Unit Economics
LTV, CAC, payback, ARPU/ARPPU/AOV, contribution margin, and net revenue retention.
10 questions
JuniorTheoryVery commonWhat is customer LTV, the simplest defensible formula for it, and the assumption baked into that formula?
What is customer LTV, the simplest defensible formula for it, and the assumption baked into that formula?
LTV is the total contribution margin a customer brings over their whole relationship. The simplest form is LTV = ARPU × gross margin ÷ churn, since lifetime ≈ 1/churn. The buried assumption is constant churn — real retention curves flatten, so 1/churn understates long-lived users.
Common mistakes
- ✗Using gross revenue instead of contribution margin as the LTV base
- ✗Treating LTV as a single-purchase or first-order value
- ✗Forgetting the constant-churn assumption behind lifetime = 1/churn
Follow-up questions
- →How would you refine LTV when retention flattens with tenure?
- →Why discount future contribution when the horizon is long?
JuniorTheoryVery commonWhat LTV/CAC ratio is considered healthy, and what is a ratio of 1.2 really telling a founder?
What LTV/CAC ratio is considered healthy, and what is a ratio of 1.2 really telling a founder?
A healthy LTV/CAC is around 3:1 — cushion over CAC to fund overhead and growth. Above 5 often signals underinvesting; below 3 is thin. A 1.2 ratio means the customer barely out-earns cost — after overhead no real profit, so it is unsustainable despite LTV beating CAC.
Common mistakes
- ✗Reading any ratio above 1.0 as a healthy, profitable channel
- ✗Ignoring that overhead sits between LTV and actual profit
- ✗Treating a very high ratio as ideal rather than underinvestment
Follow-up questions
- →Why can a very high LTV/CAC signal a problem, not success?
- →How does payback period change the story a healthy ratio tells?
JuniorTheoryCommonWhat belongs in CAC — customer acquisition cost — and which three costs do analysts routinely leave out?
What belongs in CAC — customer acquisition cost — and which three costs do analysts routinely leave out?
CAC is the fully-loaded sales-and-marketing cost to win a customer ÷ new customers acquired that period. Beyond ad spend, add the three usually left out — salaries, discounts and promos, agency and tool fees. Ad-only CAC understates cost and flatters LTV/CAC.
Common mistakes
- ✗Counting only ad spend and omitting salaries, discounts, and fees
- ✗Dividing by all customers instead of newly acquired ones
- ✗Folding product/COGS costs into CAC, which belong to margin
Follow-up questions
- →How does a long sales cycle complicate assigning cost to a cohort?
- →Should organic signups be blended into a paid CAC?
JuniorTheoryCommonWhy can a company with a 3× LTV/CAC ratio still go bankrupt on an 18-month CAC payback?
Why can a company with a 3× LTV/CAC ratio still go bankrupt on an 18-month CAC payback?
LTV/CAC captures lifetime profit but ignores timing. Payback is the months of margin needed to recover CAC. At 18 months you front the full CAC now and wait 1.5 years to get it back; keep acquiring and outflows outrun repayments, so you exhaust working capital while profitable on paper.
Common mistakes
- ✗Assuming a strong LTV/CAC alone guarantees the company stays solvent
- ✗Confusing payback period with customer lifetime or churn
- ✗Ignoring that CAC is paid upfront while margin trickles in
Follow-up questions
- →What payback period is usually considered safe for a startup?
- →How does faster acquisition worsen a long payback's cash gap?
MiddleTheoryCommonARPU, ARPPU, and AOV are often used interchangeably — what decision is each one actually for?
ARPU, ARPPU, and AOV are often used interchangeably — what decision is each one actually for?
ARPU is revenue ÷ all users, paying and free — whole-base monetization, feeds LTV. ARPPU is revenue ÷ paying users only — payer spend, for pricing. AOV is revenue ÷ orders — per-order size, for checkout and promos. Confusing them misattributes where money moves.
Common mistakes
- ✗Treating ARPU, ARPPU, and AOV as interchangeable
- ✗Swapping the denominators (all users vs payers vs orders)
- ✗Feeding AOV into LTV where per-user ARPU belongs
Follow-up questions
- →When does a rising AOV coincide with a falling ARPU?
- →Which of the three best tracks a freemium product's health?
MiddleTheoryCommonCAC 1500 ₽, AOV 2000 ₽, margin 30%, 2.5 orders/yr, 40% annual churn — is the channel profitable?
CAC 1500 ₽, AOV 2000 ₽, margin 30%, 2.5 orders/yr, 40% annual churn — is the channel profitable?
Yes. Margin per order = 2000 × 0.30 = 600 ₽; annual contribution = 600 × 2.5 = 1500 ₽. Lifetime ≈ 1/0.40 = 2.5 years, so LTV = 1500 × 2.5 = 3750 ₽. LTV/CAC = 3750/1500 = 2.5 and payback ≈ 12 months — profitable, though just under the 3× rule of thumb.
Common mistakes
- ✗Using gross revenue instead of contribution margin as the LTV base
- ✗Counting only the first year instead of the full 1/churn lifetime
- ✗Multiplying by churn rate instead of dividing to get lifetime
Follow-up questions
- →At what CAC would this channel hit the 3× rule of thumb?
- →How would a 12-month payback affect the scaling decision?
SeniorDesignOccasionalYour LTV model uses ARPU × margin ÷ churn, i.e. it assumes lifetime = 1/churn at a single constant churn rate. But your retention curves clearly flatten with tenure — a customer active two years churns far more slowly than a fresh one, so one rate mis-states different tenures. Design the cohort-based LTV estimate you would replace 1/churn with. Describe how you would use observed cohort retention, how you would handle the long tail beyond your data window, and where a discount rate enters.
Your LTV model uses ARPU × margin ÷ churn, i.e. it assumes lifetime = 1/churn at a single constant churn rate. But your retention curves clearly flatten with tenure — a customer active two years churns far more slowly than a fresh one, so one rate mis-states different tenures. Design the cohort-based LTV estimate you would replace 1/churn with. Describe how you would use observed cohort retention, how you would handle the long tail beyond your data window, and where a discount rate enters.
Drop the constant-churn assumption. Build LTV from observed cohort retention — sum over tenure months margin × active share r(t). Extrapolate the tail with a fitted survival curve (sBG or power-law), not a fixed rate, and discount future months. This credits the sticky tail 1/churn erases.
Common mistakes
- ✗Replacing one constant churn rate with another instead of a curve
- ✗Extrapolating the tail with a fixed rate rather than a fitted survival curve
- ✗Omitting present-value discounting on a long LTV horizon
Follow-up questions
- →How would you validate the tail extrapolation against held-out cohorts?
- →When does per-segment cohort LTV beat one blended curve?
SeniorTheoryOccasionalHow does a B2B firm grow revenue while losing customers, and does net revenue retention (NRR) or the growth Quick Ratio expose it?
How does a B2B firm grow revenue while losing customers, and does net revenue retention (NRR) or the growth Quick Ratio expose it?
Expansion from surviving accounts — upsell, seats, usage — can outweigh churned-logo revenue, so revenue and NRR rise as customers fall. NRR above 100% hides logo churn by netting expansion vs churn. The Quick Ratio = (new+expansion) ÷ (churned+contraction) exposes it — low means churn drags despite growth.
Common mistakes
- ✗Reading NRR above 100% as proof the customer base is healthy
- ✗Assuming revenue and customer count must always move together
- ✗Overlooking expansion revenue masking underlying logo churn
Follow-up questions
- →What NRR and logo-retention pair would you report together?
- →How does a few large accounts expanding distort blended NRR?
MiddleDebuggingRareFinance and Product read the same data as LTV/CAC = 4 and as a loss — find the disagreement.
Finance and Product read the same data as LTV/CAC = 4 and as a loss — find the disagreement.
Finance put gross revenue in LTV's numerator; Product used contribution margin. LTV must rest on margin, not revenue — after a 25% gross margin and ~15% of revenue lost to fees, support, and refunds, net contribution is ~2000 ₽, so the true LTV/CAC ≈ 0.4, not 4. Product is right.
Open full question →Common mistakes
- ✗Building LTV on gross revenue instead of contribution margin
- ✗Assuming variable costs like fees and refunds sit inside CAC
- ✗Trusting a headline LTV/CAC without checking the numerator's basis
Follow-up questions
- →Which variable costs must the LTV numerator net out?
- →How would you standardize one LTV definition across teams?
MiddleDesignRareA one-week 30%-off promo just ended. Baseline was 100 orders at a 2000 ₽ AOV with a 50% contribution margin — 1000 ₽ of margin per order. During the promo, orders rose 60% to 160, each discounted to 1400 ₽ with unchanged COGS, so margin per order fell to 400 ₽. Marketing calls it a win — 60% more orders. Was the promo actually profitable? Compute the incremental contribution margin the extra orders added, and name the cannibalization effect you must subtract before you can answer.
A one-week 30%-off promo just ended. Baseline was 100 orders at a 2000 ₽ AOV with a 50% contribution margin — 1000 ₽ of margin per order. During the promo, orders rose 60% to 160, each discounted to 1400 ₽ with unchanged COGS, so margin per order fell to 400 ₽. Marketing calls it a win — 60% more orders. Was the promo actually profitable? Compute the incremental contribution margin the extra orders added, and name the cannibalization effect you must subtract before you can answer.
No — it lost money. The 60 truly incremental orders add 60 × 400 = 24000 ₽ of margin. But the 100 buyers who would have bought at full price now earn 400 not 1000 ₽ — that 600 × 100 = 60000 ₽ cannibalization is subtracted. Net 24000 − 60000 = −36000 ₽ — the discount to loyal buyers sinks it.
Common mistakes
- ✗Judging a promo by revenue or order volume, not incremental margin
- ✗Counting new orders' margin but ignoring cannibalized full-price buyers
- ✗Forgetting the discount compresses margin on every order, not just new ones
Follow-up questions
- →How would you estimate the true incremental (non-cannibalized) orders?
- →What discount depth would make this promo break even?
