Money automation for people without a fixed salary.

00 File

Client	N26 Brazil
Product	Rules · money automation by trigger and action
Role	Product Designer · ran the beta user tests and sole author of the Rules study · Diletta Solutions, client N26 Brazil
Period	November 2021 to March 2022
Platform	iOS and Android
Tools	Figma, FigJam
Scope	the whole concept: map, benchmark, grammar, vocabulary and two flows

One line. The European version automates saving from a monthly salary. The study changed the question: instead of "what day does your salary land", it asked "what happens in your life". The trigger vocabulary became food delivery, your team scoring a goal, and instant payments.

---

01 Before the study: the beta, tested with people

The Rules study came at the end of four months of the app in beta. During that time I ran the user tests on onboarding: an account opened in 8 minutes on average, one in nine people blocked by a foreign ID, and a step indicator that looked like a button. I redesigned the step and tested it again, with variants in rotated order. With blind TalkBack users, I gathered the accessibility requirements. The detail is in section 11.

---

02 Summary

Problem. N26 Europe automates money with Rules: move an amount into a savings sub-account on a date and frequency. The anchor is a monthly salary on a fixed date. The Brazilian operation had the same engine and an audience with a different income rhythm and a different repertoire of events.

What I did. A concept map in four questions, a benchmark of three categories, a rule grammar in a single sentence, a localized vocabulary of triggers and actions, named templates as an on-ramp, and two flows end to end.

Result. The proposal, two complete flows, and three open questions on the record. Never shipped: N26 closed its Brazilian operation in November 2023.

Role. Sole author of the study. No access to usage data and no research budget, which is stated because it changes what the study can claim.

---

03 Context

N26 opened in Brazil in November 2021, two years after announcing its arrival, starting with a group of insiders. In March 2022, when I ran the study, the Brazilian team was deciding which European features to import and in what order. The operation closed in November 2023 without passing 200,000 customers.

---

04 Constraint

The vocabulary lesson from this study comes back in CPF Seguro, the fraud-protection product, where it also has to speak the language of someone who arrives suspicious.

The engine already existed. An adaptation, not a new product. Rewriting the backend was off the table.

Incremental entry was mandatory. Only what could ship in parts fit.

Zero primary data. No analytics, no research budget. Every figure in the study is public or a competitor benchmark. That is declared because it is the constraint that weighed most on the outcome.

An abstract concept. Rule and automation are not banking vocabulary for an ordinary user. Teaching was half the problem.

Money moving on its own. Automation that moves a balance without confirmation is automation that frightens people. Trust was a requirement, not polish.

---

05 The problem in numbers

The demand exists. 31% of Brazilians say they have no emergency savings at all. In the lowest income brackets that figure reaches 48%. Anbima with Datafolha.

The mechanic works. N26's Round-Ups in Europe accumulated 12 million euros in under a year by rounding up card purchases into a savings sub-account.

The European anchor doesn't fit. About 40% of employed people in Brazil worked informally in 2022, 39.1 million people. IBGE, national household survey. Informal doesn't mean no monthly salary. But with that share outside formal employment, a monthly-salary trigger doesn't describe how a lot of people get paid.

Cross the three: automatic saving has an audience, and a fixed payday anchor doesn't reach all of it. The study's sticky notes list the four income rhythms as the first segmentation, before any screen: people paid daily, weekly, fortnightly and monthly.

---

06 Exploration

The map before the screen

Four questions on sticky notes, answered before opening a single frame: What is it? What is it for? How do you use it? Why use it? The first two became, literally, the product's two opening blocks:

What are rules? Rules are automatic actions you can create to organize and save your money.

Can I customize them? The sky is the limit. You can shape rules your way, to match your lifestyle and your financial goals.

That is also where the user's vocabulary came from, and it is not the bank's: "help me control my finances", "help me lay out goals", "put money aside".

Three benchmarks, each with one question

IFTTT · how do you teach "if this, then that" to a layperson. The full flow mapped across eight screens. The finding is the language, not the layout: "If This Then That" carries the entire mental model in four words.

Medication reminder apps · how do you configure recurrence without looking like a cron job. Frequency in plain language, days as pills, time, stock count. The category that solved recurrence for a non-technical audience, and that nobody in fintech was looking at.

Onboarding across eight fintechs · how do you introduce an abstract feature in the first session. N26, PayPal, Revolut, credit score apps, loyalty apps, travel apps. The illustrated carousel is the category standard, and it is the worst of them: it teaches out of context.

What died

The dedicated onboarding carousel. Replaced by named templates inside the Rules home itself, with an explicit promise: "We built a few templates to help you automate your day to day, simply."

---

07 Decisions

D1 · The rule is a sentence, not a form

The creation screen is a sentence with blanks:

Whenever [something happens], do the following: [pick an action]

Trigger and action in running language, IFTTT style, not loose labelled fields.

Why: the expensive cost here is understanding, not filling in. A form solves the second and worsens the first.

D2 · The trigger vocabulary is Brazilian, not translated

This is the study's central decision. The trigger list is not the European one in another language:

Trigger	What it reveals
I order from a food delivery app	delivery as a spending category, not a store
My team scores a goal	an emotional event, outside the statement
It's day XX of the month	the only one inherited from Europe
I make a transaction	generic, by card type
I split a purchase into instalments	instalments are a Brazilian institution
I hit a percentage of my card limit	the limit as a behavioural ruler
Total spend for the month	a ceiling, not a goal

And the actions follow: put it in the sub-account, send me a notification by SMS, app, email or messaging, send an instant payment, round up to the next unit, add it to a spreadsheet, create a sub-account.

Why: automation is only useful if the trigger exists in the life of whoever uses it. Instant payments and instalments have no European equivalent, and "my team scores a goal" never shows up in a requirements list, it shows up in research.

What I gave up: parity with Europe. What I gained is a product that speaks the local language.

D3 · A named template is the on-ramp, not the tutorial

Templates arrive with a name and an emoji, not a description of a function:

Time is money ⏳ · Hunger control 🍔🍟 · My team helps me save ⚽ · Away-game trip 🛩 · Continental cup trip

And the template is editable: "You can customize any of the available rule templates."

Why: a name solves two problems at once. It teaches the concept through a concrete example, and it solves identification in the list. Without a name, "My rules" is a list of summaries the user rereads in full on every visit.

Question left open: if the name is optional, how do you identify the rule without it.

D4 · Trust is copy, and it is explicit

The product moves money on its own. Every risky screen carries a sentence that hands control back, and those sentences were written alongside the flow, not afterwards:

• "Your rule will never run if there isn't enough balance in the account."
• "You'll get a notification to approve your rule every time it runs."
• "You can edit, pause or delete any of your rules."
• "Don't worry, you're always in control of your rules."

Why: in a product that touches a balance, the objection is not usability, it is fear. Fear is answered with a sentence, not a padlock icon.

---

08 Craft

Two flows end to end, each proving a different pair:

Example 1 · Trigger: credit card transaction → Action: custom SMS notification. The reactive pair: the app watches and tells you. It includes the custom message screen, with keyboard, and a ready-made example: "Whenever a purchase is made above 💰 send me an SMS saying 'Stop spending now! 😱'"

Example 2 · Trigger: specific day of the month → Action: transfer to sub-account. The executive pair: the app watches and moves money.

Screens covered: empty state, suggestions, step-by-step creation, numeric keypad, sub-account picker (Trip, Emergencies, Saving), naming the rule, review, confirmation, the My Rules list with an Active state, editing, pausing and deleting.

---

09 Result

Never shipped. N26 closed its Brazilian operation in November 2023.

What the work left behind: the rule grammar as a sentence, the localized vocabulary of triggers and actions, named templates as an on-ramp, the trust copy, two complete flows, and three open questions documented for the team. The rule's name, identification without a name, and which income rhythm goes first in the incremental cut.

No product metric. This case is here for the reasoning, and it is labelled that way from the first line.

---

10 What I would do differently

This study never reached a user. In the beta I tested every flow before closing it, and here I did not. The trigger vocabulary is the study's central bet, and it came from benchmarks and cultural intuition, not primary research. "My team scores a goal" could be the best idea in the study or the worst, and I don't know which. Today I would run five twenty-minute interviews before opening Figma. Five interviews fit in a week.

I segmented by income rhythm and didn't follow it through. The sticky notes open with four rhythms, daily to monthly, and the final flow says "it's day XX of the month". The study's strongest segmentation stayed on the wall and never reached the screen. That is the failure I would take to a retrospective first.

I treated it as adapting a feature, and the data suggests positioning. If about 40% of employed people work informally, income without a fixed date shows up in onboarding, in credit, in collections too. At the time I had no place to raise that. Today I would raise it.

Credits. Individual study, at Diletta Solutions, for the client N26 Brazil.

---

11 The beta tests, in detail

The Rules study came at the end of four months in which the N26 Brazil app was in beta. During that time I ran the user tests on onboarding and redesigned what they broke.

Onboarding test, December 2021. Nine people without an account opened one at the same time, timing it and logging every friction. The average was 8 minutes, excluding outliers. One in nine could not finish: a foreign ID was not accepted. Six of nine gave the two highest ratings.

Moderated usability test, January 2022. Think-aloud sessions with a moderator and a note-taker, and closed tasks: find where to open the account, use a name different from the ID, enter a foreign phone number, open the contracts. The step indicator, which the team called the macro-step, looked clickable and did not say how much was left.

Redesign and retest, February and March 2022. I redesigned the macro-step and the address form and tested them again, along with three other micro-flows, in a Figma prototype. Each variant came in rotated order, so the first one would not win just for being first. I prioritized people with small screens, because one of the flows was text truncation.

Accessibility. A blind person wrote in asking to test the beta with TalkBack. We ran moderated interviews with blind screen reader users. The biggest blocker was the account-opening selfie. The fix they suggested themselves was voice guidance during capture ("face too far left"). The minimum requirement became a list: every button labeled, a reading order that makes sense, fingerprint login.

Participant data and spreadsheets stay with the client. Only aggregate numbers appear here.

---

Sources

• N26 in Brazil, launch and shutdown: Brazil Journal, Finsiders, Tecnoblog
• N26 Automations, Rules, Round-Ups, Income Sorter: n26.com/en-eu/blog/new-in-n26-more-control
• Round-Ups, 12 million euros in under a year: n26.com, press release
• Informal work at 40.1% in 2022, 39.1 million people: IBGE, PNAD Contínua
• Emergency savings, 31% and 48% in the D/E income bracket: Raio-X do Investidor, Anbima and Datafolha
• Copy, triggers, actions and templates: taken from the study file itself