पाठशाला Pathshala · ग्राहक Grāhak, The customer · Lesson 30 · Scale
B2B2C: serving the customer and the customer’s customer
In B2B2C the partner signs and the end user decides whether it works. Give the partner a reason to push, the end user a reason to stay, and write down which wins when they collide.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
A B2B2C company signs its contract with one customer and earns its keep from another. Most of them fail by serving only the one who signs: the partner is delighted at launch and the end users never come back.
This lesson sets out how to design for both at once: the two jobs and the two numbers, who owns the end-user relationship, the two layers of customer success, a revenue share set by what it does to the partner’s behaviour, and a rule for the day the partner’s needs and the end user’s collide. A figure shows why the share that feels fair is often too small.
Two customers, two jobs
Alex Rampell of Andreessen Horowitz defined the model in On B2B2C Business Models: a company sells to a business and, in doing so, gains customers it can keep. His examples include Affirm, Instacart and OpenTable. The attraction is acquisition without a per-customer marketing cost, and contracts that make revenue more predictable than consumer advertising. He adds that it works best when the partner does not want to be in the business you offer: lending, collections, compliance.
In India the pattern is everywhere. A school buys a learning app its students use. A dealer offers a loan from a lender through a lending app. A bank offers an insurer’s policy. An employer buys a health benefit its employees use. In each the partner and the end user hire you for different jobs. The school wants better results and fewer parent complaints; the student wants to pass Thursday’s test. The dealer wants more vehicles sold this month; the buyer wants an instalment they can pay.
Write both jobs down with the number that proves each is done. For the partner it is usually their own metric: sales, retention, cost to serve, compliance. For the end user it is activation and repeat use. The lesson on [who pays, who uses and who decides](/library/who-pays-who-uses-who-decides) maps the roles in a single deal; in B2B2C there are two deals stacked, and both have to close.
Who owns the end-user relationship
Rampell is blunt about the tension. The partner is “justifiably paranoid” that you will do other things with its most valuable asset, its customers. And unless end users see themselves as your customers, you cannot count them as yours: a clause in the terms of service does not make it so. Settle three things in the contract before launch: whose brand the end user sees, who may contact the end user and about what, and who holds the data and for what purposes.
In regulated sectors the regulator settles some of it for you. In digital lending the Reserve Bank’s 2022 guidelines, since consolidated into the Digital Lending Directions of 8 May 2025, require repayments to go directly into the lender’s bank account rather than through a lending service provider’s pool account, require the lender rather than the borrower to pay the service provider’s fees, and require a Key Fact Statement to the borrower before the contract. The partner’s app may own the experience; it cannot own the money. Checked October 2026.
Financial data moves the same way under the Account Aggregator framework: as Sahamati explains, an aggregator cannot share a person’s data without their consent, and the person can revoke it at any time. Whatever the commercial agreement says about data, the end user’s consent is the ceiling. The lesson on the [India Stack](/library/india-stack-as-market-map) maps where else this applies.
Two layers of customer success
Signing a partner is the start, not the sale. Rampell calls the fact that there are two layers of customer success, one for the partner and one for its customers, arguably the most crucial reason B2B2C partnerships fail. A partner that has signed but whose staff do not mention you at the counter produces nothing.
The partner layer: train the people who face the end user, give them a one-line pitch and an answer to the three questions customers ask, and pay or recognise them in a way their manager can see. Give the partner a dashboard in their own metric, not yours: the school sees parent complaints and test scores, the dealer sees vehicles financed. The end-user layer: onboarding, support and reminders in the end user’s language, under whichever brand the contract settled, with a route back to you when the partner cannot help.
Set the share by what it makes the partner do
Founders set the partner’s share by what feels fair or by what the first partner asked for. The useful question is different: what share makes the partner push hard enough that the end-user base grows? A partner earning a trivial amount per customer has no reason to train staff or change a counter process. Above some level, each extra point of share adds active users faster than it costs you margin; past another, it simply gives money away.
With twenty thousand eligible customers, a ₹199 monthly price and a 20 per cent share, about 8 per cent of the partner’s customers become active and you keep about ₹2.6 lakh a month. On this illustrative response your own revenue peaks near a 35 per cent share, at about ₹2.9 lakh, because the partner’s extra effort more than pays for the extra points. Raise the price to ₹499 and your revenue rises while active users fall by about two-fifths; a partner whose job was reach for its customers may not accept that trade. The shape is the argument: the share that maximises your revenue is usually higher than the one you would have offered. Learn the real response from your first three partners by offering different structures, then standardise.
In B2B2C the partner signs the contract and the end user renews it. Design so that the partner wins only when the end user does.
When the two customers collide
They will. The partner asks you to raise the end-user price because their margin is under pressure. The partner wants its own logo on everything and yours nowhere. The partner wants a feature that helps its sales team and confuses its customers. The partner wants end users’ data for a purpose the end user did not consent to.
Write the rule before the first conflict: protect the end user’s outcome, and pay the partner in ways that do not degrade it. A floor on the end-user price and a ceiling on the partner’s mark-up go in the contract. Features that serve the partner ship only if they do not lower end-user activation or repeat use. Data use follows consent. The rule is not altruism. A partner keeps renewing a product its customers use, and drops one its customers ignore, whatever the margin.
A worked example: a learning app sold through schools
A company sells a practice and revision app to private schools in Tier 2 cities, priced per student and paid by parents through the school. In year one it signs 120 schools and 40,000 students. Three months in, 22 per cent of students use it weekly; the rest never logged in. The schools are content because they were paid a share at enrolment. The parents are not, and renewals look poor.
The company changes three things. It moves the school’s share from enrolment to students active in the term, so the school earns only when students use the app. It gives teachers a weekly class report in the school’s own terms: which chapters the class is weak in before the unit test. And it sends parents a short weekly message in their language with what their child practised. By the second term weekly use reaches 41 per cent, and the schools that pushed hardest earn more than under the old structure.
The monthly two-sided review
One hour a month with partnerships, product and customer success, with two columns on every slide. Partner column: partners signed, partners active (meaning their staff produced at least one end user this month), partner earnings and the partner’s own metric. End-user column: activation, repeat use and support tickets by partner. Find the mismatches: partners with high sign-ups and low end-user use are pushing the wrong customers or selling the wrong promise; partners with high use and low sign-ups need a better staff incentive. Settle one conflict against the written rule and record the decision.
Once a quarter, redraw the revenue-share curve with what partners actually did, and change the standard terms only on that evidence. Visit two partners in person each quarter, one of the best and one of the worst, and watch how their staff mention you to a customer. The difference between them is usually a sentence, a screen or an incentive, and it is cheaper to copy than to discover any other way.
The figure and the example are illustrative. Regulatory points were checked in October 2026; nothing here is legal advice.
Sources
- Alex Rampell, On B2B2C Business Models, Andreessen Horowitz (2018) — Definition; works best when the partner does not want to be in your business; the partner is justifiably paranoid about its customers; two layers of customer success.
- Reserve Bank of India, Guidelines on Digital Lending (2 September 2022), consolidated into the Digital Lending Directions, 8 May 2025 — Repayment directly to the lender’s account; LSP fees paid by the lender, not the borrower; Key Fact Statement before the contract. Checked October 2026.
- Sahamati, What is an Account Aggregator? — No sharing without the individual’s consent; consent can be revoked at any time.