Skip to content
Swiggy case study
Share
Explore

Swiggy case study

Overview

Goal

To increase the GMV
To increase the average order value of restaurants (AOV)

Hypothesis

I believe that if
Swiggy Mobile App suggested a relevant Food item to the user during their checkout
then
the total GMV and AOV of the restaurants will increase
as I believe
that the users will add the suggested product to their cart and checkout.

image.png

Experiment description

Variations
0
Name
Description
1
Control
Users in the control will not experience the new auto suggested feature
2
Auto suggestion
Users in Auto suggestion variation will experience the new auto suggested feature
There are no rows in this table


Test dates
0
Start date
End date
1
1/1/2021
31/1/2021
There are no rows in this table
Active* users
0
Variation
No of users
1
Control
2,00,000 (50% of total active users)
2
Auto suggestion
2,00,000 (50% of total active users)
There are no rows in this table
*Active users: Users who have ordered more that 6 times for 3 consecutive months


Results

Metrics
0
Type
Metrics
Control
Auto suggestion
Percentage change
1
Input
No of users
2,00,000
2,00,000
NA
2
Output
Total orders
12,35,451
12,15,478
-1.6%
3
Output
Total suggestions shown
0
10,52,258
NA
4
Output
Total suggestions opted
0
8,54,236
NA
5
Output
Average suggestion opted per user
0
~1.7
NA
6
Output
Average Suggestion added then removed from cart
0
~0.8
NA
7
Output
Number of users who were shown the suggested product
0
1,67,830
NA
8
Output
Number of users who added the suggested product in their cart and checked out
0
98,834
NA
9
Outcome
Total GMV
24,33,83,847
29,42,64,443
+20.9%
10
Outcome
GMV from Suggested food items
0
7,94,43,948
NA
11
Outcome
Total average order value
197
227
+13.2%
12
Outcome
Average order value from the suggested product
0
68
NA
13
Outcome
Total cart abandoned
1,34,673
1,53,645
+14%
14
Outcome
Tips to delivery partners(in Rs)
5,34,789
5,15,467
-4%
There are no rows in this table

Total orders decreased by 1.6%
when Autosuggestion variant is compared with control
Cart abandoned rate increased by 14%
when Autosuggestion variant is compared with control
Total tips decreased by 4%
when Autosuggestion variant is compared with control
AOV increased by 13.2%
when Autosuggestion variant is compared with control
GMV increased by 20.9%
when Autosuggestion variant is compared with control


The Hypothesis

I believe that if Swiggy Mobile App suggested a relevant Food item to the user during their checkout then the total GMV and AOV of the restaurants will increase as I believe that the users will add the suggested product to their cart and checkout.

has been proven with the Qualitative and Quantitative data although there were some negative changes in some key metrics and the root cause of dip should be analysed in order to make sure that this feature does not had a considerable negative impact before making this feature live to 100% of the users.



Challenges

58.8%
of the users added the suggested item in their cart out of total users who were suggested an item. Possible reasons could be
The feature was not able to gauge the attention of the user.
The feature did not suggest a RELEVANT item to the user.
The feature did not allow the user to know more about the suggested product.
Cart abandoned rate increased by 14%.
Possible reasons could be
The feature was creating confusion amongst users as to why this food item is added in the cart automatically.
etc
Total tips decreased by 4%.
Possible reasons could be
Users were expecting to spend 400/- and provide a tip to 30/- to delivery partner but since we suggested a food item of 50/- which got the attention of the user, the user added the suggested food item and compensated the same by not providing a tip as the spending capability of the user was 430/- initially which was pushed to 450/- with more value to the user and led to a decrease in tip.

Improvements

For each challenge mentioned above the following improvements can be done once we prove that the reasons as valid and significant

The feature was not able to gauge the attention of the user.
UI/UX improvements
The feature did not suggest a RELEVANT item to the user.
The recommendation algorithm can be categorised as
Item which has been sold maximum number of times in the last X days
Item which has been bought with the item that is already on the cart for the maximum number of times in the last X days
Promo of a item, market/advertised item/new item
Item that has been added to the card maximum number of times due to the auto-suggestion feature
Each of the above category can have a confidence level on conversion. The category with highest confidence level on conversion will be applied for the user.
The feature did not allow the user to know more about the suggested product.
The user should be able to view the photo of the suggested item along with its description.


Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.