RevenueOS for platforms and marketplaces
Every marketplace founder is very good at one side of the business, usually the side they came from. But a marketplace doesn't grow by adding supply or demand. It grows by matching them, and you can add a great deal of one side without the matching improving at all.
Summary
- A marketplace doesn't grow by adding supply or demand. It grows by matching them, which means running the same five pillars twice, once for each side, and then managing a third thing neither side owns: whether they actually find each other.
- Plan both sides in matched units, at the level where the matching actually happens. A platform can be oversupplied nationally and short in the three cities that produce most of the demand, and the national number will show everything looking fine.
- GMV measures how much moved through your platform. It doesn't measure how much of it was yours, what it cost you to make happen, or whether any of it will happen again.
Every marketplace founder we meet is very good at one side of their business.
Usually it's the side they came from. A founder out of the industry can bring supply: the sellers, the drivers, the clinics, the manufacturers. A founder out of consumer tech can bring demand. They build the side they know, it grows nicely, and then it stops working.
It stops because a marketplace doesn't grow by adding supply or demand. It grows by matching them. And you can add a great deal of one side without the matching improving at all.
That's what makes this the hardest of the business models to run. You're operating two businesses that have to grow at roughly the same pace, in the same places, at the same times, and neither of them is yours.
Different dials, same system
As always, what's genuinely different is narrower than it seems.
A marketplace runs the same five-part system as everyone else (planning, demand, execution, review, predictability). What changes is that you have to run it twice, once for each side, and then manage a third thing that neither side owns: whether they actually find each other.
Planning: plan both sides, in the same units
Most marketplace plans we're shown are a demand plan with a supply footnote.
Plan both properly, and plan them in matched units. Not "add 500 sellers and grow orders 40%", but how many sellers, in which city, with what inventory or availability, are needed to serve the orders you're forecasting in that city.
Marketplace balance is local. A platform can be oversupplied nationally and short in three cities that produce most of the demand, and the national number will show everything looking fine. Plan at whatever level the matching actually happens: city, pin code, category, time slot.
And plan the ratio, not just the totals. How many active sellers per active buyer does your model need? What happens to match rate when that ratio moves? Very few marketplaces can answer that, and it's the core question of the business.
Demand: two engines, different economics
You're acquiring both sides, and they cost different things.
Supply is usually harder to acquire, cheaper to retain, and slower to replace. A good seller took months to sign and took months more to become productive.
Demand is usually easier to acquire, faster to , and expensive to keep buying.
Which means the two sides need different playbooks, different teams, and different measures. A single "growth" function measured on one blended number will always over-invest in whichever side is easier this quarter.
Also worth tracking: what share of each side is actually active? Registered sellers are not sellers. Downloaded apps are not buyers. Most marketplace dashboards are inflated with people who signed up once. Measure activity in the last 30 days and the picture usually changes substantially.
Execution: matching is the whole job
This is where marketplaces look nothing like a sales business. There's no deal to close. The job is to make sure that when someone arrives wanting something, they find it, book it, and have a good experience.
The numbers that matter here:
Match rate. What share of demand actually gets fulfilled. Every unfulfilled search is a customer learning that your platform doesn't work for them, and the second-order cost is far larger than the lost transaction.
Time to match. How long it takes. In some categories this is the entire product.
Fulfilment quality. Cancellations, no-shows, quality complaints, disputes. These damage both sides at once, which is what makes them more expensive on a marketplace than in a normal business.
Repeat rate on both sides. A seller who lists once and never returns, and a buyer who orders once and never returns, are the same failure with different names.
The take rate question
This is the pricing decision that defines your business, and it has four parts.
Who pays? Supply, demand, or both. This isn't only an economics question. It changes who feels like your customer and who feels like your cost.
How much? Enough to build a business, low enough that neither side is looking for a way around you.
Does it move? A rate that tapers as a seller grows rewards your best supply and makes leaving expensive. A rate that's flat regardless of volume quietly pushes your largest sellers toward doing it themselves.
What justifies it? This is the question that decides whether your survives. A rate you charge purely for access is the most vulnerable rate there is. A rate justified by services (payments, logistics, quality guarantees, financing, demand generation, dispute resolution) is defensible, because leaving costs the seller something real.
Which brings up the risk every marketplace lives with. Once the two sides know each other, what stops them transacting directly? The fix is never enforcement. It's making the platform genuinely more useful than the direct relationship.
If the honest answer is "nothing, they just haven't thought of it", you don't have a business model, you have an introduction service.
Subsidies: the question nobody asks early enough
Almost every marketplace subsidises one or both sides to get started. That's reasonable.
What's not reasonable is doing it for four years without ever defining the exit.
Three questions to answer before you start a subsidy, not after:
- What is it buying? Liquidity in a specific city, supply in a specific category, a habit in a specific customer group. "Growth" is not an answer.
- What does success look like? The measurable state at which the subsidy is no longer needed, usually a match rate or a repeat rate at which the market works on its own.
- When do we taper, and by how much? With a date attached.
A subsidy without those three answers stops being an investment and becomes a permanent cost that everyone has learned to call growth.
Review: the rhythm
- Weekly: active supply and active demand by city or category, match rate, unfulfilled demand, cancellations, subsidy spend.
- Monthly: contribution per transaction after everything, which means payments, support, fraud, logistics and incentives. Take rate actually realised versus your published rate. Repeat rates on both sides. Cost to acquire each side.
- Quarterly: take rate and packaging, subsidy taper, city and category expansion, and the services you're adding to justify the rate.
That second line in the monthly is the one most marketplaces skip. Your published take rate and your realised take rate are different numbers, once you count seller incentives, buyer coupons, payment costs and waived fees. The gap is usually large.
Predictability: what moves before GMV does
Match rate by city and category, which predicts demand churn before the demand actually churns.
Active supply growth against active demand growth. When these diverge, you have a problem forming that won't show up in GMV for a quarter.
Repeat rate on both sides, by cohort.
Contribution per transaction, which tells you whether growth is making the business better or just bigger.
Subsidy as a share of GMV, trending. If it isn't falling, the market isn't maturing. You're just buying transactions.
GMV is the number that lies
One direct word, because it's the most misused metric in Indian startups. GMV measures how much moved through your platform. It doesn't measure how much of it was yours, what it cost you to make happen, or whether any of it will happen again.
A platform doing ₹100 crore GMV at a 6% take rate, with 5% going out in incentives and payment costs, is a ₹1 crore business with a very impressive headline.
Report net revenue and contribution per transaction internally. Keep GMV for the contexts where it's genuinely the right measure, and don't let it drive your decisions, because it will reliably point you toward growth that costs more than it's worth.
The dashboard
Active supply and active demand, the ratio between them by city and category, match rate, time to match, GMV, net revenue, take rate published and realised, contribution per transaction, subsidy as a share of GMV, repeat rate on both sides, acquisition cost for each side, and cancellation and dispute rates.
Where to start
Pick your three largest cities or categories. For each, work out your match rate and your contribution per transaction after every cost.
You'll almost certainly find they're very different from one another, and that the market you thought was your best is the one being carried by a subsidy you stopped questioning a long time ago.
How useful was this article?
One tap. It tells us what to write more of.
The next one
Get what we publish next, by email.
Working notes on raising, borrowing, protecting, growing and structuring capital in India. One email a week at most, and you can leave any time.
We use your address only to send this. See our privacy policy.
We store your address to send you these emails and nothing else. See our privacy policy.
Related reading
