, long contracts, the client base) protect you as long as you do not grow beyond market capacity; count market share in the district, not in the country.
Formula derivation and sensitivity tables
- Derivation of marginal CAC. From N = a x S^b it follows that dN/dS = a x b x S^(b-1) = b x N / S. Marginal CAC equals dS/dN = S / (b x N) = CAC_avg / b; at b < 1 it is above the average.
- Derivation of the optimum. Profit Π = V x a x S^b - S. The condition dΠ/dS = V x a x b x S^(b-1) - 1 = 0 gives S^(1-b) = V x a x b, hence S* = (V x a x b)^(1/(1 - b)). At V = RUB 3,200, a and b = 0.75 we get S* from the table.
- Derivation of the experience curve: if each doubling of volume multiplies cost by λ, then after k doublings c_k = c_1 x λ^k = c_1 x (V_k / V_1)^(log2 λ); for λ = 0.85 the exponent is about -0.234.
- Numeric table of S*, RUB m per month (rows b = 0.65..0.85, columns client contribution -20..+20%): b 0.65: 2.7 / 3.8 / 5.1 / 6.7 / 8.6; b 0.7: 3.6 / 5.4 / 7.6 / 10.5 / 14.0; b 0.75: 5.4 / 8.6 / 13.1 / 19.2 / 27.2; b 0.8: 9.5 / 17.1 / 29.0 / 46.6 / 72.1; b 0.85: 23.9 / 52.4 / 105.7 / 199.6 / 356.5.
- Brand: sensitivity to the client contribution: at V ±10% the optimum changes by ±46% (factor (1 ± 0.1)^(1/(1 - b))), that is a contribution 10% higher calls for a budget almost 46% bigger.
Control checks
- Check: the optimum S* of RUB 13.1 m is confirmed by the condition CAC_m = V (RUB 3,200 at a contribution of RUB 3,200); profit at S* is higher than at neighboring levels (code).
- Brand: the client contribution V is computed over 12 months with repeat purchases and returns, not on the first order.
- Retailer: b and the learning rate are example assumptions; confirm them from your own history and do not carry them across categories.
- Brand: the experience curve holds for repeatable operations (sewing, cutting); for a new technology the rate must be measured again.
- Retailer: define the tipping point (from gain to loss of scale) by marginal, not average, indicators.
Where the data comes from
| Data source | Fields |
|---|---|
| Ad campaign reports | Spend, impressions, clicks, orders, new customers by period and channel |
| CRM and cohort analysis | Customer contribution by cohort over 12 months, repeat purchases, returns |
| Buying and production | Prices by volume tier, minimums, operation cost by batch |
| Logistics and warehouse | Fixed and variable warehouse cost, throughput, expansion steps |
How to build it step by step
- Step 1. Collect a series of pairs (acquisition spend, new customers) by week or month and channel.
- Step 2. Fit customers against spend on logs: the slope gives the response exponent (below 1 means diminishing returns).
- Step 3. Compute customer contribution over 12 months by cohort: revenue less cost, returns, delivery, commissions.
- Step 4. Build a table by budget level: customers, average CAC, marginal CAC, profit.
- Step 5. Find the budget where marginal CAC equals customer contribution; this is the upper bound and the working budget sits below it with a margin.
- Step 6. Estimate buying scale effects from price tiers and MOQ and logistics scale from fixed costs.
- Brand: Step 7. Estimate the experience curve by batch and test the learning rate on recent data.
- Step 8. Build a barrier register with cost and time to overcome; pick the bottleneck.
- Step 9. Build growth scenarios (core, adjacent segments, new countries) with different customer contribution and CAC.
- Step 10. Approve a growth budget with a stop rule: when marginal CAC exceeds the threshold, growth is paused.
- Brand: Step 11. Brand: compare the run and the micro-batch on unit price including stock carrying cost, and check D2C growth for wholesale cannibalization.
- Retailer: Step 12. Retailer: compute the marginal store contribution and the effect on neighboring stores and online before opening a new point.
Formulas
| Measure | Formulas | Note |
|---|---|---|
| Response to spend | Customers N = a x S^b, S is budget, b < 1 is the response exponent | Parameters are estimated from own data; each channel has its own. |
| Average and marginal CAC | Average CAC = S / N; marginal CAC = dS / dN = average CAC / b | Decisions use marginal CAC. |
| Growth tipping point | Marginal CAC = V (customer contribution), so S* = (V x b x a)^(1 / (1 - b)) | Units: S in the same monetary units as V; check dimensions. |
| Experience curve | Unit cost at volume n = c1 x r^(log2 n), where r is the learning rate (0.85 means -15% per doubling) | Applies to operations, not to material prices. |
Formulas in notation
- Return on budget
N = a × S^b
S is budget, N new clients, b < 1 the diminishing return exponent (0.75 in the example).- Optimal budget
S* = (V × b × a)^(1 / (1 − b))
V is the client contribution over 12 months; the condition CAC_m = V.- Profit
Π(S) = V × N(S) − S
The maximum is reached where marginal contribution equals marginal cost.
Worked example
Exercise: response to acquisition spend, budget in RUB m a month
| Budget, RUB m | New customers | Average CAC, RUB | Marginal CAC, RUB | 12-month profit, RUB m |
|---|---|---|---|---|
| 1 | 793 | 1,261 | 1,682 | 1.54 |
| 2 | 1,333 | 1,500 | 2,000 | 2.27 |
| 4 | 2,242 | 1,784 | 2,378 | 3.18 |
| 8 | 3,771 | 2,121 | 2,828 | 4.07 |
| 16 | 6,342 | 2,523 | 3,364 | 4.30 |
| 32 | 10,667 | 3,000 | 4,000 | 2.13 |
Notional: response exponent 0.75, customer 12-month contribution RUB 3,200. Average CAC grows more slowly than marginal CAC, so by the average, growth looks profitable while by the marginal it no longer is. Tipping point: a budget of about RUB 13.1m a month, 5,461 customers, marginal CAC RUB 3,200 equals contribution; profit is highest on the grid at RUB 16m. An experience curve at 0.85 lowers operation cost from RUB 1,000 to RUB 614 after three doublings and to RUB 377 after six. Numbers are notional.
Figures
- Average CAC
- Marginal CAC
- Client contribution
Data as a table
| Measure | Average CAC | Marginal CAC | Client contribution |
|---|---|---|---|
| 1 | 1,261 | 1,682 | 3,200 |
| 2 | 1,500 | 2,000 | 3,200 |
| 4 | 1,784 | 2,378 | 3,200 |
| 8 | 2,121 | 2,828 | 3,200 |
| 16 | 2,523 | 3,364 | 3,200 |
| 32 | 3,000 | 4,000 | 3,200 |
Average and marginal CAC by budget. Marginal CAC grows faster than the average; the crossing of marginal CAC with the client contribution of RUB 3,200 gives the optimal budget. Horizontal axis: budget, RUB m per month.
Data as a table
| Measure | Profit over 12 months |
|---|---|
| 1 | 1.54 |
| 2 | 2.27 |
| 4 | 3.18 |
| 8 | 4.07 |
| 16 | 4.3 |
| 32 | 2.13 |
Twelve-month profit by budget. Profit grows up to the optimum and falls after it: an excess budget destroys the return. Horizontal axis: budget, RUB m per month.
Data as a table
| Measure | Unit cost |
|---|---|
| ×1 | 1,000 |
| ×2 | 850 |
| ×4 | 722 |
| ×8 | 614 |
| ×16 | 522 |
| ×32 | 444 |
| ×64 | 377 |
Experience curve: unit cost as volume grows. Each doubling of cumulative volume cuts unit cost to 85%: after six doublings 38% remains.
- 2.7 … 357 RUB m
Data as a table
| -20% | -10% | 0% | +10% | +20% | |
|---|---|---|---|---|---|
| b = 0.65 | 2.7 | 3.8 | 5.1 | 6.7 | 8.6 |
| b = 0.7 | 3.6 | 5.4 | 7.6 | 10.5 | 14 |
| b = 0.75 | 5.4 | 8.6 | 13.1 | 19.2 | 27.2 |
| b = 0.8 | 9.5 | 17.1 | 29 | 46.6 | 72.1 |
| b = 0.85 | 23.9 | 52.4 | 106 | 200 | 357 |
Optimal budget: return exponent and client contribution. The budget S* in RUB m per month at different b and client contribution ±10% and ±20%: the closer b is to one, the sharper the optimum grows.
Specifics by category
- Apparel
- Scale in buying comes through fabric MOQ: raising the order from 3,000 to 6,000 m gives a price-step discount (an assumption of 4-7%) but doubles the aging risk. Compare the discount with the stock carrying cost and the markdown probability. The experience curve works on repetitive sewing (a basic T-shirt) and weakly on complex dresses. Compute 12-month customer contribution after returns and delivery: in fashion returns eat a visible part of the CAC headroom.
- Footwear
- Scale is strongest in tooling: last and mold cost per pair equals tooling cost divided by the run; at RUB 300,000 and 1,500 pairs that is RUB 200 per pair, at 6,000 pairs RUB 50 (an assumption). So model MOQ is high and the risk of broken size runs grows: doubling the run does not double the fast sizes, demand concentrates at the center of the run. A learning rate of 85-90% (an assumption) holds for sewing and lasting, not for sole molding with fixed molds.
- Accessories
- Scale is limited by recognition and unit price: a bag or jewelry is bought rarely and for image, so CAC grows faster than for basic apparel. Buying is bound by leather and hardware MOQ, materials are expensive and changeover shows in price. For precious-metal jewelry scale is capped by the metal price (which does not scale) and the marking circuit; costume jewelry scales as a mass category. A high ticket allows a higher marginal CAC if repeat purchases are confirmed by data.
- Kids
- Growth is limited by demographics and season windows: the customer changes size every 1-3 years, so customer contribution is computed over the family life cycle (younger children, school, winter), not for one child. Buying scale hits the height run: with a 6 cm step a big run without fast turnover ages as the child grows. Entry barriers are higher: TR CU 007/2011, tests and a document for each group, with certification cost included in model cost.
- Men
- A high basics share gives an experience curve and buying scale: the same SKUs run for years, the learning rate is estimated more reliably and MOQ pays off. Acquisition is costlier (the audience switches brands less) but customer contribution is higher through repeats: look at 24 months, not 12. Check marginal CAC against the repeat-purchase share of basics. The scale barrier is fit and quality stability as runs grow: a pattern deviation on the sleeve shows up later in returns.
- Women
- A higher fashion share: the experience curve is weaker (complex styles, short runs) and the buying gain is capped by markdown risk. Scale comes from audience breadth and channels, not from the run of one style. CAC is higher because of competition for traffic in womenswear; customer contribution depends strongly on fit returns and the full-price share. Check that growth was not bought with discounts: marginal CAC is computed on contribution after price cuts, otherwise the break point is distorted.
Common mistakes
- Deciding on average CAC -> budget grows past the tipping point -> compute marginal CAC and customer contribution.
- Customer contribution measured by revenue -> losses on buyout and returns -> 12-month contribution after cost, returns and delivery.
- A ready-made learning rate instead of own data -> a false saving -> estimate from batches.
- Widening assortment with growth -> volume per style falls, scale effect lost -> keep depth and shared materials.
- Growth with no stop rule -> budget not revisited when bids rise -> a stop rule on marginal CAC.
Russian specifics
- Buying from large and small suppliers in Russia is governed by the Civil Code and competition rules; volume discounts and priority supply terms are set out in the contract. Rule: Civil Code part 2: Civil Code, part two (including Arts. 513, 713, 990, 1027) (rule check not completed).
- Acquisition spend is subject to the advertising law and ad labeling; include labeling and data operator costs in CAC. Rule: Law 38-FZ of 13.03.2006: Federal Law of 13.03.2006 No. 38-FZ on advertising.
- Competition restrictions (collusion, abuse of dominance) as market share grows are governed by the competition law; the risk is low for small and mid-size firms, check for large ones. Rule: Law 135-FZ of 26.07.2006: Federal Law of 26.07.2006 No. 135-FZ on protection of competition (rule check not completed).
- Brand: priority supply and volume discount terms with factories are set in the contract; competition law limits as market share grows in a niche are checked by a lawyer (unverified).
- Retailer: retro bonuses and shelf terms with brands must not make a supplier dependent on exclusive conditions; lawfulness of the terms is checked by a lawyer (unverified).