Case Study #5
Microcopy in onboarding and withdrawals
Two microcopy fixes at the points where a user commits: one call-to-action word rewritten for an audience spread across six onboarding stages, and a withdrawal confirmation whose Yes and No told the user nothing. The call-to-action change lifted conversion on the step by 10%.
Outcomes
+10%
Conversion lift on the invitation step
1 word
Words changed to get it
6
Onboarding stages one button had to cover
Yes / No
Replaced by buttons that name the outcome
Context
Admirals puts a new customer through a multi-stage path: register, open a live account, pass suitability testing, upload documents, get approved, fund the account. Communication runs alongside it. Lifecycle emails fire at each stage, and in-product screens carry the irreversible actions.
Two surfaces, one problem. Both had been written from the company’s point of view: what we want the user to do next, and what our systems call it. Neither had been written from the seat of the person reading, who is somewhere in the middle of the process and often not where the copy assumes.
Problem
In email. The call to action said Deposit. The message went to everyone at the invitation stage, and that audience was not in one state. Some had no live account yet. Some still had suitability tests ahead of them. For those people the button pointed at an action they could not physically take, and it told them the message was not meant for them.
In product. A withdrawal cancellation screen asked Do you want to cancel this withdrawal? with Yes and No beneath it. “Cancel” carries two readings: cancel the withdrawal itself, or cancel what I am doing in this window. With that ambiguity, Yes and No carry no information. The user cannot tell what they are agreeing to, and their own money is what moves or does not move.
Approach
1. Replaced the specific verb with a state-agnostic one
Before: Deposit
After: Continue
The audience for that email was spread across six onboarding stages, so any concrete verb excluded part of it. “Continue” resolves to whatever the next step is for that particular reader, whether that is opening an account, taking a test, or funding.
Conversion on the step rose by 10%.
2. Named the outcome on the button instead of asking for agreement
Before: Do you want to cancel this withdrawal? → Yes / No
After: Do you want to stop this withdrawal? → Yes, stop withdrawal / No, proceed with withdrawal
Two changes sit in that. “Cancel” became “stop”, which nobody can read as dismissing the dialog. And the buttons state the consequence instead of expressing consent, so the choice is legible without going back to re-read the question.
Results
- 10% more conversions on the invitation step, from one word.
- A withdrawal confirmation a user can answer without re-reading it.
- A rule the rest of the lifecycle copy now follows: when one message serves several stages, the button names the direction, not the transaction.
My contribution
- Auditing lifecycle email calls to action against the stage the reader was in
- Rewriting the invitation call to action and defining the state-agnostic rule
- Rewriting the withdrawal confirmation question and both button labels
- Setting the pattern for confirmation dialogs: name the outcome, drop Yes / No
- Measuring the conversion change on the invitation step
This case describes UX copy work I led at Admirals between 2025 and 2026. Screen copy is paraphrased to respect employer confidentiality.