Designing a regulated self-service journey at Three UK
How I used research and content design to turn a high‑demand assisted journey into self-service.
Interviewers:
Astrid DecropHead of Product Design
James HaryottSenior Product Designer
Presented by:
Tolu Akinyemi
Senior Content Designer
Designing a regulated self-service journey at Three UK
How I used research and content design to turn a high‑demand assisted journey into self-service.
Interviewers:
Astrid DecropHead of Product Design
James HaryottSenior Product Designer
Presented by:
Tolu Akinyemi
Senior Content Designer
Context
One journey inside a larger transformation
Programme objective
Move high-volume customer tasks from assisted support into self-service across Three’s app and website.
My role
I led content design across selected journeys, from discovery and research synthesis through journey definition, prototyping, detailed interaction content and delivery.
Agile delivery
I worked in cross-functional teams with product managers, product and journey designers, researchers, engineers, analysts, and colleagues in legal, compliance and customer care.
Before the case study
A quick quiz before we get into the case study…
Question 1/2
What help centre article do you think got the most traffic on Three's website?
The answer
Unblocking adult content
The most visited help page on the Three UK website
A feature blocked by default as a regulatory requirement until the customer's age is verified
Customers have to call or visit a contact centre, to change this setting
Affected about 2 million pre-pay customers*, roughly 20% of Three UK's customer base at the time
A digital task that could not be completed digitally
* Pre-pay customers were the least likely to have had their age verified already.
The real challenge
Moving a sensitive, regulated account change into self-service without losing user trust.
My approach
Systems thinking before screen-level decisions
I treated the brief as a service-design problem. I mapped the complete customer journey and the system around it. Research, system behaviour and operational constraints shaped the content before I wrote individual screens.
Evidence-led UCD
My evidence-led UCD process
01
Discover
Gathered customer, behavioural and support evidence.
02
Define
Turned evidence into user needs, constraints and success criteria.
03
Prototype
Mapped the journey and created interaction-content prototypes.
04
Test
Tested findability, comprehension, trust and recovery.
05
Iterate
Turned findings into journey and content changes.
06
Deliver
Aligned with product, legal and engineering through implementation and QA.
07
Measure and learn
Monitored customer behaviour and support demand to inform improvements.
Problem framing
One journey, three requirements
User need
Manage a sensitive account setting privately and confidently.
Risk if unmet
Customers remain dependent on an embarrassing, less private assisted journey.
Business need
Move a high-volume request from assisted support into digital self-service.
Risk if unmet
Avoidable contact demand, higher service costs and greater pressure on advisers.
Compliance requirement
Keep adult content blocked by default until the account holder’s age has been verified.
Risk if unmet
Potential regulatory action, financial penalties and reputational damage.
Research
Evidence* identified where customers could lose trust
*Multiple rounds of unmoderated usability testing involving 20 participants.
Research evidence
What it changed
100% preferred a self-service route.
Remove unnecessary assisted contact from the main journey.
70% expected the assisted conversation to feel embarrassing.
Keep the task private, direct and matter-of-fact.
60% thought entering card details meant payment.
Explain the purpose of the card check before displaying the form.
75% of users asked when the change would take effect.
Confirm the system’s dependable processing time with engineering.
75% assumed the setting might also control Wi-Fi.
Explain that the filter applies only on Three’s mobile network.
Research insight
80% of participants assumed ‘adult content’ meant just pornography
What this meant for the content
I needed to show that the filter covered more than pornography. Explaining its wider protective role helped reframe the feature and defuse some of the stigma created by its sexual association.
The filter also covered
Sexual assault
Abuse
Child abuse and incest
Animal cruelty or animal death
Self-harm and suicide
Eating disorders and body image
Violence
Kidnapping and abduction
Death or dying
Pregnancy and childbirth
Miscarriage and abortion
Blood
Mental illness
Ableism
Racism and racial slurs
Information architecture
Navigation to the feature prioritised the user's mental model
I used tree testing to compare different routes and labels before deciding where the setting should live. The final location reflected customers’ mental model of controlling what a mobile number could access.
01Sign in
02Controls
03Adult Content Filter
90%
of test participants found the setting on their first attempt during tree testing sessions.
The route avoided Three’s internal service structure and grouped the setting with other account controls.
Content strategy
The journey needed to answer seven questions
Location
Where can I find the setting?
Action
What can I change?
Consequence
What happens when I change it?
Scope
Where does the protection apply?
Verification
Why do you need my card details?
Timing
When will the change take effect?
Recovery
What happens if something goes wrong?
This sequence kept supporting information close to the decision it affected instead of placing every explanation on the first screen.
Journey design
Content strategy translated into design
01 · Start the journey
Where do I start?
Keep the journey inside the signed-in account so permissions and the relevant mobile number can be established.
1 of 8
What the work made reusable
A reusable pattern and two writing rules emerged from the work
Scalable outcome
A reusable content pattern emerged from the work
I turned six recurring customer questions into a reusable content pattern, then applied it to other account-change journeys involving risk, verification or delayed processing.
Applied to other journeys
SIM replacement and activation
Roaming and spending controls
Number transfers and ownership changes
Parental controls
Premium-rate and chargeable service controls
Account security settings
Payment method changes
The reusable structure
01
Current state
What is the setting now?
02
Consequence
What will the change do?
03
Scope and limitations
Where will it apply, and where will it not?
04
Eligibility or verification
What must the customer prove?
05
Processing expectation
How long can the change take?
06
Confirmation or recovery
What happened, and what can the customer do next?
Research-informed writing guidance
Two content rules I added to our guidelines
Testing showed where these established principles mattered in this journey. I made them explicit in our guidelines for future account-change flows.
Rule 01
Describe the outcome, not the control
Say what an action will allow or prevent, not just what the toggle is called.
Applied in this journey
‘Allow sensitive or adult content’ makes the on state explicit.
1 of 2
Reflection
What I would improve beyond the MVP
MVP compromise
Reused Three’s existing but suboptimal Barclays card-check integration.
Avoided the cost and delay of procuring and integrating a new provider.
Improved clarity, reassurance and recovery within that constraint.
Post-MVP recommendation
Replace the card check with purpose-built third-party identity and age verification.
Verify age directly using document and/or biometric checks.
Remove payment ambiguity and the need to enter card details.
Further improvements
Measure where customers abandon verification and why.
Result
A digital dead end became a private self-service journey.
Private self-service. Customers could change the setting without explaining a sensitive preference to an adviser.
Regulatory requirement preserved. The MVP had compromises, but adult content remained blocked by default until the customer’s age was verified.
Almost 100% reduction in related demand over three months. Measured by reduction in customer and contact centre requests.
The data indicated a strong contribution from self-service, but did not isolate every external factor.
Question 2/2
Thank you, Astrid.Thank you, James.Any questions for me?
