The Milkround Builder, a new onboarding experience
Streamlining and simplifying the onboarding experience leading to a 38% growth increase
Role
Product Designer
Tools
Adobe XD, VWO, Jira
Company
The Modern Milkman
Context
In my first month at The Modern Milkman, I proposed a concept called The Milkround Builder, a tool designed to improve first-time customer conversion and increase average basket value.
The idea was sparked by analysing drop-off data and seeing a clear opportunity in the sign-up journey
Problem
The existing onboarding process relied heavily on the user manually adding products after registration. This led to a high number of Sign-Up No Purchase (SUNP) users, customers who created accounts but never placed an order.
There was no structured pathway guiding users to a first successful shop.
Solution
I designed The Milkround Builder, a step in the onboarding flow that automatically curated a suggested basket of products based on postcode, delivery frequency, and seasonal popularity.
The tool aimed to reduce decision fatigue and make it easier for customers to complete their first order.
Impact
Although I was made redundant before the feature launched, I later learned that The Milkround Builder shipped and contributed to a 38% increase in growth.
First things first
Mapping the journey
I began by analysing SUNP (Sign-Up No Purchase) behaviour using product analytics and stakeholder insights.
-
Identified the most common drop-off point: immediately after registration
-
Found users were overwhelmed by choice and lacked confidence in what to order
-
Noted patterns in successful first baskets (milk + juice + bread) that could be reused



Understanding the user
Persona Development and Discovery
To better understand buying psychology, I ran informal interviews and gathered insights from support and growth teams.
-
I observed recordings and heatmaps of recent customer interactions and to explore what made their first order easy or difficult
-
Shadowed customer service reps to hear recurring points of confusion
-
Created personas and Journey Maps of the main processes
Emma - The Busy Parent
Age
37
Occupation
Primary School Teacher
Location
Suburbs
Tech Savvy?
Moderately
Shopping style
Weekly planner
Goals
-
Get essentials delivered consistently (milk, juice, bread)
-
Reduce midweek supermarket trips
-
Support ethical/local businesses
Frustrations
-
Doesn’t have time to browse product lists
-
Easily overwhelmed by choice
-
Needs the service to “just work”
What they'd love
-
A smart, ready-made starter bundle
-
Ability to adjust quantities easily (e.g., 3x milk, not 1)
-
Reassurance that she can edit or pause later
Janet - The Traditional Shopper
Age
64
Occupation
Retired
Location
Rural village
Tech Savvy?
Not really
Shopping style
Habitual, Brand loyal
Goals
-
Replace her old doorstep milkman who retired
-
Keep her morning routine uninterrupted
-
Buy British, plastic-free products
Frustrations
-
Unsure how to navigate new tech
-
Doesn’t want to scroll endlessly or "build a basket"
-
Fears being locked into subscriptions
What they'd love
-
A simple, guided flow that feels like talking to a person
-
Visual confirmation of what she’s getting and when
-
A printable or emailed summary after setting up
Developing a hypothesis
I worked with growth and product leads to align on the goal: "reduce SUNP rate and increase first-order value, without being pushy"
-
Agreed that the tool should feel helpful, not salesy
-
Kept flexibility: users could edit or remove items freely
_2x.png)


So let's build it
I created mid-to-high fidelity flows in Figma, designing a light, conversational interface that could sit naturally within the onboarding journey.
I introduced a short 3-step builder:
-
Choose your household size
-
See a pre-filled suggested basket
-
Confirm, edit, or start fresh
To bring the concept closer to reality, I worked closely with:
-
Engineering: to understand the feasibility around the dynamic basket population
-
Growth: to align on messaging and CTA performance
-
Product: to prioritise the feature against other roadmap items
What would have happened
Although we didn’t reach live A/B testing before I was made redundant, we conducted:
-
Clickthrough tests with internal users and recent sign-ups
-
Success metrics were defined: conversion to first order, average order value uplift
-
Built flexibility into the designs for future testing (e.g., number of items, visual layout)

What I learned
You can't control the outcome, only the process
Being made redundant before the feature launched was tough. But I realised that great work can still have impact even if you’re not around to see it through. The fact that it shipped, and performed well, gave me confidence that strong foundations and ideas matter more than individual presence.
Perfect doesn't mean practical
I spent a bit too long polishing early UI and flows, which slowed initial feedback. In future, I’d get scrappy prototypes in front of users sooner to validate faster. "Done, testable, and flexible" beats "perfect but late."
Collaboration beats assumption
Some early ideas I had didn’t survive contact with engineering constraints or growth goals. That’s okay, working closely with others surfaced better solutions.
Next time, I’ll involve engineers and marketers even earlier to stress-test assumptions from the start.