How to Calculate LTV for Marketplaces: Formulas, Cohorts, and What Investors Want to See
Definition: Marketplace LTV = AOV × Take Rate × Purchase Frequency × Customer Lifespan × Contribution Margin %. Unlike SaaS LTV, marketplace LTV must be calculated on net revenue (the take rate), not gross merchandise value. Using GMV inflates LTV by 5-15x and produces a false-positive on unit economics. — From The Two Numbers by Lech Kaniuk
The standard LTV formula -- ARPU times gross margin times one over churn -- doesn't work for marketplaces.
It was built for SaaS businesses where one customer type pays a recurring subscription at a predictable margin. Marketplaces have none of that. You have two user types (buyers and sellers), revenue that's a fraction of each transaction, margins that shift depending on which costs you include, and retention curves that look nothing like a subscription product.
Most marketplace founders get LTV wrong in one of two ways. They either calculate it on gross merchandise volume (GMV) and present a number 10-20x too high, or they use the SaaS formula on their take-rate revenue and miss second-order effects like frequency increases and cross-category expansion. Both errors lead to the same place: a fundraise deck that falls apart under diligence. I've lived this from the operator's seat — at iTaxi, the ride-hailing company I ran as CEO, our LTV only made sense once we modeled it on take-rate revenue and real ride frequency, not gross fares.
Here's how to get it right.
The Marketplace LTV Formula
Marketplace LTV = AOV × Take Rate × Purchase Frequency × Customer Lifespan × Contribution Margin %
Each component matters and each one has a trap.
AOV (Average Order Value) is what the demand side -- the buyer -- pays per transaction. Not what you keep. What they pay. For Uber, that's the ride fare. For Airbnb, the booking total. For Etsy, the cart value. This number is the base of everything, so measure it from actual transaction data, not averages from your pitch deck.
Take Rate is the percentage of GMV your marketplace captures as revenue. This is the number that separates marketplace economics from everything else.
| Marketplace Type | Typical Take Rate | Examples |
|---|---|---|
| Product marketplaces | 5-15% | Etsy (6.5%), eBay (13%) |
| Service marketplaces | 15-30% | Fiverr (27%), Upwork (10-20%) |
| Managed marketplaces | 30-50%+ | DoorDash (~35%), Uber (~25%) |
| SaaS-enabled marketplaces | 5-15% + subscription | Shopify, Mindbody |
Higher take rates mean higher revenue per transaction but usually come with higher variable costs. Managed marketplaces like DoorDash take 35% but also handle logistics. That spread between take rate and contribution margin is where the real economics live.
Purchase Frequency is transactions per time period. This is the variable with the widest range across marketplace categories. A DoorDash customer might order 4-6 times per month. An Airbnb guest books 1-2 times per year. An Etsy buyer purchases maybe 3-5 times per year. You need to measure this per cohort over time because frequency changes -- usually it increases for retained users and the blended average drops as churned users dilute the denominator.
Customer Lifespan = 1 / Churn Rate. If your annual churn is 40%, your average customer lifespan is 2.5 years. But be careful here. Marketplace churn is harder to define than SaaS churn because there's no cancellation event. A buyer who hasn't transacted in 6 months -- are they churned or dormant? Define your churn window based on your category's natural purchase cycle, not an arbitrary 30-day cutoff.
Contribution Margin % = (Revenue - Variable Costs) / Revenue. Not gross margin. Contribution margin. This distinction kills more marketplace LTV calculations than anything else. More on this below.
Worked Example
Example: Service Marketplace (Homeowners + Cleaners)
AOV: $150 per cleaning
Take rate: 20% ($30 revenue per transaction)
Purchase frequency: 2x per month (24x per year)
Customer lifespan: 2.5 years (40% annual churn)
Contribution margin: 65%
LTV = $150 × 0.20 × 24 × 2.5 × 0.65 = $1,170
That's $1,170 in contribution profit from the average buyer over their lifetime. If your blended CAC to acquire that buyer is $200, your LTV:CAC is 5.85:1. Strong.
But change one input and the math shifts fast. If churn is actually 60% instead of 40%, lifespan drops to 1.67 years and LTV falls to $780. If contribution margin is 45% instead of 65% (because you underestimated variable costs), LTV drops to $810. Stack both errors and you're at $540 -- less than half the original number.
This is why Lech Kaniuk wrote in The Two Numbers That Build or Break Every Business that the inputs to LTV matter more than the output. One wrong assumption cascades through everything.
The Contribution Margin Trap
Gross margin and contribution margin are not the same thing and the gap between them can wreck your LTV calculation. Here are the variable costs you need to deduct from each transaction's revenue before calculating LTV:
Payment processing: 2.5-3.5% of GMV (not of your take rate). On a $150 transaction where you keep $30, payment processing eats $4.50-5.25 -- that's 15-17% of your revenue gone before you do anything else.
Customer support per transaction: allocate your support costs per ticket, multiply by the ticket rate per transaction. For most marketplaces this runs $0.50-3.00 per transaction.
Fraud and dispute costs: chargebacks, refund fraud, identity verification. Ranges from 0.5% to 3% of GMV depending on category.
Insurance: Airbnb's Host Guarantee, Uber's rider insurance. These aren't optional in regulated categories.
Refund and cancellation costs: the full cost of a cancelled transaction isn't zero -- you've already processed the payment, notified the supplier, possibly allocated inventory.
Delivery and fulfillment: for managed marketplaces (DoorDash, Instacart), this is the single largest variable cost.
Per-transaction technology costs: API calls, mapping services, SMS notifications, real-time pricing calculations.
Using gross margin instead of contribution margin
Marketplaces often hit positive gross margins well before positive contribution margins. The gap can be 12-18 months. A marketplace showing 70% gross margins might have 35% contribution margins once you include all per-transaction costs. Using gross margin in your LTV formula doubles the real number.
Founders have walked into Series A meetings with LTV:CAC ratios of 8:1 that collapsed to 2.5:1 when the investor's associate rebuilt the model with actual contribution margins. That's not a rounding error. That's the difference between "we're scaling" and "we need to fix unit economics first."
Three Approaches to Calculating Marketplace LTV
Approach 1: Simple Historical Average
Total Cumulative Revenue from a Cohort / Number of Users in That Cohort
Take everyone who signed up in Q1 2024. Add up every dollar of revenue (your take rate, not GMV) they've generated through today. Divide by headcount. This is quick. It's also wrong in almost every useful way. It treats all users as identical, ignores the time dimension, and can't project forward. Use it as a sanity check, not a planning tool.
Approach 2: Cohort-Based Analysis (The Gold Standard)
This is what you actually want. Here's how to do it step by step:
Step 1: Define your cohorts
Group users by their acquisition month (or week, if you have the volume). January 2024 cohort, February 2024 cohort, and so on.
Step 2: Track cumulative revenue per user
For each cohort, calculate the average cumulative revenue per user at month 1, month 2, month 3, and so on. Revenue means your take-rate revenue, not GMV.
Step 3: Plot the retention curve
Chart what percentage of each cohort is still active at month 1, 2, 3, etc. "Active" means completed at least one transaction. You'll see the classic L-shaped curve -- steep early drop-off, then flattening.
Step 4: Fit a curve to project forward
The most common models are logarithmic (cumulative revenue = a * ln(t) + b) or shifted geometric (retention at month t = a / (1 + b*t)). Fit to your actual data points and extend. If you have 12 months of data, you can reasonably project to 24-36 months. Beyond that you're guessing.
Step 5: Multiply by contribution margin
Apply your contribution margin percentage to the projected cumulative revenue. That's your LTV.
You need a minimum of 6-12 months of cohort data to make this reliable. Less than 6 months and you're fitting a curve to noise. If you're pre-product-market-fit, the curve is moving between cohorts anyway, which means older cohort data may not predict newer cohort behavior.
The thing that separates good marketplace operators from everyone else is comparing cohort curves. Are newer cohorts performing better or worse than older ones at the same point in their lifecycle? If your October 2024 cohort has higher month-3 cumulative revenue than your April 2024 cohort did at month 3, your marketplace is improving. If the curves are deteriorating, no amount of growth spending fixes that.
Approach 3: Predictive Models
Upwork has publicly discussed predicting a freelancer's LTV within 24 hours of signup, using behavioral signals from their first session: profile completeness, response time to first invitation, bid quality, category selection. Fiverr uses similar early-signal models to allocate marketing spend before cohort data matures.
This is where the field is heading. Machine learning models that ingest first-session behavior, demographic data, acquisition channel, and marketplace state (supply density in the user's category/location) to predict 12-month value. The best marketplace data teams can predict LTV with 70-85% accuracy within the first week.
For most startups, this is aspirational. But knowing it exists shapes how you should think about data collection from day one. Track every behavioral signal. You'll need it eventually.
Supply-Side vs. Demand-Side LTV
Calculate LTV on the demand side. Buyers generate your revenue. The LTV:CAC ratio that determines your growth economics is a demand-side calculation.
But ignoring the supply side is how marketplaces die slowly. Track these supply-side metrics as operational health indicators:
Supply retention rate. What percentage of active suppliers are still active 3, 6, 12 months later? Uber's driver retention has historically been brutal -- around 50% at 12 months in many markets. That means Uber needs to constantly recruit new drivers, which is a cost that shows up in demand-side LTV through longer wait times and surge pricing.
Supply lifetime GMV facilitated. How much total transaction volume does the average supplier generate? A top Airbnb host might facilitate $50,000+ in annual bookings. Losing that host doesn't just remove supply -- it removes high-quality, well-reviewed supply that drives demand retention.
Cost to replace a churning supplier. Recruiting, onboarding, background checks, initial incentives. For Uber this runs $1,000-3,000 per driver in mature markets. For Airbnb a new host takes months to build reviews. Replacement cost is the hidden tax on supply churn.
Share of wallet. What percentage of a supplier's total business runs through your platform? An Etsy seller doing 90% of their business on Etsy is locked in. An Upwork freelancer doing 20% of their work on Upwork will leave when they build enough direct client relationships. Share of wallet predicts supply churn better than any satisfaction survey.
Supply collapse is a lagging indicator
A marketplace with high demand-side LTV but collapsing supply retention is a ticking time bomb. The demand-side numbers look great until supply thins out, match quality drops, and buyer retention follows supply out the door. By the time it shows up in your demand LTV, you're 6-9 months behind.
How Liquidity and Network Effects Show Up in LTV
If your network effects are working, you should see LTV increasing over time as cohorts mature and as the marketplace scales. Specifically:
Lower incentive costs. Early cohorts needed $20 referral credits and 50% discount codes to convert. As the marketplace builds liquidity, newer cohorts convert with less subsidy. That shows up directly in CAC but also in LTV through better contribution margins (fewer per-user promotions eating into revenue).
Better retention. A marketplace with 10,000 suppliers retains buyers better than one with 500 suppliers because match quality improves. If your month-6 retention rate is 35% at 5,000 supply units and 48% at 20,000 supply units, network effects are real and measurable.
Improving pricing power. As you become the default place buyers go for a category, your ability to increase take rates grows. Airbnb's take rate has crept upward over the years. Etsy raised transaction fees from 3.5% to 6.5%. That pricing power flows straight into LTV.
Sarah Tavel at Benchmark describes this as a three-level hierarchy of engagement:
Level 1: Growing engaged users
Users complete their core action (book a stay, hire a freelancer, order food). At this level, LTV is driven purely by transaction economics.
Level 2: Accruing benefits, mounting losses
Users build up assets on the platform -- reviews, reputation, saved preferences, purchase history, loyalty credits. The longer they stay, the more they'd lose by leaving. LTV starts compounding because retention improves as tenure increases.
Level 3: Self-perpetuating
Users generate content, reviews, and referrals that attract new users. The marketplace grows with less paid acquisition. LTV improves because CAC drops and organic users tend to retain better.
As your marketplace moves up Tavel's hierarchy, LTV should compound. If it doesn't -- if your newer cohorts aren't performing better than older ones at the same lifecycle stage -- your network effects may be weaker than you think.
The leading indicator of all of this is liquidity, measured as fill rate or match rate. What percentage of buyer searches or requests result in a completed transaction? If your match rate is 60% and climbing, LTV will follow upward. If match rate stalls or drops, LTV will follow it down, usually with a 2-3 month lag. Watch match rate weekly. It tells you where LTV is going before LTV gets there.
What Investors Want to See
The marketplace founders who close quickly bring specific metrics. The ones who struggle show up with blended averages and hand-waving. Here's what Series A and Series B investors actually ask for:
Cohort-level LTV:CAC, not blended. A blended LTV:CAC of 4:1 means nothing if your paid acquisition cohorts are at 1.5:1 and your organic cohorts (who would have come anyway) are at 15:1. Break it out by channel, by cohort vintage, and ideally by geography or category.
LTV improvement trajectory. Are your newer cohorts worth more than older cohorts at the same lifecycle stage? Investors call this "improving vintage curves." It's the single strongest signal that network effects are working. Flat or deteriorating vintages at scale are a red flag.
Separate demand-side and supply-side metrics. Demand LTV:CAC for the growth model. Supply retention, supply CAC, and supply satisfaction for the operational model. Smart investors know a marketplace can have great demand economics and still fail on supply.
Share of wallet on both sides. What percentage of a buyer's total category spend goes through your marketplace? What percentage of a seller's business? Higher share of wallet means higher switching costs, better retention, and more defensible LTV.
Dollar retention (net revenue retention). Are existing users spending more over time? A net revenue retention of 115% means your existing user base generates 15% more revenue this year than last year, before you add a single new user.
Established framework alignment. a16z published their 16 marketplace metrics. Point Nine Capital's marketplace napkin sets baseline expectations for each stage. Speedinvest published Series A marketplace benchmarks by category. Investors know these frameworks.
Investor Benchmarks
LTV:CAC of 3:1 or better at Series A. Below 3:1, you need a clear story about why it improves.
CAC payback under 12 months. Preferably under 6. Marketplace transactions are lower-margin than SaaS subscriptions, so payback periods can stretch. Anything over 12 months requires serious capital reserves.
Net revenue retention above 100%. This means existing users are growing, not just not churning. The best marketplaces run 110-130%.
As Lech Kaniuk covers in The Two Numbers That Build or Break Every Business, these benchmarks aren't arbitrary -- they reflect the minimum economics required for a marketplace to fund its own growth without burning through capital faster than it generates returns.
The Disintermediation Risk
Here's the thing that can make your entire LTV model fiction: if buyers and sellers can easily transact off-platform after they find each other, they will.
A freelancer found on Upwork has every incentive to take the relationship off-platform once trust is established. The client saves 10-20% in fees. The freelancer earns more per hour. Both win -- except Upwork, which loses all future LTV from that match.
Multi-tenanting dilutes LTV
When users are active on multiple competing platforms simultaneously (a driver on both Uber and Lyft, a host on both Airbnb and Vrbo), your share of wallet is lower, retention is weaker, and LTV projections built on single-platform behavior overestimate reality by 20-40%.
The marketplaces that sustain high LTV have built specific defenses against disintermediation:
Trust infrastructure. Airbnb's review system, host verification, and guest identity checks create trust that doesn't transfer off-platform. A host with 200 five-star reviews on Airbnb starts from zero on a direct booking site.
Integrated payments and escrow. Fiverr and Upwork hold funds in escrow until work is delivered. Moving off-platform means the buyer loses payment protection and the seller loses guaranteed payment.
Insurance and liability coverage. Uber's $1M per-ride insurance policy, Airbnb's Host Guarantee -- these protections disappear the moment a transaction moves off-platform.
Reputation portability (or lack of it). Your 4.95-star rating on Uber is worthless on Lyft. Your 500 five-star reviews on Etsy don't come with you to your own Shopify store. Every review a user accumulates is a sunk cost that increases switching costs.
If your marketplace lacks these defenses -- if discovery is the only value you provide -- your LTV projections need a disintermediation discount. Undefended marketplaces lose 30-50% of theoretical LTV to off-platform leakage. Build that into your model or get surprised by it later.
Putting It Together
Your marketplace LTV divided by your two-sided CAC gives you the ratio that determines whether your marketplace will scale or stall. Get either number wrong and the ratio is meaningless.
The formula is the starting point. Cohort analysis makes it real. Contribution margins make it honest. And tracking supply-side health makes sure the number stays real six months from now. As Lech Kaniuk wrote in The Two Numbers: know your actual LTV, know your actual CAC, and let the ratio tell you whether to step on the gas or fix the engine.
Frequently Asked Questions
Related
Go Deeper
This post covers the basics. "The Two Numbers That Build or Break Every Business" by Lech Kaniuk includes:
- The complete marketplace LTV methodology with real company examples
- Cohort analysis frameworks for two-sided marketplaces
- The contribution margin audit to find hidden variable costs
- How to present marketplace unit economics to investors
This article gives you the marketplace model. The book shows how to connect LTV, CAC, payback, and strategy in one operating system.
Still Building — a newsletter for founders
One email per week. Unsubscribe anytime.
Written by Lech Kaniuk, author of "The Two Numbers That Build or Break Every Business."