How to Build a Banking App: Neobank, Digital Wallet, and Embedded Banking

Payroll companies, gig platforms, and vertical SaaS operators all eventually hit the same wall: their customers live financially inside the product, but the money flows out to a third-party bank the operator cannot see or control. The natural answer is to bring the bank account in-house — issue a card, hold a balance, and become the financial layer for your users. That is a banking app build, and it is meaningfully different from adding a payment button.

How much does it cost to build a banking app?

A BaaS-backed MVP — deposit accounts, a debit card, ACH transfers, and push notifications — costs $120K–$200K and ships in 20–28 weeks. A full neobank with lending, savings goals, and multi-currency support runs $350K–$600K over 36–52 weeks. Regulated infrastructure with your own banking charter starts at $2M and takes 24–36 months. Build scope drives cost more than technology choice here.

The BaaS-backed MVP is the right starting point for 95% of the operators reading this. You do not need a banking charter to launch. You need a BaaS provider — Unit, Treasury Prime, or Marqeta are the most common — that partners with an FDIC-insured sponsor bank, and you build your product on top of their APIs and compliance infrastructure.

The $120K–$200K range covers KYC/AML flows, card program setup, core account management screens, and the compliance review cycles that BaaS providers require before you go live. Engineering alone is not the cost driver. Onboarding and compliance review cycles are.

How does a neobank actually make money?

Interchange fees, subscription tiers, lending margin, and partner financial products are the four revenue lines — but their relative weight depends entirely on whether you are building a consumer neobank or an embedded banking feature inside an existing SaaS product. Getting the model wrong in V1 shapes the wrong features.

Interchange fees are the primary engine for consumer neobanks. Every time a user swipes your debit card, the merchant pays a processing fee and your card program earns a share — typically 0.5–1.5% of the transaction. According to Forbes, 2023, Chime, the largest US neobank, was processing over $8 billion in annual transactions and generating roughly $600–$700 million in revenue from interchange alone.

Subscription tiers work when you can deliver features users will pay for directly. Revolut charges $9.99–$16.99 per month for premium tiers covering travel insurance, higher ATM withdrawal limits, and multi-currency accounts. This model requires a higher-engagement user base than interchange alone.

Lending margin — interest on overdraft facilities, personal loans, or buy-now-pay-later products — is where larger neobanks find their second growth curve. It carries more credit risk and regulatory complexity, which is why it belongs in V2 or V3, not at launch.

Partner financial products — insurance, investment accounts, credit builder tools — distributed through your account interface earn referral fees or revenue shares without requiring you to hold the underlying risk yourself.

For operators building an embedded banking product rather than a standalone consumer neobank, the math is different. The primary value of the bank account is retention and data, not interchange. The account keeps users financially inside your platform, and monetization follows from that stickiness.

Who actually builds a custom banking app?

Four specific types of operators consistently reach this decision. If you are not on this list, the right first question is whether embedded banking is genuinely your product or a distraction from it.

Payroll and HR platforms adding earned wage access (EWA) need to be the destination for those funds. A third-party bank account creates friction and leakage — the employee gets paid, then moves money somewhere else. An embedded deposit account and a card issued by the platform closes that loop. The platform becomes the financial home, not just the payroll processor.

Gig economy platforms with a retention problem are the second group. On-demand platforms that pay workers weekly via standard bank transfer compete on rate alone. Platforms offering a spending account, instant payouts, and basic savings tools have a retention advantage. According to the Aspen Institute, financial wellness tools rank among the highest-valued benefits for gig workers, who often lack access to employer-sponsored financial products. The card and the account are a retention tool dressed as a financial product.

Credit unions and community banks that need a digital-first experience without replacing core banking are the third group. Core banking replacement costs $10M or more and takes years. BaaS-backed mobile apps let a credit union compete with Chime on mobile experience without touching existing compliance and risk infrastructure.

Vertical SaaS companies whose customers are already financially inside the product are the fourth. A property management platform where landlords collect rent, pay contractors, and track expenses is a natural host for an embedded business account. The account is not a new product — it is the financial layer the existing workflows already need.

Build vs. buy vs. white-label: when does custom actually win?

Custom is not the right first answer for most operators. Here is how to work through the three paths.

Use an existing neobank (Chime, Revolut, Monzo) when your users need a personal account, you have no product differentiation to offer, and your goal is distribution, not a product. There is no build required, but there is also no product — you are pointing users at someone else's app.

Use a BaaS provider (Unit, Treasury Prime, Marqeta) when you need banking features embedded in your own product and brand. This is the right path for the vast majority of operators. You own the UX; the BaaS provider owns the charter, compliance infrastructure, and the sponsor bank relationship. Unit allows companies to go live with an FDIC-insured account product in weeks rather than years, without needing their own banking license. Build cost: $120K–$600K depending on scope. Ongoing cost: per-account and per-transaction fees to the BaaS provider.

Build regulated infrastructure (apply for a banking charter or build a full sponsor-bank relationship) when you are processing over $1 billion annually and the economics of the BaaS fee structure no longer hold against owning the underlying infrastructure. This path takes 24–36 months, costs $2M at minimum before transaction volume, and requires dedicated regulatory counsel and a compliance team. Almost no startup should start here.

What features belong in V1, V2, and V3?

Building everything at once is the fastest way to ship nothing. The phasing below reflects what actually matters at each stage — not what sounds impressive in a pitch deck.

V1: Launch — prove the account is useful

V1 budget: $120K–$200K. Timeline: 20–28 weeks.

V2: Growth — add reasons to keep funds in the account

V2 adds $80K–$150K to the build, depending on lending scope.

V3: Scale — expand the financial product surface

V3 is where the platform starts to look like a financial services company, not just a fintech feature. Budget: highly variable, $200K–$400K additional.

What engineering problems eat your budget?

Most banking app overruns trace back to two specific failure modes, both of which are foreseeable before a line of code is written.

How does sponsor bank dependency risk affect your build?

Single sponsor bank dependency is the most common architectural mistake in BaaS-backed builds, and it can cost $200K–$500K to unwind. After 2023, when regulators increased scrutiny on BaaS partnerships, several sponsor banks exited the market — some with 60 to 90 days' notice. American Banker, 2024 documented the resulting scramble: neobanks on single-sponsor structures faced forced replatforms or shutdown on compressed timelines.

The fix is architectural: abstract the BaaS provider dependency from day one. Design your account model, card issuance flow, and transaction ledger so that the sponsor bank is a pluggable integration, not a hard dependency baked into every API call. That abstraction costs roughly $15K–$25K in additional V1 engineering. A forced replatform costs $200K–$500K and nine months you do not have.

Why does card fraud hit immediately after launch?

New debit card programs attract fraud within days of going live. Fraud rings test new programs with small transactions, then escalate. Programs without velocity limits, real-time transaction monitoring, and instant card freeze capability face chargebacks that can exceed $50K–$100K in the first 90 days. Some BaaS providers include basic fraud tooling; most require explicit configuration.

Building velocity rules, device fingerprinting, and instant card controls into V1 is cheaper than the chargebacks. This is the single most common place founders try to defer cost to V2 and then absorb a painful fraud event instead.

What does a real banking app build look like?

KYC onboarding is always the longest integration. The technical work of connecting to a KYC vendor — Alloy, Persona, or Jumio — is straightforward. The time goes into configuring the decision rules: what happens when ID verification partially fails, how manual review queues work, how you handle address mismatches for users who have moved recently. Plan for three to four weeks on KYC configuration alone, not the two days the API documentation implies.

Card program setup with a BaaS provider involves a compliance review of your product, your user base, and your marketing claims before you go live. This review can take four to eight weeks. It is not an engineering dependency — it runs in parallel — but teams that do not start the BaaS application at the beginning of the project get stuck waiting for it at the end.

"The operators who ship fastest are the ones who treat BaaS onboarding like a project track, not a vendor procurement. Start the compliance review on week one, run it in parallel with engineering, and you are not waiting 60 days for approval after the app is done," says Ashit Vora, co-founder of RaftLabs. "In every banking app build we have scoped, the teams who skipped that step lost an average of six to eight weeks at the worst possible moment — right before launch."

According to S&P Global Market Intelligence, neobanks that reach profitability do so primarily on the strength of debit card interchange and tight cost discipline on fraud and customer acquisition — not on novel feature sets. The product advantage is the account experience, not the feature count.

How does RaftLabs approach banking app builds?

Every banking app project we take on starts with a sponsor bank and BaaS provider audit before we write a line of code. The provider you choose affects your compliance timeline, your card program configuration options, your fraud tooling, and your fee structure. Choosing the wrong provider is a $50K–$100K mistake that shows up 18 months into the build.

Our default architecture for BaaS-backed builds isolates the banking provider dependency behind an internal service layer. That decision adds two to three weeks to V1 and saves months of replatforming if your sponsor bank relationship changes. We include V1 fraud controls — velocity limits, device signals, instant freeze — in every card program build, not as an optional line item.

If you are evaluating whether a banking feature belongs in your product roadmap, we can give you a scope and a cost range in one conversation. Book a 30-minute scoping call and we will tell you which BaaS provider fits your user base, what the compliance timeline looks like, and where the real cost risk is.

RaftLabs is a product development studio. We build fintech products, SaaS platforms, and AI-integrated applications for founders and operators at growth-stage companies.