Rachael Hunt
← All work

Sign-up redesign

Research-led redesign. I mapped where people got frustrated signing up, then redesigned it into one guided flow that 80% of users preferred.

Role
Lead designer, with a product manager and the engineering team
Product
Best Health patient booking, Best Practice Software
Market
Australian patients
Methods
Hotjar, interviews, journey mapping, usability testing
The completed Best Health sign-up in a browser: email verified, password rules ticked, personal details, mobile verified and terms accepted, with a Create my account button
The redesigned sign-up, completed: one guided page with email and mobile verification built in.

The problem

To book an appointment, patients must have a Best Health account. Users found the sign-up long and confusing. Hotjar recordings and user interviews showed the old journey jumping through six stages: picking a time, an unexpected login, setting up an account and verifying email, verifying a mobile number, creating the account, and then a separate form asking for details they'd already given.

"I already gave you my details, you know who I am."
A patient, on the final step of the old journey
Journey map of the old sign-up across six steps, with patients' feelings dropping from hopeful to frustrated
The previous patient journey, mapped from Hotjar and user interviews.
The old journey, step by step
StepPatients feltWhat they saidPain point
1. Selected timeHopeful"Perfect, I'm free then"Whether the selected time is held is unknown
2. Login into accountConfused by detour"What is Best Health? It's not my clinic. Why an account?"Account value and clinic / account relationship unclear
3. Set up account & verify emailConfused by form order"Why verify now?"Fragmented order and context switching
4. Verify mobileImpatient"Another! How many steps left?"Second contact check adds effort
5. Create accountRelieved"Now I can book"Account creation ≠ booking confirmation
6. Complete patient detailsFrustrated"I already gave you my details, you know who I am"Unexpected form; repeated info

What I did

I redesigned sign-up as one guided flow, built around progressive disclosure, so patients always know what's next and why it's needed.

  • Explain the value up front. A short panel says what an account is for (managing appointments, booking for family), encouraging patients to manage their own bookings rather than call the clinic.
  • Clear numbered sections. Headings and copy link each step to the information it needs.
  • Verification in place. Email and mobile are verified inline with clear success states, following established MFA patterns.
  • Ask once. Address and date of birth are gathered during sign-up, which removes a screen from booking.
  • A calmer look. A modern, soothing palette suited to health software, for people who may be feeling unwell.
Step 1: email field empty, Send a code to verify my email disabled Step 2: email entered, Send a code to verify my email ready Step 3: Successfully verified your email
Verifying an email in three steps, with a clear success state.

Entry

The start of the new sign-up on a phone, with only the first section open

Email verified

The sign-up on a phone with the email verified and the password section next
On mobile, each section opens as the previous one is done.

Outcome

The redesign tested well in moderated usability testing. Together with the booking fixes, it removed 2 steps and reduced drop-offs.

80%clear user preference for the redesign
83%described it as modern; 67% friendly, 58% clean
Intuitivethe flow was understood without instruction

Try the prototype

How I'd use AI now

This project was done before AI tools were part of my process. If I did it today:

  • Use AI to sift Hotjar recordings and interview transcripts for where people hesitate or give up, and draft the first journey map from them, then check every step against the evidence.
  • Build the flow as a working coded prototype in a day, so testing uses real interactions (verification codes, password rules) rather than static screens.
  • Draft the tone and helper text in several versions and test the strongest against each other.
  • Keep patient details out: anonymise recordings and transcripts before any AI tool sees them.