Benchmarking CAC, LTV, and churn against fintech category norms

The fastest way to be wrong about your own unit economics is to compare a $180 consumer checking CACCACCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → against a blended "fintech average" that quietly includes B2B payments companies closing $50,000 annual contracts. The arithmetic is fine. The judgement is worthless. Whether your numbers are healthy, flattered by mix, or failing slowly depends almost entirely on choosing the right comparison set and then adjusting it before you compare.

This lesson assumes CAC assembled the fully loaded way the true-CAC lesson describes, and LTVLTVLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → built from deposits, interchange, spend or credit balances as the modelling lesson sets out. What follows is the comparison layer on top.

Why "fintech" is not one category

"Fintech" covers business models whose unit economics have almost nothing in common. Lumping them produces averages nobody can act on.

• Interchange-led consumer accounts (neobanks): a few pounds or dollars per user per month, enormous volume, thin margin per account.
• Cross-border and payment take-rate: revenue is a percentage of money moved, so a customer sending £4,000 a quarter is worth ten of one sending £400.
• Deposit-led savings and lending: revenue is net interest margin, so LTV moves with the rate cycle and with balance behaviour rather than with app usage.
• BNPL (Buy Now, Pay Later): merchant fee plus interest, repeat purchase frequency decides everything, individual lifecycles measured in weeks.
• B2B payments and infrastructure: few customers, high contract value, long sales cycles, real switching costs once integrated.
• Wealth and trading: revenue tracks AUM (assets under management) or trading volume, so a market drawdown cuts LTV without a single customer leaving.

Each carries a structurally different acceptable LTV:CAC ratio. Benchmark a neobank against a payments infrastructure business and you will always look wrong, in one direction or the other.

How mix flatters a blended number

Most numbers that pass a benchmark check are blends hiding a worse marginal reality. Four ways this happens:

Organic dilution. Wise has said for years that the large majority of its new customers arrive through word of mouth rather than paid channelspaid channelsVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition →. If two thirds of your signups are referral-driven, a blended CAC of £15 can sit alongside a marginal paid CAC of £120. Plan next year's growth off the blend and the economics break the moment paid becomes the marginal channel, because it always does.

Revenue tail. Revolut reported group revenue of roughly £1.8 billion for 2023 against a customer base that ended the year near 38 million, which works out around £45 to £50 per customer-year. That mean is carried by paid plan subscribers, FX-heavy users and business accounts. The median retail user sits far below it. Comparing your median-ish user against someone else's mean is a category error that flatters you by a wide margin.

Cohort vintage. A blended LTV:CAC still contains customers acquired when CPMs were cheaper and approval rules looser. Rebuild the ratio on the last two quarters of cohorts only, and it usually drops.

Churn definition. Account closures and dormancy are different numbers. A neobank can report 5% annual closures while a third of the base has stopped transacting entirely, because closing an app costs the user nothing and nobody bothers. If a published churn benchmark does not say which one it counts, it is not comparable to yours. The engagement signals the retention lesson treats as leading indicators are what tell you which side of that gap you are on.

Worked comparison: same label, three different answers

Illustrative profiles, built on industry-reported ranges (estimates, various sources 2023 to 2025, including a16z fintech benchmarks and CB Insights sector reports; note a16z invests in the category it benchmarks and CB Insights sells the research).

Interchange-led consumer account (Europe): fully loaded CAC of €30 to €80, revenue around €4 per user per month, gross margingross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition → near 55%, average life 30 months. LTV lands about €66, so LTV:CAC runs anywhere from 0.8:1 to 2.2:1 depending purely on the paid-versus-organic split. In this category the acquisition mix, not the product, decides whether the business works.

Cross-border take-rate: revenue per active customer closer to £80 a year (Wise disclosed just over £1 billion of revenue across roughly 13 million active customers in its year to March 2024), an assumed 65% gross margin and a four-year life because the underlying need recurs: rent abroad, a salary in another currency, family remittances. That gives an LTV near £208, and at a £40 CAC a ratio above 5:1. Same consumer wallet, very different answer.

B2B payments infrastructure: CAC of $2,000 to $8,000, ARPU of $500 to $2,000 a month, 65% gross margin, 48 months or more of life. Midpoint LTV around $31,200 and a ratio near 6:1. Hold this business to 1.9:1 and you would tell it to overspend wildly on acquisition.

A healthy ratio is not the same as a healthy business. Marcus by Goldman Sachs gathered well over $100 billion in consumer deposits by leading with a high savings rate, and deposit-gathering CAC per account looked cheap next to any neobank. The value side never arrived: spread income compresses when rates move, cross-sell into cards and loans underperformed, and Goldman's consumer platform ran up billions of dollars in cumulative pre-tax losses before the retreat, including the 2023 halt on new Marcus personal loans and the 2024 sale of GreenSky. Cheap acquisition into a revenue model you do not control is a benchmark that passes and a business that fails.

Adjust for structural factors before you compare

1. Regulatory drag on CAC. KYC (Know Your Customer) and AML (Anti-Money Laundering) onboarding, enforced in the US by FinCEN and in the EU under AMLD, adds verification drop-off that a non-regulated consumer app never pays for. Published consumer-app CAC benchmarks are therefore structurally low for you.
2. Interchange economics by geography. EU interchange is capped by regulation at 0.2% (debit) and 0.3% (credit) of transaction value, well below typical US levels. A European neobank's LTV is suppressed by law, not by weak execution, which is why US and EU neobank ratios should never be pooled.
3. Payback period, not just the ratio. A 1.9:1 ratio with an eight-month payback funds itself; 4:1 with a 30-month payback needs external capital for years. When funding is expensive, payback is the number the board should be benchmarking.
4. Contract length in B2B fintech. A three-year payments APIAPIApplication Programming Interface: a standardised interface that lets applications communicate and exchange data without knowing each other's internal workings.View full definition → contract has a different churn profile from a monthly consumer subscription. Annualise consistently before comparing.

Churn benchmarks by category (estimates)

• Neobanks: annual churn often cited around 20 to 30% for non-primary relationships, meaning users who keep the app but bank elsewhere (estimate, various 2023 to 2024 industry reports).
• BNPL: repeat purchase rate matters more than classic churn; Klarna and Affirm both disclose repeat-usage shares in investor materials, and UK FCA moves to bring BNPL into regulation may compress repeat-usage assumptions.
• B2B payments: logo churnlogo churnChurn rate is the percentage of customers or revenue lost over a period. It measures how fast a business loses its existing customer base.View full definition → under 5% a year is considered healthy given switching costs, with net revenue retentionnet revenue retentionNet Revenue Retention measures the percentage of recurring revenue retained and grown from existing customers over a period, including upsell and expansion, net of downgrades and churn.View full definition → above 110% a common target.

Where to find real published data

Don't rely on vendor blog posts alone. Cross-check against:

• CB Insights Fintech reports for funding and category-level trend data
• Annual reports and listed-company filings: Wise and Revolut publish revenue and customer counts, and US-listed fintechs such as Affirm, SoFi and Nubank disclose CAC, ARPU and cohort retention
• a16z's fintech benchmarking pieces for operator-oriented ratio ranges

When a benchmark doesn't specify sub-category, geography and churn definition, treat it as low-confidence and find a narrower source before it reaches a board deck.

🎬 [VIDEO: "Fintech Unit Economics Explained" - youtube.com/@a16z - a16z operating partners walk through CAC, LTV, and payback period differences across fintech business models]

A quick sanity-check framework

1. What sub-category am I actually in, and does the benchmark's source share it?
2. What is the comparable's monetisation model: interchange, take-rate, subscription, interest spread?
3. What is the regulatory and geographic context behind their ARPU and my own?
4. Am I comparing blends or margins? Marginal paid CAC and recent-cohort LTV, not the all-time average.
5. Same time window on both sides, and the same definition of a churned customer?

Key Takeaways

• Never benchmark against generic "fintech" averages; match sub-category, geography and monetisation model before comparing CAC, LTV or churn.
• LTV:CAC near 2:1 can be sound for a European interchange-led neobank while the same figure would signal trouble for B2B payments infrastructure, where 5:1 or higher is normal.
• Blends flatter: organic-heavy mixes hide marginal paid CAC, mean ARPU hides a revenue tail (Revolut's roughly £45 to £50 per customer-year is carried by a minority of users), and closure-based churn hides dormancy.
• Cheap acquisition does not rescue a weak revenue model. Marcus gathered deposits at low cost and Goldman's consumer platform still absorbed billions in losses before retrenching.
• Use primary filings and annual reports over unsourced "average fintech CAC" claims, flag industry figures as estimates, and benchmark payback period alongside the ratio.