Redesigning Prepay activation

Reimagining a complex acquisition journey in 10 weeks, balancing conversion, retention and customer trust while using AI to radically accelerate the design process

Role

Lead Designer · Sole designer

Industry

Product design · Activation · Acquisition & retention · AI-assisted workflow · Customer testing

Duration

10 weeks

Project overview

The Challenge

Our existing Prepay activation journey had fallen out of step with both our product strategy and the experience customers received elsewhere in our digital ecosystem.

Important commercial behaviours were either poorly supported or disconnected from activation. Auto Renew needed to become a part of the experience to help reduce churn, while customers wanting to bring their existing number to One NZ were only directed to a separate web form after activation.

At the same time, the experience needed to move into Flutter to better align with our self-service app.

As the sole designer, I was asked to rethink the end-to-end journey and deliver a working first-draft prototype within three weeks, with the wider project running over 10 weeks.

The challenge wasn't simply to redesign the screens. I needed to balance acquisition and retention goals with customer trust, simplify several technically complex journeys, and move quickly enough for BAs and delivery teams to develop requirements in parallel.

The project also gave me an opportunity to rethink how I worked. With ambitious scope and limited design capacity, I built an AI-assisted workflow to accelerate research, exploration, prototyping and critique while keeping design judgement and real customer validation at the centre of the process.

My role

I was responsible for the end-to-end experience from discovery through to delivery, working closely with BAs, product and technical stakeholders.

I began by reviewing Prepay activation, eSIM, physical SIM and number-porting experiences across the New Zealand market. I documented complete journeys rather than isolated screens so I could understand established patterns, friction points and opportunities to simplify our own experience.

I also interviewed internal stakeholders to deepen my understanding of the product and its operational constraints. Frontline teams were particularly valuable because they could highlight recurring customer pain points and the parts of activation that generated confusion or support demand.

A key constraint shaped the project from the outset: stakeholders needed a working first-draft prototype within three weeks so requirements and technical discussions could progress in parallel.

Rather than simply compressing my usual design process, I used the constraint as an opportunity to rethink it.

Key design decisions

Balancing Auto Renew with customer trust

Increasing Auto Renew adoption was an important commercial objective. Automatically renewing Prepay plans helps maintain continuity for customers while reducing the risk of churn.

But it also created an important design tension: how could we encourage Auto Renew without customers feeling that they had been pushed into a recurring payment arrangement?

I wanted the decision to be explicit rather than hidden in the journey.

I designed deliberate moments of consent into the experience. Customers review the relevant terms, actively confirm their agreement and reaffirm the decision before continuing.

The aim was to support the commercial objective without resorting to a dark pattern, making the value of Auto Renew clear while ensuring customers understood what they were agreeing to.

Bringing number porting into the journey

Porting presented a different problem.

Previously, customers wanting to bring their existing number to One NZ were directed to a separate manual web form at the end of activation. This disconnected an important acquisition behaviour from the core journey and introduced unnecessary friction. We know customers can be attached to their number, yet we don't currently mention porting at all until post activation.

I brought porting into the activation experience itself.

After selecting a plan, customers are prompted to consider bringing their number across, with the available 5GB incentive explained at the point where the decision has context.

The prompt remains easy to dismiss, and customers are reassured that they can complete the task later if they aren't ready. A second opportunity is provided towards the end of activation.

The post-payment porting journey was particularly complex. Several rounds of iteration were required to translate operational and technical requirements into a sequence customers could understand and complete confidently.

Reducing friction around account creation

Account creation introduced another tension between our wider digital strategy and the immediate needs of Prepay customers.

Prompting customers to register or sign in at the beginning of activation supports our broader self-service ecosystem. But Prepay also includes customers who simply want to insert a SIM, activate their service and get connected as quickly as possible.

Adding too much identity setup upfront risked creating abandonment before customers had achieved their primary goal.

I challenged the amount of account setup we were asking customers to complete during activation and advocated for deferring non-essential One Identity steps to subsequent app onboarding.

Although this introduced additional technical complexity, it protected the core activation journey from unnecessary friction while still supporting our longer-term account strategy.

A different way of working

Using AI to increase design leverage

Being the sole designer with a three-week prototype deadline meant I needed to find ways to increase my output without compromising the quality of the thinking behind it.

I built an AI-assisted workflow around Orca, Claude Code, Figma MCP and ChatGPT Enterprise.

Rather than treating AI output as finished design, I used these tools to generate starting points, explore alternatives and handle lower-level production work that I could then critique and refine against our design system, product constraints and UX principles.

Using Orca and Claude Code alongside Figma MCP allowed me to move rapidly between requirements, journey logic, wireframes and working prototypes.

Multiple tasks could also run in parallel within Orca while I concentrated on higher-value problems such as journey structure, customer needs, stakeholder decisions and edge cases.

This changed the relationship between design and requirements. Instead of BAs interpreting static documentation, I could put an interactive representation of the journey in front of them early, helping requirements and design evolve together.

Moving beyond Figma

The workflow also changed my reliance on traditional design tools.

I used Claude and Google AI Studio to create working HTML prototypes based on earlier design exploration, while our design system provided the foundation for established components and patterns.

Where appropriate, Material 3 helped fill gaps during rapid exploration.

Figma remained useful, but it was no longer necessarily the starting point or destination for every design task.

This allowed me to spend less time manually producing screens and more time thinking about the behaviour of the complete journey.

Validation

Using AI as critique, not customer research

With the first prototype taking shape, I needed to identify obvious weaknesses quickly before putting it in front of customers.

I used GPT to stress-test the journey from a range of simulated customer perspectives and levels of digital confidence.

I treated this as critique rather than research: a fast way to expose potential comprehension problems, missing states, assumptions and areas worth investigating.

The exercise surfaced several useful issues around wording, hierarchy and journey logic that I was able to address before formal customer testing.

It took less than an hour to set up and run, making it a useful additional iteration loop rather than a replacement for customer evidence.

Validating with real customers

Once the initial issues had been addressed, I worked with Optimal Workshop to launch unmoderated customer testing within 48 hours.

Approximately 20 participants worked through the journey, providing evidence about how customers understood the proposed experience and where further refinement was required.

AI helped me rapidly synthesise the findings and identify themes across the sessions, which informed the final rounds of iteration.

The combination worked particularly well: synthetic critique helped remove obvious problems cheaply and quickly, while real customer testing was reserved for the assumptions and design decisions that genuinely required customer evidence.

Outcome

The redesigned Prepay activation journey is currently progressing towards launch.

Within the 10-week project I delivered and validated the end-to-end experience, established interaction approaches for Auto Renew and integrated number porting, challenged unnecessary identity friction, and provided delivery teams with a working prototype to support requirements and implementation.

The project also established an evolution in my ways of working as a designer.

AI allowed me to compress parts of research, exploration, prototyping and synthesis that would previously have taken considerably longer. More importantly, that additional capacity gave me more time to engage with BAs and stakeholders, challenge requirements and concentrate on strategic design decisions.

Post-launch, the effectiveness of the new experience can be evaluated against measures including activation completion, journey abandonment, Auto Renew adoption and number-porting uptake.

What I learned

AI works best as leverage, not authority

The biggest shift wasn't simply producing screens faster. It was being able to delegate parts of the design process while retaining responsibility for the decisions.

AI could generate wireframes, explore alternatives, critique journeys and help build prototypes, but the quality of the outcome still depended on understanding the customer, recognising constraints and knowing when to challenge the output.

Our existing design expertise and design system became more important, not less, because they provided the standards against which generated work could be judged.

Prototypes can become a shared thinking tool

Being able to move rapidly from requirements to interactive prototypes changed conversations with BAs and stakeholders.

Instead of waiting for requirements to be fully documented before designing, we could use the evolving experience to identify gaps, discuss trade-offs and refine requirements together.

That brought design closer to product definition rather than treating it as a downstream production activity.

Synthetic critique can improve real research

The AI-assisted critique was one of the more surprising parts of the project.

It wasn't a substitute for customers, but it proved useful as an additional iteration loop before customer testing. Problems that didn't require genuine customer evidence could be identified and challenged early.

That meant real participants could be used to investigate the questions that mattered most rather than spending valuable sessions uncovering issues we could have caught ourselves.

Speed creates room for more strategic design

Using Orca, Claude Code, Figma MCP and AI-generated prototypes reduced the amount of time I spent manually producing artefacts.

That didn't mean doing less design. It meant shifting more of my time towards understanding the journey, working alongside BAs, resolving complexity and advocating for the customer.

The account-creation decision was a good example. I pushed to remove non-essential One Identity steps from activation and defer them to app onboarding, despite the additional technical complexity.

Protecting the customer's ability to get connected quickly was more important than making the implementation easier.

For me, that was the most valuable outcome of the new workflow: AI accelerated the production of design, giving me more capacity to focus on the decisions that actually required a designer.

Product design · Prepay activation · AI-assisted design · Customer testing · Prototyping · Claude Code · Figma MCP