11: Your Order is on Time

Introduction

Over the past few years, companies like Zomato and Swiggy have gained a large market share in the food delivery ecosystem. They handle close to one and a half million orders every single day. It’s clear that any policy decided by these companies with regards to delivery times would have a clear and massive impact across the wide regions in which they operate.

This project aimed to explore whether the short delivery time policies of these corporations resulted in their delivery partners rushing to be on time.

Related Work

There has been a lot of work done by journalists in conjunction with researchers dissecting the exploitative tactics used by these companies to grow their customer base. This particular piece from The New York Times shed light on the hectic working conditions that the delivery partners operate in.

Most of these investigations however have been focused on the United States and are from a couple of years ago, from a time when these apps were just getting started and the ecosystem was relatively immature. Since then, there have been significant changes in the regulations surrounding these industries and the business models they rely on. Addioantly, the regulatory framework would also change across national boundaries which would result in the regional versions having different policies compared to their American counterparts.

Research Questions

The primary question that this project asked was whether delivery partners were likely to overspeed due to the apps policies? Do they feel any pressure to deliver the order in a narrow time frame? We expanded our inquiry to also include questions about working conditions and their view of the fairness of the policies.

To answer the questions raised, we started by collecting data about orders. The exact pipeline that we used can be seen in the flowchart below.

Here, we began by collecting order data from users by collecting the emails that they received after delivery of each order. We processed and extracted specific attributes, including but not limited to order and delivery time, restaurant, etc.

After this, we used Google Maps API to get location information about these restaurants and get an estimate about the time it would take to travel from there to the delivery location. We solely focused on orders delivered to IIIT so the delivery location is the same for each order. For the analysis, we grouped the data by the restaurant so that the distance could be normalized across a group and we can get a base estimate about the travel time from that particular location.

To complement our analysis of the collected quantitative data, we also conducted interviews with the delivery partners. These were done in an open ended format and in the native language format of the partners so that they were at ease expressing their thoughts and views on the topics we discussed. These interviews allowed us to include their perspectives and get a much more thorough understanding of the ground reality which might not have been possible from the narrow window provided by the data that we had.

We initially started our data collection by building a web scraper that extracted the required data from the user’s account. This presented some challenges in the context of privacy as users were hesitant to provide their account credentials. Due to this, we pivoted to the less invasive method of asking for their emails where they had complete control over the information that they were sharing.

Results and Discussions

The final results from this were different from our initial hypothesis. Owing to the previous work done, we were expecting the deliveries to be made with a rush. However, from our analysis of the data, we found that the vast majority of the orders were delivered in a time frame that ensured that the travel speed was well below the speed limit inside the city.

We found that the apps actually have specific policies in place to avoid encouraging rash driving from the partners. They intentionally don’t share the exact promised delivery time with them so they are not rushing to meet the deadline. Additionally, in case an order is delivered late, the drivers are not penalized. The cost is mostly borne by the restaurant as the primary cause for delay is the preparation time of the food. 

Conclusions

In conclusion, we can say that there is no significantly observable rush that is caused by the apps. Our interviews also concluded that the delivery partners are satisfied with their working conditions and don’t feel pressured to dash to their deliveries.

Future Work

This project solely focused on the food delivery ecosystem and was limited to the established corporations i.e. Zomato and Swiggy. There is a new rising industry revolving around quick delivery. In fact, during the course of the semester, many grocery delivery services have started one of which is Ola Dash. It is clear that there is a lot of interest in capturing the market, from players large and small. 

Thus, the methodology deployed here can be used to explore the state of rush in these corporations whose entire business model revolves around rapid deliveries.