What Is Embedded Finance? A Strategy Guide for Non-Financial Companies Deciding Whether to Build, Partner or Wait

Most writing on the subject is produced by the companies selling the infrastructure behind that button, so it tends to assume the answer is yes. This guide is written from the other side of the table: for founders, product owners and commercial leaders at companies that are not financial institutions and are trying to decide whether a financial product belongs inside their customer journey at all — and if it does, whether to build it, partner for it, or deliberately wait.

A working definition

Embedded finance is a financial product — a payment, a loan, an account, a card, an insurance policy — offered inside a non-financial company's own product, at the moment the customer needs it, under that company's brand or alongside it. The defining feature is not the technology. It is a split of roles: the company that owns the customer relationship is usually not the company that holds the licence, the balance sheet or the regulatory liability.

That split is the whole subject. Every strategic question about embedded finance — what you earn, what can go wrong, what you can walk away from — comes down to who sits in which role and what the contract between them says.

Three parties sit behind every embedded finance product

Even when the customer sees one screen, there are almost always three parties behind it.

• The brand — you. You own the customer, the moment of need and the data about how that customer behaves. You usually do not hold a financial licence.
• The licensed provider — a bank, an electronic money or payment institution, a lender or an insurer. It holds the licence, carries the regulatory obligations (anti-money-laundering, consumer protection, capital) and often the money or the credit risk itself.
• The enabler — the API and operations layer that connects the two: onboarding flows, ledgers, card issuing, decisioning, reporting. This is what the industry calls Banking-as-a-Service when the provider is a bank. Sometimes the enabler and the licensed provider are the same company; often they are not.

The arrangement works when each party is paid for the thing it is genuinely best at: you for distribution, the provider for regulated risk, the enabler for plumbing. It fails when one party ends up carrying a cost the contract did not price — most often the brand discovering that it is the one answering customer complaints, chasing failed payouts and explaining frozen funds, with none of the controls that would let it fix anything.

That failure is not hypothetical. When the US middleware provider Synapse collapsed in 2024, end users of several consumer apps built on top of it found their balances frozen for months while the banks, the middleware and the apps disputed whose ledger was correct. The apps had the customer relationship and the reputational damage; they did not have the records. Any embedded finance plan should be stress-tested against that scenario before it is signed.

The five forms, and what each really asks of you

"Embedded finance" covers products with very different economics and very different obligations. Treating them as one decision is the most common planning error.

The pattern across the table: the revenue column is the one vendors lead with, and the right-hand column is the one that decides whether the product survives its second year. A useful exercise is to fill that right-hand column in for your own business, in your own words, before any vendor demo — and to notice which items nobody in the room currently owns.

The economics: what a "yes" actually pays

Take the wholesale distributor from the opening. The figures below are illustrative, but the structure is the one that matters.

The portal handles €40 million of orders a year. Around a third of buyers ask for terms, and today the sales team grants them case by case, carrying the receivable on the distributor's own books. An embedded B2B financing partner offers to take that over: approved buyers get 30–90 days, the distributor is paid within a few days less a fee, and the partner carries the default risk on approved buyers.

• The visible line: the partner offers the distributor a small share of the fee it charges. On €13 million of financed volume, even a generous share is a modest number — a rounding error against the distributor's gross margin.
• The line that actually matters: cash. If the distributor currently waits an average of 55 days to be paid on that €13 million, roughly €2 million of working capital is tied up in receivables at any moment. Being paid in days releases most of it.
• The line people forget: conversion. If even a few percent of abandoned or delayed baskets now close at checkout, that usually outweighs the fee share several times over.
• The line that can erase the rest: who absorbs the fee. If the distributor pays it from its own margin to keep prices unchanged, the deal is a financing cost dressed as a revenue stream. If buyers pay it as a visible cost of longer terms, the economics look very different — and so does the buyer's reaction.

Run this model with your own numbers before anything else. A surprising number of embedded finance projects are approved on the fee-share line alone, which is almost always the smallest of the four.

Build, partner, or wait

Once the economics are clear, the decision is rarely binary. There are three real options, and two variants inside "partner."

Build: become the licensed party

You apply for your own licence — in Europe typically as a payment institution or electronic money institution under the Payment Services Directive (PSD2) regime, or as a lender under local consumer or commercial credit law — and you take on capital, safeguarding, compliance staff and supervisory reporting. For a non-financial company this is almost never the right first move. It makes sense only when financial services have become a large share of your revenue, the partner's margin is now the main thing standing between you and a better product, and you are prepared to run a regulated business permanently, not as a project.

Partner, variant one: white-label program

The product carries your brand; a licensed provider and an enabler run it underneath. You control the experience and capture more of the economics, but you also take on program-management obligations — customer identification flows, complaints handling, often a share of fraud losses. Before choosing this, read our guide to KYC and KYB: if your customers are businesses, the onboarding burden is heavier than most program plans assume.

Partner, variant two: referral or marketplace model

The financial product stays under the provider's brand, surfaced inside your journey at the right moment. You earn less per transaction and control less of the experience, but your regulatory footprint is small and switching providers is a commercial decision, not a migration project. For most companies testing whether customers will actually use a financial product, this is the right starting point.

Wait: a legitimate answer

Waiting is the right call when transaction volume is too small for any provider to give you meaningful terms, when customers are not actually asking for the product, or when you have no data advantage the provider could not get elsewhere. Waiting is not standing still: it means instrumenting the journey so that you can see, in a year, exactly how many baskets stalled on payment terms or how many customers left for a competitor that offered an account.

Five questions to settle before you talk to any provider

1. Where is the customer's moment of decision, and is money the thing blocking it? Embedded finance works when it removes a real obstacle at a specific step. If you cannot point to that step in your funnel data, you are buying a feature, not solving a problem.
2. Who answers the phone when it goes wrong? Map complaints, disputes, failed payouts and frozen accounts to a named team on each side of the contract. The party with the customer relationship always gets the call, whatever the contract says.
3. Who holds the authoritative record of the money? If your provider or enabler disappeared tomorrow, could you reconstruct every customer's balance from your own systems? If the answer is no, that is the first thing to negotiate.
4. What is your data worth, and what are you allowed to share? Your transaction history may be the best underwriting data the provider will ever see. Price it accordingly — and check what customer consent actually permits before you send it anywhere.
5. What does exit look like? Notice periods, data portability, who owns the customer accounts, and how long a migration would take. A partnership that is cheap to enter and impossible to leave is not cheap.

Where consulting fits — and where it doesn't

None of the five questions above is a technology question, which is why embedded finance projects that start with an API comparison so often end in a renegotiation. The work that decides the outcome is commercial and regulatory: modelling the economics honestly, choosing a structure, writing the provider brief, and negotiating the contract so that liability, data and exit are priced in. That is the kind of work we describe in what fintech consulting actually means, and it is the core of our fintech and Web3 strategy practice. If you are comparing advisors for it, our checklist for choosing a fintech consulting partner covers what to test before you sign.

The underlying argument is not new to us, and it is not specific to finance. It is about owning the moment of decision instead of renting it from an intermediary. In our work with a resort hotel in Izmir, the growth came from moving the booking step out of third-party channels and into a funnel the hotel controlled and could measure. Embedded finance is the same instinct applied to the payment step — with the important difference that the money, unlike a booking, is regulated, and someone has to be licensed to hold it.

A one-page decision memo

Before any vendor shortlist, write one page that answers these headings. If a heading cannot be filled in, that is the work to do next, not a detail for later.

• The step: the exact point in the customer journey the product changes, and the current drop-off there.
• The four economic lines: fee share, cash released, conversion gained, and who pays the fee.
• The structure: build, white-label, referral or wait — and why the other three were rejected.
• The ownership map: licence, money, customer record, complaints, fraud losses — one named party for each.
• The exit: what happens to customers and balances if the partnership ends in eighteen months.

A company that can write this memo clearly is ready to talk to providers and will get far better terms from them. A company that cannot is not ready yet — and that, too, is a useful answer. If you would like a second pair of eyes on yours, request a strategy review and we will go through it with you.