​

Problem statement

Decrease in order volume from the last 3 months by almost 40%

Dig into the data to gain insights Run the right experiments and come up with possible solutions to improve orders in the next quarter

​

Overview

Swiggy's order volume has decreased by 40% in the last 3 months. Data from across cities and categories is available - Mumbai, Bangalore and Delhi seem to be the major causes - contributing more than 90% to the overall drop of 5.3% (rate+mix effect)

Diving into the categories gives an insight into particular categories in particular cities causing the downfall

​

Data insights

Bangalore - Munchies & Snacks conversion and mix is the main dragger

​

Delhi - Similar to Bangalore, munchies & snacks are the main cause - but here it's only the impact of the conversion rate

​

Mumbai - saw a dip in the rate of "other category" items

​

Experiment

Bangalore - the goal is to increase both conversion and mix of munchies across categories

• Bundle offers for munchies with other items like fruits and personal hygiene - free delivery on ordering a bundle of munchies for orders above 99/ (normal free delivery is for orders above 199/-)

Delhi - the goal is to increase the conversion of munchies and snacks

• Offers on bundles of munchies and snacks - Buy 3 get 2 free

​

Hypothesis

To increase the conversion of munchies and snacks in the overall mix of items and increase the overall volume, we will offer discounts and bundle offers for buying munchies. We expect that the order volume will go up with this

​

Null hypothesis

Offering bundled discounts on munchies and snacks won't increase their conversion

​

Goals

Increase the conversion rate of munchies and snacks in Bangalore and Delhi leading to an increase in the order volume and hence leading to an increase in revenue

The goal is to increase the conversion in Bangalore by 6% and in Delhi by 4.2%

​

Success metrics

Increase in conversion of munchies and snacks in Bangalore and Delhi

Best case Increase in conversion in Bangalore = 6%

Best case Increase in conversion in Delhi = 4.2%

Best case Increase in mix in Bangalore = 20%

​

Experiment design ​

Variation design

Control group - No additional offer on munchies and snacks

​

Offer to be displayed in the cart page of Swiggy Instamart - after the user has added all the required items to the cart and is one step away from the checkout

Group A - 1+1 offer on popular snacks and chips and free delivery over 200/-

Group B - Suggest snacks when the cart value is less than 200/- -> nudge with a free delivery option for cart value > 200/-

Group C - Offer 60% off on munchies if the cart value is above 500/-

​

Success metrics

Increase the conversion rates for munchies and snacks in the cities of Delhi and Bangalore

Worst case - increase the conversion to the previous levels as of Dec'23

​

Sample size

Bangalore - 27000 sample size for a 6% detectable effect in conversion

​

Delhi - 73000 sample size for a 4.2% detectable effect in the conversion

​

Since the monthly orders in these cities are in the millions - it shouldn't be an issue for us to test this sample of users

​

A/A testing

Ensure that all the instrumentation is the same with the control group and test group to ensure no bias whatsoever

User segmentation

Ensure random segmentation of users without any bias and any scope for inherent differences

​

​Experiment Results

Ran the experiment for the 4 groups - Control and groups A, B and C in the cities of Bangalore and Delhi

The experiment was carried out for one month and the conversion results are as follows

​

​

Group C had the highest increase in the conversion delta throughout the experiment

Release Decision

Nudge high cart value customers with extensive offers on snacks to increase the cart size, and conversion of snacks leading to an overall increase in revenue

Experiment learnings

A significant drop in the conversion of snacks could be attributed to no offers on these segments. The customer with a cart value close to 400-500/- expects additional value-based addition to his cart from the company and a nudge of higher discounts on a discretionary category can increase its conversion leading to an overall increase in revenue to ​

Next steps

Track order values, cart values etc. to ensure the conversion rates of other categories that are already healthy do not drop off for any reason

For this experiment - track the conversion rates of the snacks category to reach its December peak ASAP

Conclusion

We observed a 40% drop in orders in the March 2024 quarter compared to the December 2023 quarter

On diving deeper into the data, we observed a huge delta in the conversion rates in Mumbai, Delhi and Bangalore. Looking at the category-level data in each of these cities, we could see a drop in conversion in the munchies and snacks category

So we based our experiment on improving this conversion.

We experimented with different types of offers to customers to nudge them to increase the order value and conversion.

The final release decision is to promote discounts on munchies to customers with a higher cart value, with an insight to attack the higher discretionary and value spending of these customers to increase the overall revenue of swiggy instamart. ❤