Nu (Nubank)
Fintech
Employees
9,000
Clients / users
114.2 M
Products / annual scale
~250.0 M*
ICA
12,689
IPA
27,778
Exceptional. The employee who works the most is the risk algorithm. They invest so software decides, not the human.
← Back to Strategy
Strategy / Costs
The automation metric that actually matters: clients and products per employee
Stop counting tools. Start dividing two numbers.
If the same structure cannot operate more clients and more products, there was no real automation.
The formula
This is the first reality test. It is simple. It is uncomfortable. And it is harder to fake than any dashboard.
ICA — Autonomous Capacity Index
Clients
Employees
IPA — Amplified Productivity Index
Products
Employees
Then add the economic derivatives: real cost per client and real cost per product.
Two indicators that matter
ICA — Autonomous Capacity Index
Clients per employee. ICA measures how many customers each person in the team can sustain. A high ICA says one thing: technology is operating, not payroll.
IPA — Amplified Productivity Index
Products per employee. In digital models, IPA helps reveal how much complexity, portfolio depth and commercial capacity each person can sustain.
Who is really operating more per employee?
Fintech does not scale by speeches. It scales by architecture, scoring engines, infrastructure and automation built into the operating core.
ICA
IPA
Nu (Nubank)
Fintech
ICA12,689
IPA27,778
Revolut
Fintech
ICA5,250
IPA12,000
Comparative benchmark
Now the uncomfortable part. Divide operating scale by structure. The result tells more truth than any automation presentation.
Revolut
Fintech
Employees
10,000
Clients / users
52.5 M
Products / annual scale
~120.0 M*
ICA
5,250
IPA
12,000
Very high. They invest in multi-currency infrastructure, ML for fraud and chatbot-driven operations instead of traditional banking bureaucracy.
Calculate your benchmark
Now enter your numbers. The logic is exactly the same as the comparative benchmark: employees, clients or users, and products or annual scale.
Sector productivity proxy
This chart is not showing ICA or IPA. It shows a sector-level proxy: annual labor productivity change in commercial banking and insurance. The purpose is simple: to remind the reader that even when companies talk constantly about automation, sector productivity does not improve in a straight line.
Commercial banking
Insurance
Banking
2021: 11.4% · 2022: -6.9% · 2023: 5.1% · 2024: 1.8%
Read: there was a rebound, but the close is weak again. There is no sustained structural improvement.
Insurance
2021: 4.0% · 2022: 4.9% · 2023: -0.9% · 2024: 1.0%
Read: more stable than banking, but still without strong and sustained acceleration.
Uncomfortable read
If the sector barely closes at 1.8% or 1.0%, many companies are celebrating bots, prompts and APIs without truly transforming operating capacity.
Read it this way: the sector gains productivity, then loses it, then partially recovers, then closes weak again. That is why automation claims should always be tested against real operating capacity.
What companies celebrate instead
| Vanity statement | Reality test |
|---|---|
| “We deployed 14 bots” | Irrelevant if ICA stayed flat. |
| “We launched copilots” | Irrelevant if clients per employee and products per employee stayed flat. |
| “We published 32 APIs” | Irrelevant if real cost per client did not improve. |
| “We saved time in one task” | Irrelevant if operating capacity did not change. |
What to measure by industry
| Industry | Vanity metric | Serious metric | Economic derivative |
|---|---|---|---|
| Fintech / Banking | Bots, copilots, licenses, channels | ICA = clients per employee | Real cost per client |
| Insurance | RPA, CRM, quoters | ICA = clients per employee | Real cost per client / policy |
| Payments | APIs, new rails, integrations | Transactions or merchants per employee | Real cost per transaction / merchant |
Reading notes
- * Product figures for Nu and Revolut are directional operating approximations to express multi-product platform scale.
The conclusion that does not lie
Nu operates at roughly 2.4x the autonomous scale of Revolut with only slightly fewer employees. That gap is not in people. It is in architecture.
Hard conclusion
Real automation is simple to test. Divide two numbers and analyze the reality of your company.
- ICA: clients per employee.
- IPA: products per employee.
- Then look at real cost per client and real cost per product.
- Everything else is derivative, intermediate or useless if these numbers do not move.