nomad+22%
YoungFintech
How Young made onboarding an always-on experimentation programme
+31%
activation · measured against a randomised holdout
In short
Young's onboarding converted signups into accounts but not into active customers. Markin turned activation into a continuous experiment: a per-customer path to the first meaningful transaction, tested against a randomised holdout, lifting activation 31% on treated cohorts.
Results
Every decision carries a control group+31%
Activation on treated cohorts
vs randomised holdout, full measurement window
+16%
90-day ARPU
activated customers reaching a second product faster
6 days
From hypothesis to readable result
previously a quarterly roadmap item
Anonymised range across Markin deployments in large B2C bases, read over a full measurement window. It is not an industry benchmark and not a forecast for your base.
At a glance
- Industry
- Fintech / consumer banking
- Markets
- Europe
- Base
- High-volume consumer signups
- Motions live
- Activation, product cross-sell, dormancy, KYC recovery
- Stack
- Warehouse, product events, Braze, in-app, support
- First measured lift
- Week 5
The challenge
Young was very good at acquisition and much less good at what happened next. A large share of new customers finished signup, funded nothing, and were quietly gone within a month. The funnel looked healthy right up to the point where revenue starts.
Onboarding was a single sequence for everyone, changed a few times a year, and every proposed improvement competed for the same engineering roadmap. By the time a variant shipped, the cohort it was designed for had already lapsed.
Nobody could say which step actually caused activation, because nothing was measured against a control.
What Markin did
Data and context, intelligence, action01 / Data & context
The first 30 days, event by event
Markin reads signup, KYC, funding, card and transaction events as they happen, so the activation state of every customer is known in hours rather than in a monthly cohort report.
02 / Intelligence
Activation as a hypothesis space
Instead of one onboarding flow, the intelligence layer proposes and prioritises many small hypotheses about what unblocks this specific customer: a funding method, a missing KYC document, a first use case worth the effort.
03 / Action
Continuous tests, retired automatically
Each hypothesis runs as a live experiment against a holdout through Braze, in-app surfaces and support. Winners scale, losers are retired without a roadmap discussion.
Experiments that ran
Hypothesis to measured lift| Hypothesis | Segment | Action | Channel | Measured lift |
|---|---|---|---|---|
| Funding friction, not motivation, blocks activation | Signups with completed KYC and no funding after 48h | Funding method matched to the customer's bank, one tap | Push + in-app | +21% funded accounts |
| A first use case beats a feature tour | Funded accounts with no transaction in 7 days | Single suggested first transaction based on declared intent | In-app | +31% activation |
| KYC drop-off is recoverable if we act inside the hour | Abandoned verification sessions | Contextual help on the exact failing step, not a generic reminder | In-app + support | +18% KYC completion |
| The second product should follow behaviour, not tenure | Active customers showing a savings pattern | Second product offered at the behavioural trigger | In-app + email | +16% 90-day ARPU |
How it rolled out
Week 0
Events connected
Signup, KYC, funding and transaction events connected read-only. Historical cohorts replayed to size the activation gap.
Week 2
First experiments live
Three activation hypotheses running simultaneously, each against its own holdout.
Week 5
First readable lift
Activation reads +31% on the treated cohort. Two hypotheses are retired in the same week.
Week 12
Always-on programme
Activation, dormancy and cross-sell running continuously, with new hypotheses entering weekly instead of quarterly.
In their words
“Onboarding stopped being a roadmap item. It's a programme that improves every week without asking engineering for anything.”
Head of Growth, Young
The stack Markin worked with
No migration, no new system of recordQuestions buyers ask about this case
Does running this many experiments require engineering time?
Not after the connection. Markin composes and launches each variant through the channels already in place, so a new hypothesis costs a review, not a sprint.
How was the Young result measured?
Every decision carries a randomised holdout. The lift quoted is the difference between the treated cohort and that control group over a full measurement window, not a before-and-after comparison.
Did they have to move their data to Markin?
No. Markin reads from the warehouse, CDP and product events already in place. Nothing is copied into a new system and nothing is locked in.
Which team ran it day to day?
The existing growth team. Markin does the analytical work, a fintech team of that size would need a dedicated data-science pod to produce, and routes every action into the tools they already operate.
How long until the first measurable lift?
A first motion goes live in about 30 days and the first statistically readable result lands inside the following six weeks. Scaled impact typically reads at 90 days.
