WHAT

Increased churn from Talabat Grocery

WHY

Aggressive pricing by competitors

SO

Timely nudge users to restock

HOW

Based on ML algorithm to make nudges feel more personalized

CONTEXT

Investing in long-term user retention amid competitors' pricing war

COMPETITORS

With competitors matching our delivery speed and sometimes outpacing us on discounts, Talabat was looking to invest in long-term user retention instead of entering a price war.

SOLUTION

Using ML algorithm to personalize re-ordering experience

USER'S REORDERING CYCLE DEFINED BY ML ALGORITHM

The ML algorithm defined by our data scientist placed user's unique ordering cycle into 3 stages.


  1. Before-cycle: Before users are expected to restock items from their previous orders

  2. On-cycle: At the user's time to order +- 2 days

  3. After-cycle: Past the time to order


This documentation ONLY covers the first iteration: On-cycle

BEFORE: TALABAT HOME

AFTER: TALABAT HOME

COMPONENT BREAKDOWN

DESIGN CONSTRAINTS & TRADE-OFFS

A crowded homepage with competing priorities made space a design constraint.

One non-negotiable set by the homepage team was the placement of our solution: it will stay below the subscription banner, which often meant it was below the fold for some users depending on their devices and regions.

Leveraging on horizontal scroll

I designed the widget to be compact and used horizontal scrolling with clear visual affordances to maximize the available space.

Making content the main driver

I believed setting the right context through words in this widget was more important than the number of categories shown.

BUSINESS CHALLENGE

Navigating high-stakes decisions involving 6M+ monthly users

This project involved working closely with high-stakes decision-makers, where I quickly realized scalability was key to getting buy-in, and therefore creating multiple strategies and scenarios to advocate for the value of the solution.

NARRATIVE: SCALABILITY

NARRATIVE: STRATEGY

PROJECT THREE

BUILDING REORDERING HABITS.

IMPACT

1st Iteration

+1.3%

Returning Users

ROLE

Lead Product Designer

DURATION

2 sprints (Dec 23)

LinkedIn

CHOI

Resume

HOW DID RESEARCH & DATA INFORM DESIGN?

I applied Hook’s model to design a contextual and relevant user experience

Recognizing that behavior is shaped by motivation and incentives, I began with user interviews to uncover their underlying goals and jobs-to-be-done.

APPLYING HOOK'S MODEL TO THE SOLUTION

Conviction built through data: reordering was the key driver of habit formation

Based on the result of a successful experiment we conducted where we navigated users to the categories they usually refill, the conversion rate from the category page increased by 11.1%.