User research, visual, and interaction design.
Q1 - Q4, 2022
Grab is a super-app which provides users with transportation, food delivery, and digital payments services via a mobile app.
I worked on the GrabFood vertical to help user get the best value from their purchase, while minimizing the money spent on offers.
Because of the market needs, we need to keep providing competitive offers. Otherwise, it would be difficult to keep users in our platform.
However, doing so has hurt the business, especially when the economy is not doing so good.
Here are some negative impact that we found:
Too much money burned to subsidize users.
Average order value does not cover the money spent on offers.
We need to create offers on a daily basis. But the current offers creation tool is too complex, slowing down the operations workflow.
Because we want to stop over-subsidizing, we provide offerswith certain conditions, which did not go as planned.
Users often add items to their baskets but leave them before checking out.
Based on our qualitative study, the two most common reasons for drop-off are:
"Minimum order value is too high for me to make a transaction."
"Offers are not interesting enough for me to make a transaction"
Analyzing the drop off rate in the funnel. Looking for insights between people who purchase with and without discounts.
User interviews with a couple of user archetypes in Indonesia. Understanding their browsing and purchase behaviour.
Designed the ideal end-to-end experience in this phase and validated the design with usability tests.
Due to the economic crisis, the company's strategy changed to focus on saving the "cost of operations". This means we will tighten the resources for promotions and discounts, as well as human resources (including engineering effort).
So instead of launching the full-fledged feature, we scoped down and shifted our focus to AI targeting and offers creation optimization (the backend tool).
Offers sit within the GrabFood ecosystem. Which means the goal for the user is to make a purchase and order food. How might we introduce tier discount without intervening with the main jobs to be done, while keeping it easy to discover?
Tier discount should not disrupt the main jobs to be done, which is to order food.
Tier discount is just another type of offer. It shouldn't have special treatments.
Tier discount is not a marketing gimmick. So we should be upfront and communicate with clarity.
Tier discount means having multiple offers within a single entry point. We want to introduce a new type of offer while keeping it easy to understand.
How might we communicate that there are multiple tiers within the offer?
For the detail page, we need to communicate that there are multiple tiers instead of just one.
.png)
In the checkout funnel, the offer list is a way for users to get more value for their purchases. It is not the main jobs to be done. Looking from that angle, we decided to keep it simple.
Instead of cluttering the offers list with new components, we played around with the content.
The principles we set up in the beginning helped us in keeping tier discount non-intrusive. We treat tier discount the same as other offers.
In the offers detail page, we use a tab navigation to show the tiers and their respective terms and conditions.
With the new entry point, we want to make it easier for users to understand the tier that they are eligible for. When user is eligible for a tier, we will automatically select the tier for them.
User will always see the highest eligible tier by default.
When user is eligible for the higher tier, we will disable the lower tier. We want to give users the best value based on their basket size. However, they can still browse the higher tier.
When browsing for food, it's natural for users to add or remove items in their basket. As the basket value changes, we need to inform the user when the tier they selected has been upgraded/downgraded or if they are now ineligible for the offer.
We want to highlight the changes, but not intrude the whole journey. Even if the tier changed in value, user will have the option to keep going, or to take a look at the changes.
To validate the design solution, we conducted usability test sessions with 14 users from Singapore, Indonesia, Vietnam, and Malaysia. The test was done using Maze while we have a zoom video call with the participants.
The proposed design achieved the primary objectives of the project:
✅ Tier discount concept: Majority of the users can comprehend that they may get a range of discounts based on certain conditions. The information in the offer card sufficiently communicated the concept.
✅ Contextual offer titles: The offer card design and content is straightforward for users to understand in checkout stage. They understand the discount that they are eligible for and that they can get a higher discount (i.e. $9 off) if they spend more.
✅ Tier change handling: Users can understand the situation when there is update in their offer due to change from their order. Or when there is a downgrade when they removed something from their basket.
Due to the economic crisis, the company's strategy changed and is now focusing on saving the "cost of operations". This means we will tighten the resources for promotions and discounts, as well as human resources (including engineering effort).
So instead of launching the full-fledged feature, we scoped down and shifted our focus to AI targeting and offers creation optimization (the backend tool).
The AI targeting was built by the engineers and data scientist with no design effort. So in this case study, I will only focus on offers creation optimization (the backend tool).
To save the "operational cost" and optimize the backend tools, I needed to understand the problems that the business operations team are facing. First, I tried to understand the current workflow of an Ops agent.
To create an offer, an Ops agent have to open the internal tools and fill in a bunch of forms manually. It looks fine at a glance. However, it gets overwhelming over time. Because they need to do this manual labor hundreds of times a week. It's not efficient and time-consuming.
Then, I had some discussions and interviews with the team to understand and validate their pain points with the current offers creation workflow.
Most of the time, the differences are only in the value and the minimum spend. Repetitive and redundant.
Ops needs around 5-10 minutes to create an offer, depending on the complexity of the offer conditions. Time-consuming.
They are concerned that tier discount would add complexity to the offer creation process.
We want to save time and remove the complexity of the tool. So, with tier discount, we try to automate as much things as possible.
Instead of typing and selecting manually, more than 30 fields in the creation process are automated.
Tier discount allows Ops to create multiple offers with different values in one go.
With faster creation time, we will be able to help more clients in shorter time. Which leads to a revenue increase.
During the experiment, we didn't spend any effort on the User App. But with the AI targeting, we are able to show different offers to different types of users without any design change.
Based on the user's behaviors, we can surface different offer tiers values. We consider factors such as how often they make purchases and what is their average purchase amount, among other things.
We needed to save engineering effort for other crucial projects. So during this experiment, we didn't make any design changes in the user app. Here's how tier discount will look to the users.
The project started with a very high confidence level. We were sure that this would be the next big thing and that everyone would be on board with this idea.
However, the economic crisis changed the trajectory of this project, and we had to phase the project before finally getting the trust to launch the ideal version.
When I was exploring the tier discount UI, I got wild ideas here and there. It was fun, and I enjoyed the process. I'd like to think that I zoomed out and explored various possibilities that tier discount could be.
However, it turns out the most straightforward solution was the one that worked the best.
As I zoomed back into the problem and user journey, I realized that tier discount is not the main jobs to be done at that point of time. The user only had one job, which was to complete the purchase.
Remembering the main jobs to be done and setting up the principles early on in the process helped me focus on the problem and not stray too far from there.
Improving the likeliness of conversion by making discounts more achievable and less demanding.
