+150 XP

Embedded distribution and partner-led acquisition

# Embedded distribution and partner-led acquisition

In August 2020 Klarna agreed to appear at checkout for Shopify's US merchants, and handed Shopify warrants over roughly a tenth of Klarna's equity as part of the arrangement. Around the same period Shopify was rolling out its own instalment option, Shop Pay Installments, underwritten by Affirm, inside the checkout flow it controls. Klarna got distribution no ad budget could buy. Shopify kept the surface, the default setting, and the right to change its mind.

That is partner-led acquisition in one paragraph: volume you did not pay for, on real estate you do not own, governed by a counterparty whose interests only partly overlap with yours.

Why embedded changes the demand economics

Paid acquisition means buying attention from people who were not thinking about a financial product. Embedded distribution puts the product in front of someone mid-decision: the shopper at checkout, the driver checking today's earnings, the merchant staring at a cash-flow gap. You are intercepting demand rather than creating it, and the conversion gap is an order of magnitude, not a few points.

Klarna reaches on the order of 100 million active consumers through several hundred thousand merchant partners. Very few arrived from a Klarna campaign. They arrived because a shop they were already buying from put a button in front of them. Cost the channel using the allocation method the CAC lesson sets out; what matters here is where the volume comes from.

Four surfaces where this happens:

  • Checkout: instalments at the point of purchase, once the buying decision is made.
  • Super apps: Grab Financial Group sells lending, insurance and payments to drivers, merchants and riders who opened the app for a ride or a meal, and holds digital bank licences in Singapore and Malaysia to do it.
  • Platform data: Shopify Capital advances cash to merchants priced off transaction history Shopify already holds. No lead gen, no application funnel.
  • Device and wallet: Apple puts card and payment features in front of hundreds of millions of users through software already on the phone.

The trade-off

The volume is cheap and it is not yours. The distributor owns the brand the customer remembers, the placement and usually the relationship afterwards. Your job moves from buying media to winning, defending and tuning placement inside someone else's product. The credibility a money product needs is still required, but here much of it is borrowed from the partner's brand, which is why partner selection carries more weight than creative.

The BaaS layer that makes it possible

Embedded finance runs on BaaS (Banking as a Service): infrastructure that lets a non-bank offer banking or payment features by plugging into a licensed bank through APIs.

Three roles:

1. The distributor, the brand the customer sees: the retailer, the super app, the handset maker.

2. The BaaS provider or program manager, the technology layer joining everything.

3. The chartered bank, the licensed institution holding deposits or issuing credit. Regulation requires a real bank somewhere in the stack.

Apple Card shows the split cleanly: Apple owned the surface, the design and the customer's mental model of the product, while Goldman Sachs held the licence, the balance sheet and the servicing obligations.

For how these layers fit together, the Federal Reserve's material on banking-fintech partnerships is a reasonable starting point.

Evaluating a BaaS or distribution partner

Reach and fit

Does the partner's audience match the product? Instalment lending wants high-frequency retail checkout. Working-capital lending wants platforms full of small-business owners. A hundred thousand merchants with live transaction data are worth more than ten million casual app users.

Placement quality

Ask for the exact screen, the position on it, and whether the option is default-on or opt-in. Being one of four logos under a card form is a different business from being pre-selected. Ask also what happens when the partner tests a competitor in the same slot, because it will.

Compliance posture

Regulators have sharpened scrutiny of bank-fintech arrangements, particularly around KYC (Know Your Customer, the identity-verification rules), dispute handling and consumer protection. The cost lands unevenly. In October 2024 the CFPB fined Goldman Sachs $45 million with roughly $20 million in consumer redress over Apple Card servicing and dispute failures, and fined Apple $25 million too. Owning the customer surface can mean owning the enforcement action, even when another firm runs the operations.

The blunter failure mode: a bank partner under a consent order freezes new programme onboarding, your channel stops taking applications overnight, and you have no ad budget standing by to absorb the shortfall.

Data and control

Who owns the customer? Who may market to them later, through which channel? What data comes back, at what granularity, how fast? A placement that converts well but returns nothing beyond a settlement file caps lifetime value at the first transaction.

Structuring rev-share economics

Common revenue sources

  • Merchant fees per transaction for instalment products, typically well above card acceptance cost. The bulk of Klarna's revenue comes from merchants rather than consumers.
  • Interchange on embedded debit and credit cards, frequently shared with the distributor.
  • Interest and lending margin on embedded credit, plus late fees where the jurisdiction permits them.
  • Platform fees paid by smaller distributors that want finance features they cannot build.

How the split usually works

Rev-share runs in both directions. The fintech pays the distributor for audience access, which is the norm when the distributor is a large retailer or a platform with millions of merchants. The distributor pays the fintech for licence and infrastructure when it is a small platform adding a payments feature.

There is no standard percentage. Whoever holds the scarce asset (audience, licence, underwriting skill) captures more of it. Scarcity moves: as more providers offer an interchangeable instalment product, the audience becomes the scarce side and renewal terms worsen for the fintech every cycle.

The unit-economics check

Contribution per customer =
    (Revenue per customer x your share)
  - (Integration and servicing cost)
  - (Expected losses, for credit products)
  - (Compliance and support cost)

Then compare to:
  Embedded CAC (the cost to win the placement, prorated per customer)

Two things break this model in practice. Credit losses: a channel converting impulse purchases hands you a different risk mix than one gated behind an application, and a few hundred basis points of extra loss erases a thin revenue share. Concentration: if one partner drives most of your new customers, your marketing plan is really a contract renewal, and the revenue should be discounted accordingly.

Knowledge check

1. What is the fundamental reason embedded distribution can achieve a lower CAC than traditional paid acquisition channels?

2. A BNPL button at an e-commerce checkout illustrates which core principle of embedded distribution?

3. How does a marketer's job change when shifting from direct-to-consumer acquisition to embedded distribution?

MULTIPLE CHOICE

4. Select ALL correct answers about the tradeoffs of embedded distribution compared to traditional acquisition.

Select all the correct answers.

MULTIPLE CHOICE

5. Select ALL correct answers that are valid examples of embedded distribution as described in the lesson.

Select all the correct answers.

Winning and keeping placement

Make integration painless

Distributors pick the partner that creates least work: clean APIs, high approval rates, low fraud, few support tickets landing in their inbox. Reliability is the pitch.

Prove you lift the partner's metrics

The strongest claim is about their P&L: higher checkout conversion, larger baskets, better repeat rate. Bring the measurement design as well (holdout or geo split), because a distributor that cannot attribute the lift will file you under cost.

Optimise the moment, not the campaign

Your creative surface is a few dozen pixels plus a cost disclosure. Button wording, position relative to the pay-now option, and how plainly total cost appears move more volume than any brand campaign, and the wording has to stay inside the rules the compliant-messaging lesson sets out.

Defend the relationship

Placement is rented. Distributors renegotiate, dual-source, or build in-house, and exclusivity is what they sell most dearly. Klarna's 2025 agreement to become the exclusive instalment provider in Walmart's OnePay app displaced Affirm from that slot, which is a fair summary of the category: one contract can move a large share of a competitor's volume without either brand buying a single consumer ad.

When embedded is the wrong channel

Products needing an advisory relationship, complex wealth management or business banking with a named contact, do not build one in a two-second checkout tap. If your strategy depends on owning the brand and the data, handing both to a distributor buys volume now and forecloses direct economics later. Embedded suits high-intent, low-friction products where context does the persuading, and distributors whose traffic is large enough that a modest attach rate is still a real business.

Key takeaways

  • Embedded distribution intercepts demand inside someone else's product instead of creating it, which is why the conversion economics differ by an order of magnitude from paid channels.
  • The BaaS stack splits brand from licence: Apple owned the Apple Card surface while Goldman Sachs carried the servicing obligations, and the CFPB still fined both in October 2024.
  • Screen partners on intent fit, exact placement and default status, compliance posture and data rights. A bank partner frozen by regulators takes your channel with it.
  • Model contribution per customer against the cost of the placement, and stress-test credit losses and partner concentration, since either can invert a cheap-volume deal.
  • Placement is rented, not owned. You keep it by being easy to integrate and by proving lift in the partner's own numbers, and you lose it to whoever offers exclusivity next.