Decrease Drop-offs from Swiggy from the food selection page

By: Lakshay Bahl

Mentor: Siddhartha Roy

Why this problem?

Usually giving too many choices to the users might result in the user making no decision at
all. Giving too many choices to the user overwhelms them and puts them in a loop of
overthinking and over analysing. This called Analysis-Paralysis.

Platforms like Swiggy work really hard to Onboard as many restaurants as possible to provide
more food options to the customers.

However, user becomes a part of a loop where s/he is scrolling through multiple listings and
not ordering anything.

This concept is known as Paradox of choice.

Breaking the Business logic

1. User Has too many choices to chose from
2. Due to Analysis-Paralysis User gets overwhelmed and is not able to make a choice
3. User Drops of the App
4. User Stickiness takes a Hit (Workable metric)
5. Swiggy Loses out on Revenue

Breaking the business metrics

Revenue= # orders x Avg commission charged (Outcome metric)

#orders= # users coming on the food page * %users placing order

# users coming on the food page= #Users who can select what to eat + #Users who cannot
select what to eat.

#Users who cannot select what to eat= #Spending less than 5 mins + #spending more than
5 mins.

#spending more than 5 mins are the people who come to the app to order but cannot
decide and drop off.
Breaking down the user journey:

Opens the App Makes a Choice Add to cart Make Payment Enjoy Food

Orders usual Food

Makes a Choice to try something new

User Pain-Point as
per Survey Importance
Knows Knows
where to what to eat
eat from

23% If you know what to eat then you can use filters and existing ratings
to make a decision

Not Many Users reported that where to eat is a big confusion for
15% them. User Interviews revealed if they know where to eat from,
users are usually aware what best dishes are available at that place.

This is the primary problem which needs to be solved, No

62% Solutions exist which solve this problem of giving decision making
support to the user

Why Solve this worth solving

With Such Staggering Number of users active on Swiggy, the App needs to solve the
problem decision making In order to successfully Satisfy these users.

There is a population which is going back to Big cities to their offices, and this Demographic
is usually too busy to cook meals. This is a huge opportunity for a food delivery business to
capitalise on.
 Breaking down the business impact:

Food Delivery business is Ultra-competitive. Most users Don't stick to one App to Order
food and every Business spends more and more in acquiring customer leading Very Low
Customer Loyalty

In Such a Space User Experience Can become Competitive Advantage

Solving for Paradox of Choice can mean:

User making quick choice = Revenue

Less Anxiety while making decision = Long-term stickiness

Solutions:

Near Me
See what others are eating around you

FEATURES
Near Me feature takes access of the GPS Location from the Phone and Scans in 1 Km
Radius around the user
It maps previous orders of the user and matches dishes Similar to that in a 1-week Period
The feature can work on parameters like Cuisines, Average Order Value, Portion-size, and
Time of the day.
These parameters can be mapped between the User and popular dishes order around
them to give relevant recommendations

UTILITY

Users get influenced by what others are eating

This gives user a custom list of meal options which are similar to their order profile,
hence making the decision of making the order easy
 Swiggy-Scratch
Scratch to reveal the biggest discounts

FEATURES

Based on making Purchases worth a certain amount, Swiggy Users can get a scratch card
The Scratch can have a dish which is heavily discounted.
eating-places can promote themselves on Swiggy by giving out meals in these scratch
cards
Swiggy has to work on relevance of the scratch card by working on parameters like
average order value, what users of similar persona rated high and proximity of the eating-
place.

UTILITY

Good discounts are a big decision making point while ordering food
Swiggy gives a boost to their revenue by increasing number of orders but also
Restaurants promoting themselves

Swiggy-Shuffle
Let us decide your Menu
FEATURES

Swiggy-Shuffle is a game which generates a Set of meal options that user has never tried
before.
The game generates excitement in the user to see what new is on the menu for them today
The user can enter portion-size, type of food they want to have and view the Meal options
The Feature can work on algorithm accounting for Demographic, discounts and order history

UTILITY
Directly helps user in deciding what to order and solves the problem
Solves the pain-point of what to eat and where to eat from
 score=
reach Impact confidence effort
(R*I*C)/E
Out of 5 Out of 5 Out of 5 Out of 5

4 3 3 4
This requires GPS access, Highly Depends on what
which people turn on at people are ordering, the Reliability is questionable
Realtime mapping of 9
purchases in 1km radius is
the time of putting reliability of the feature required.
address, hence an extra- works on other people
Near me step is required

3 5 4 5 12
Highly effective for
people who order High Confidence due to Lot of Machine learning
Only accessible after
frequently, they will be direct motivation to try a and mapping past trends
certain number of
able to unlock the card new dish and new will be required to give a
Swiggy-Scratch purchases are done
more frequently restaurant reliable output

5 4 5 3
Directly tackles the
Positioned right at the
home page, can reach
Highly effective gives a
mix of discounts and New problem
Doesn't require lots of
engineering to build it.
33.33
everyone and suited for options to try Simple logic can solve it
Swiggy-Shuffle all users

Suggested user journey:

User opens the Swiggy app and goes to the food ordering section
User clicks the Shuffle button
user enters the budget, portion size, type of food
The user views the listings of meals offered at more than or equal to 30% discount
user places the order and enjoys food

product specifications:

Swiggy Shuffle
Your Menu
Veg-only

700

Let us decide your menu Price range

Portion Size

Small Medium Large

Serves 1-2 Serves 2-4 Serves 4-6

See whats cookin

Discover Fill-in the filters Shuffle Select

User can select through When the user presses Shuffle The user views result as 3 options that they have
Users Can Discover the
various parameters to the algorithm searches for never ordered before.
feature on the home
get a more accurate Dishes and restaurants which
page of the App
result. user has never tried before The listing listing of Dishes Can be:
Relevance > Discount > Expected delivery time
Discount is also a parameter
These filters can be left which algorithm considers while
empty as well giving results
Discovery
Place the Swiggy Shuffle button below the search bar and above the listings of restaurants
and meals.

Filters
let the user enter a price range. Minimum to maximum.
User can decide the portion size; Small = Serves 1, Medium = Serves 2 to 4, Large = Serves 4
to 6.
For small meals limit the minimum value to 150, for medium 250, and for large meals at 400.
Place a button to toggle to veg-only food.
Give the users the following categories to select from: Snacks, Drinks, home cooked, healthy,
dessert, main-course meals.
Give an option for user to not select any category.
Show user a button which shows them the output.

Algorithm

Portion size, category, and veg/non-veg will be taken as an input from the restaurant owner
as an input on the meal on which he is offering discount.
The Listings on the output page will be done in the order of Relevance > Discount >
Expected delivery time.
We can define relevance by-
Example: if the user has entered drinks as the category, then scan the last 10 drinks ordered
by the user. Build an algorithm to show other drinks ordered by people who have ordered
similar drinks. Rank this list according to number of mutual orders. Assign a relevance score
to each listing as the number of mutual orders.
This output is further sorted according to the discount offered. higher discount means
higher ranking. Discard all the listings below 30% discount.
The list will be also ranked according to the appx delivery time. Discard all the results with
delivery time more than 45 minutes.
Final Algorithm: to be allocated to DS team.
Incase no category is selected by the user, implement the same algorithm on the last 10
orders by the user.

Menu Screen:
Display top 3 listings according to the result of the algorithm.
After this continue with normal ordering and delivery process.
Edge/error cases:

If all the restaurants are closed at that time, then in output screen, ask them to
try later.

Put a button to go back to home screen.

Your Menu

Sorry, all the restaurants are

closed at the moment.

Try again later!

Go back to home

If there are no food options in that budget or category. Ask the user to adjust
the price/category again.

Your Menu Your Menu

Sorry, we don't have any Sorry, we don't have any

options in this budget. options in this category.

Try adjusting the budget! Check out other categories?

Go back to Shuffle Go back to Shuffle

Success metrics:

Recommendation to order ratio

% users filling in the input filters and clicking on the Shuffle CTA.
% Decrease in the time to order from home page to placing order
% Increase in the orders per week
% Increase in the avg revenue per user
% decrease in users dropping from home page after spending more than 10 mins on the
menu page.