Cameron Ross

Optimising a registration form to improve research participant data quality

Time to read:

3 minutes

Objective

Improve data quality and increase the conversion rate of the research programme registration form.

Problem

Accurate registration data is essential for a smooth participant experience and fulfilling partnership obligations with NHS Blood and Transplant.

  • 2.21% of user accounts have incorrect birth dates, names, or both, necessitating labour-intensive manual matching.
  • Users are unable to complete in-person appointments due to incorrect information, with an average of six such cases per month, harming the participant experience.
  • Incorrect email address entry prevents users from verifying their emails, halting the registration process.

Discovery

We assumed that while a name may change between registrations (marriage, divorce, name change by deep poll etc), birth dates remain constant, highlighting an opportunity to improve form accuracy.

To understand the registration form requirements, I mapped each data point to stakeholders across the business which clarified the importance of each each piece of data.

A screen shot of a miro board showing which business areas are stakeholders for different registration form pieces of data
Mapping data points to do business stakeholders

This stakeholder engagement provided transparency on the piece of work and reassurance that the downstream impact of any changes was fully understood.

Best practices & behavioural science

Breaking down complex forms into multiple pages is a well-documented industry standard. Multi-page forms improve data quality and completion rates by reducing cognitive load.

In behavioural science, the “goal gradient effect” suggests that users become more motivated as they progress, speeding up form completion.

User insights

A screenshot showing a single page registration form, where name, date of birth and phone number are visible
The existing registration form

The existing form was a single page, which usability testing showed met user expectations.

However, insights into the data quality revealed that the form design could be contributing to data quality issues.

A screenshot showing a single page in a registration experience asking the user for their date of birth
“One thing per page” registration form example screen

Previous tests of the “one thing per page” design confirmed it was easy to navigate and understand, validating its application to the registration form.

  • Participants navigated through the screens with ease 
  • Participants understood what was being asked of them on each screen 
  • Participants commended that the screens were in a familiar format, and referring to gov.uk services

User need

As someone who wants to join the health research programme, I want to provide my details to create my account and continue the sign-up process

Hypothesis

We believe that simplifying the registration process by splitting it into multiple screens will improve data quality and form completion rate by reducing cognitive load and leveraging the goal gradient effect.  

Design and ideation

Confident in the solution following thorough discovery, I designed and built the new screens with engineering collaboration, using our design system components. The solution included:

  • One data point per page to reduce cognitive load
  • An email confirmation field to capture typos
  • A “check your details” screen for data review before submission
Screenshots of three pages of the new registration form. The first is collecting the users name, the second their date of birth, and the third shows a summary of data entered into the form.
Examples of three screens created for the new registration form

Experimentation

In an AB test, the multi-page form increased the registration rate by 1.1 percentage points at the 95% confidence level.

An increase in email address validation errors was noted, likely due to the added confirmation field, suggesting incorrect emails were caught before registration submission.

Outcome and impact

Given the positive results of the AB test and high usability confidence, the new form was rolled out to all users. This change is expected to yield 1,769 more registrations and 660 fully signed-up participants per month.

Following longer term data analysis, the profile mismatch percentage among NHS Blood and Transplant participants was seen to have dropped from 2.21% to 1.15%, a 48% reduction.

Monthly appointment cancellations due to incorrect data fell from 6 to 3.5 on average, a 40% reduction.

Continuous Improvement

Further optimisation opportunities were identified during discovery, such as removing outdated hint text, which further increased the overall registration rate by 0.4%.

Future efforts will explore building trust by explaining why each data point is needed and showing progress to motivate users.