Boosting Milkrun’s Basket Size via Predictive Design.

Milkrun’s customers were loyal but placed small, costly orders. As the sole designer, I found that lowering the effort to shop grew revenue faster than adding incentives. Three lean features lifted AOV from $61.40 to $63.00, grew baskets from 12.5 to 12.7 items, and cut interaction cost by 35%, while checkout conversion held steady.

Growth StrategyBehavioral DesignPredictive DesignLean PrioritizationBusiness Impact
see it in action

Worked with

Head of Growth
Head of Product
Product Owner

My Role

As the sole designer, my brief was to find quick, low-effort ways to grow Average Order Value (AOV) and basket density without slowing down checkout.

Duration

Part of F26 Strategy
May - Aug

The Problem

Low-density baskets. Small orders meant delivery cost often outweighed margin. Returning users had to hunt for their weekly staples every time, so they forgot items and checked out early.

The Challenge

Speed vs. profitability. Speed is Milkrun’s promise. Anything that slowed checkout would cost more than it earned, so every intervention had to feel like a shortcut, not a barrier.

The Approach

Our data showed baskets were consistently small. I sorted every idea by effort and time to ship: instant fixes first, bigger builds later. Bundles and similar features were taken off the table for this phase because they would take months. That left three changes we could ship fast.

Key Outcomes

6 Months after shipping

35% ReductionInteraction Costs: users found what they needed faster, the key to retention
$61.40 to $63.00 (+2.6%)Growth in Average Order Value (AOV)
12.5 to 12.7 (+1.6%)Items per Basket (Density)

Solution

1. Turn “Past Purchases” into a Fast Lane

I moved past purchases from a buried account setting into the main navigation as Order Again, with Previously Purchased, Specials on Your Favourites and Popular tabs. Most revenue came from recurring purchases, and the rebrand was about convenience.

A third-party AI recommendation tool was an option. Instead, I used existing order history with simple flags to hide unavailable items and skip what was already in the cart. It shipped in weeks and kept external dependencies low.

Order Again flow
Order Again flow: 1. Home page. 2. Tap Order Again to open the Order Again page. 3. Tap Frequently Purchased to see the items you buy most.
Order Again icon
I designed the Order Again icon to make reordering effortless. The heart marks your favorite items, the products you buy again and again. The circular arrow signals repeating that order. After adding Order Again to the mobile app, I brought the same feature to the Milkrun website.
Order Again tabs

Previously Purchased uses order history, with a couple of flags to hide unavailable items and skip what’s already in the cart. Users find their daily essentials straight away, without scrolling through months of orders.

Specials on Your Favourites merges past behaviour with current promotions. It triggers loss aversion, not wanting to miss a deal on a known staple, to drive immediate conversion.

While the other tabs focus on replenishing what users bought before, Popular is a low-friction discovery tool that expands the user’s share of wallet.

Order Again in the design system

I integrated the new tab into the new design system I was developing. This redesign aimed at maximizing Customer Lifetime Value (LTV) by reducing the friction of the weekly restock, ensuring Milkrun remained the primary grocery utility for our power users.

2. Catch Forgotten Items Before Payment

A single “Have You Forgotten?” page shows 5 to 7 staples the user buys weekly but missed this session, with one-tap add and an instant skip.

Stakeholders worried an extra page would hurt conversion, so I built a prototype to show how light it felt. It sat after “Proceed to checkout” but before payment, when users are committed but not yet in the payment flow. The prototype won the room.

We also went back and forth on whether “Proceed to checkout” belonged above or below the products. I argued for a sticky CTA so the way out was always visible. Engineering pushed back on the extra build effort, most strongly on the website, with some resistance on the app too. The Head of Product backed the sticky version in the end.

To de-risk the launch, we released it to a small group of customers first and only rolled it out wider once conversion held steady.

Have You Forgotten screen
Live app screenshot: the “Have You Forgotten?” screen sits between checkout and payment, surfacing frequently purchased items users can add in one tap or skip.

3. Make the Next Reward Visible

A progress banner above the bottom navigation fills as the cart nears a threshold. It’s built to track any goal: free delivery, bundle offers or loyalty tiers. Seeing the reward at the right moment nudges one more item instead of an early checkout, which grows basket density and AOV.

Its specific impact on order value is still being measured.

Progress banner
Progress banner filling
Live app screenshots: the progress banner sits above the bottom navigation and fills as the cart gets closer to unlocking free delivery.

Saved for Phase 2

Two features had higher long-term value but were technically heavy, so I roadmapped them for later cycles and let the quick wins stabilise unit economics first.

Fixed and Mixed Bundles would offer fixed and mixed multi-buy incentives, such as “3 for $10 snacking deals”.

Add to Order would let users add items to a live order while it’s being picked, cutting the cost of split deliveries.

The Impact: Driving Unit Profitability

Key Success Metrics

Growth in Average Order Value (AOV): $61.40 to $63.00 (+2.6%): The “Have You Forgotten?” page became one of the most effective points for adding items before payment, so more bags left the dark store with a higher total value while checkout conversion held steady.

Expansion of Basket Density: 12.5 to 12.7 items per basket (+1.6%): By making “Buy It Again” a primary navigation pillar, we reduced the friction of the weekly shop, resulting in users adding more items to each delivery.

35% Reduction in Interaction Cost: By transforming the search-heavy journey into a predictive “one-tap” experience, we significantly reduced the time and physical effort required to shop. Users could now rebuild most of their recurring basket without visiting a Product Detail Page (PDP) or using the search bar.

Using existing order history APIs instead of a new third-party AI tool kept the platform technically scalable, with a low external dependency load.

Key Learnings & Reflection

Momentum is a competitive advantage. In q-commerce, a great design that ships in weeks beats a perfect one that takes months, and “Have You Forgotten?” proved it in revenue. Effort also beats incentives: lowering the work to shop was the fastest, cheapest revenue lever, and I now check it before reaching for discounts. Ultimately, design is a business lever. Working as the sole designer taught me to frame every idea by engineering effort and revenue impact, bring a prototype when the room is unsure, and be intentional about deciding what not to build and what to save for later.

↑