Product Design. Delivery Hero. 2019-2021.
Strategies and solutions for cross-selling to solve problems varying in different markets, brands, and business verticals.
This is a case study of how Cross-Selling evolves from "selling" to be a valuable assistant to our users, recommending items during their ordering journey, and improving business metrics.
Cross-selling is recommending additional, complementary products to a customer who is already making a purchase. Cross-selling effectively is an institutional strategy to create higher quality revenue. For eg: if a user is buying a burger at Mcdonald's and the waiter asks if they want a Pepsi or fries with their order.
We launched cross-selling for Foodpanda in June 2019 with a very simple algorithm. As we grew to become a global service, challenges with different market needs, the need to constantly improve cross-selling and stay relevant with our recommendations became more evident.
Product design and strategy, UI design, Prototyping, Interaction Design
Team- PM, 2 mobile developers, 1 web developer, 2 backend engineers
Research Team: 1 researcher, 1 research ops
Before I joined the team, the team had launched the MVP on checkout, the easiest implementation considering that cross-sell recommendation is always tied to the items in the cart where the intention of buying is the strongest.
The trigger and opportunity
From the first launch data suggested that the interaction with cross-selling on the cart was only 8% with a conversation rate of 2.8%. Over 90% of users overlooked cross-sell for various reasons. The avg basket value increased by 1% in the quarter. While as a team we were making substantial money for the company, we saw a huge opportunity for improvement. In one of our usability research, we uncovered problems that the data did not tell us like attention to price, items, and details on the cart.
A foundational research study was conducted to identify the value users find in recommendations in their ordering process, how they interact with the x-sell component, and what perceptions they have about it.
Our major focus was to understand motivators and what role item-level recommendations influence users' food decision-making.
Over a course of two weeks, we interviewed 10 participants from Asian and Nordic markets. This was to understand different cultural influences while decision-making.
The research was conducted by a researcher and my role was to create a brief for the foundational research, focus areas of problems, and expectations from the research. I helped the researcher as a note-taker, involved in debriefings, and in the observations and synthesis process. I conducted a session with my cross-functional team to derive insights, set goals and solution directions.
"... I already know what to order before arriving in the cart "
Cross-selling was too late in the funnel, the users had already made up their decision by going through the menu extensively.
On the cart, as users had a definitive goal of checking the total price of their order, users paid attention to items in the cart, total price, delivery time than cross-sell.
"... If I have to order coke, I go back to the menu to add it"
Users tend to go back to the menu after seeing a specific item from a particular category. Cross-sell in some context act as a reminder than a recommender. This is further supported by continuous back and forth between menu and cart which is on an avg 3 times. If the first two items are irrelevant users tend to not scroll through other items in cross-sell.
"... If a restaurant has very cheap rates, I would suspect that something is wrong"
While Nordics participants had this thought process, Asians are more pro-deals and discounts. The cultural difference between Asian market and the Nordic market suggested that one strategy doesn't not fit all. We have been using the same strategy for all countries and treated users the same.
Vision for cross-sell
Taking a step back, we got back to the drawing board and numerous ideation sessions and collaborative workshops to re-think how we can bring value to users rather than be perceived as a selling service.
Our success metrics were aligned with our company OKR i.e increase order share and conversion rate with a higher basket value from cross-selling recommendations
Challenge: Serving 20+ countries with different cultures, buying behaviors, and languages, how do we make sure our strategies cater to our users the best possible way.
at the right journey
at the right moment
with the right product
(stong product relation)
(strong buying intention)
Behaviours & Patterns
Collaborating with research, data, insights, user goals, and motivations, we planed our strategies based on four main user roles:
Roles behave differently towards:
The Ordering process (how they interact and navigate the app) Recommendations (how do they perceive and value recommendations and triggers)
Cross-selling (how do they interact with it, how do they value, perceive it, and why)
Also, elements within the app they pay attention to or could trigger them. Besides the level of decisiveness, we identified that promotions and quality were the main triggers for different roles about recommendations and cross-selling.
Role 1: “Always open to new things”
💡On this role users tend to explore new places and types of dishes even when they have one in mind
💡They may browse for more options even after they made a decision on the main dish rather than completing the order right away
💡This role is highly triggered by app recommendations on the both levels: before the main dish and after
💡When they were asked to add something to their order they would check on c-sell first rather than going to the menu.
Role 2: “It depends”
What is special about this role?
💡Users on this role tend to change their decisiveness level frequently and be either explorative or decisive depending on the context or their mood in terms of cravings, time, and previous orders (trying not to be repetitive).
Role 3: "Do not want to risk, but if we have deals"
💡Users on this role are not usually triggered by recommendations as they usually have a particular place or food in mind, but if they see a good promotion they can change their mind.
Role 4: “I know what I want: quality”
💡Users on this role have a specific and clear idea in mind what they want to order before opening the app and they are not highly triggered by promotions or recommendations as they are confident about their decision and do not want to risk.
💡Promotion on the main dish or the restaurant could cause a distraction for such users as they would consider it as a sign of bad quality
💡In the app they more often go to favorite or saved restaurants
Persuasion during food decision-making :
Helping users choose additional items from the drinks and desserts category when their buying intention is high and focused. Limited recommendations, one from each drink, dessert, and side dishes helped users in choice reduction.
Problem Space: On average, a user takes 6 minutes at our Menu. Within these 6 minutes, they add the first item in 4 minutes. In the other 2 minutes, our hypothesis was that the user spends browsing, giving cross-selling the opportunity to recommend more products.
iOS, Android, Web, mweb
While on the menu on average, 30% of Foodpanda session users interact with at least one item modifier. The buying intent of our users is strong as they choose multiple toppings and in the mental model of choice reduction. We introduce cross-sell to leverage these two behaviors of choosing and choice reduction by making it possible to select Drinks, dessert ( the browsing time for categories are extra 2 min out of six) in one user flow.
Item modifier with three items from only three categories
Variation with more than three recommendations
Items with no toppings. Items that do not have topping opens a half bottom sheet that shows product details. To keep the cross-selling experience consistent we introduced recommendations with only one item.
Ideally, every item on the menu has a half bottom sheet but we excluded this interaction on cross-selling items to withdraw unnecessary product loop.
We excluded the category of the main items from the recommendation. Eg: If a user opens Coca-Cola, the recommendation shows only desserts and side dishes.
Platforms: iOS, Android, mweb
Platforms: iOS, Android, mweb
We tested each variation against its control individually in the first round of tests, and a variation against each other in different markets to gather insights about which user journey brings the most delight and value to our users.
Persuasion just after leaving the menu:
The goal of this strategy is to show cheaper product recommendations to the user in a focused area just before they arrive on the cart.
Variation with full bottom sheet with ten items on scroll
Variation with items with images only, only four items
We increased the cross-sell order share, conversion and add to the cart on an avg by 40%* In a particular test it increased the cross-sell order rate by 100+%
*due to confidentiality I cannot share the exact figures
We launched the full rollout in SG along with tests running in 10+countries in one quarter.
Cycle of experiementation:
We went beyond co-occurrence and similar users' data to test with the right number of items to be shown, ranking of relevant items, how many times should we show a bottom sheet in a session, how users react to items with images and without images, discounts that may influence decisions. These experiments helped us to gather insights about user preferences in different countries.
One size doesn't fit all
We used the same strategy for countries showing items from a range of categories to the users. Data showed us that different countries behave differently while choosing items from cross-selling. Eg: TW 30% add Drinks, 24% add Big Mains, 22% Small Mains. Contrarily, users from Hungary add 45% main dishes.
This shows that users have different expectations in cross-sell and make decisions differently, opening a question "how can we make cross-sell relevant to each behavior"
Only 18% of users swipe at least once through cross-sell, showing that most users do not explore further than the first items suggested. One of the user research tells us that users do not explore the cross-sell lane if the first item is not relevant to them. For most countries, users add a from wide variety of categories.
Filters allow users to choose items from the top two categories that are most relevant for the country. Providing filters in the cross-sell swimlane would help users with more choice from the desired category helping them to complete a meal in the cart rather than having to go back to the menu.
Platforms: iOS, Android, web, mweb
Social proof to generate more confidence in dish choice.
From one of the research, we saw that users are more inclined to buy items that are validated by other users. We took a bold decision to run a fake door test i.e running a UI A/B test experiment with a very simple tag to move faster and validate our assumptions.
While we recommend items to the user, we do not take the country, user context into account. For eg: during Chinese New year mooncake becomes more popular for a month but co-occurrence data would never pick that up.
To learn faster about user expectations and validate user expectations, we introduced a fake door experiment for tending items on the cart. The first item for one session would always be shown as trending. Once the item is added to the cart the trending tag disappears.
Platforms: iOS, Android, mweb
An indicator to show the number of purchases made by other customers for a recommended item. With the same hypothesis, we introduced another fake door experiment that would add a real quantity to the social proof. All items on cross-selling would have the tag.
A tool to build customized market-driven strategies.
With different market needs and user segmentation. Cross-Selling needs to be tailored in order to be more useful to the users. With requests coming in from 20+ countries the development load on our backend engineers was high and it impacted the rate at which we could experiment with cross-selling. The back office provided the Product Specialists with the ability to alter cross-selling recommendations according to insights, deals, and user behavior. They can run experiments to gather insights on which kind of tailored recommendations work best for end-users.
The goal of the tool was to make our internal users (Product Specialists) independently run, override X-sell recommendation algorithms according to market and user needs. The tool also enabled users to track data and learnings of each experiment. This saved time for our developers and made our internal users experiment and iterate on algorithms faster.
Results And Impact
We improve cross-sell revenue impacting the flowing metrics
1.) Increased order share by 26%
2.) Increase total Cross-sell revenue by 10% quarter on quarter
3.) GMV increased by 24 %
4.) Increased cross-selling item value from 0.9€ to 1.66 €
By the end of fixing the basics, we functioned with an experimental mindset, A/B testing every feature, algorithm, and tracking business and UX metrics. In Q2 2021, we ran 10 completed experiments, the highest achieved by any squad.
Designing strategies is basically designing invisible UI
A lot of decisions and ideations that I took as a part of the team were UX and data-related. These decisions impacted the algorithm, pricing strategy, user context, etc. A lot of times when designing for recommendations the best design skill is to ask a lot of questions and apply product thinking to impact how things work, how the experience is measured, and which is the best solution to induce maximum impact.
Faster and Cheaper is better than perfect and later
We made numerous decisions and innovated strategies to reduce launching experiments sooner, this paid off as we learnt something new about the user behaviour evrytime.
Customer value is not a constant, but a moving target.
Assuming that every market behaves the same to cross-selling is not be the best strategy. Every market has not only behavourial differences but cultural as well.
Continually test and learn
Constantly testing new approaches on multiple platforms, and at various stages of the consumer decision journey helped us to understand which strategies work the best for which markets and drive maximum user and business value