Instamart’s take on lightning speed, Episode I: Your go-to items

Amrita Ganguly
Swiggy Design
Published in
6 min readMay 31, 2022

Instamart has set its sights on breaking ground by pioneering lightning grocery delivery in India. But the efforts are beyond just lightning speed delivery.

How it started:

Instamart’s foundational vision was not only to deliver groceries within minutes, but also aimed at designing our system to accelerate users’ buying journey.

With that thought in mind, we had previously built our “Frequently Bought Items” collection, which reflected mass purchase data or popularity, curated through numerous experiments. The principle for this collection was driven by customer intent, which was one of three pillars when we set out to design Instamart experience. This collection was enriched with a popularity-backed data science model, which went on to improve our add-to-cart rate by ~4%.

Frequently Bought Items Widget

While this was a good interim step, the holy grail for this collection has always been to make it much more meaningful and relevant for our repeat users. Every individual comes to our platform with varied needs and wants, and the intent for this collection was to evolve, to capture individual users’ buying patterns and history. It was only inevitable that it transitioned to a smart engine that personalized items to cater to that individual.

How did we approach it?

Think about the occasions where you have frequented your favourite restaurants or cafés, and the people serving you there know your usual orders or purchases. We wanted to bring the same experience to our customers on the platform.

Over the last year, our team waded through an ocean of purchase data, gleaning buying behaviours of the customers. While evaluating the insights, we uncovered a few patterns from both qualitative and quantitative standpoint, namely:

  • how frequently were they re-purchasing certain products, or what was the time gap from their last purchase of a certain product?
  • what time of the day, week, or month they were purchasing certain products or from certain categories
  • how many items at an average have re-purchase status in an individual purchased items bucket
  • what is the attention span of a browser-first user?
  • …. and many more small details that make up the bigger picture!

Based on our findings from the research, the data science team designed a system built on 4 parameters: periodicity, frequency, recency and popularity.

Trial & error:

Is it possible we reach our conclusion before experimenting and observing? The answer is mostly no. And this feature is not an exception.

Before we concluded on our current interface, we repurposed the same UI and ran some copy experiments for validation. There were some problems we hypothesised with the same UI as the FBI widget (frequently bought). There was widget blindness as it was the same as before. We concurred that users won’t be able to differentiate the value proposition of the new engine, as they looked the same.

We went through a phase of iteration that we did for the copy to capture the essence and for the UI to exercise our most important design principle — noticeability.

Generally, architecting simple things are the hardest, and this final exercise was going to test all of us. We had (many) spirited debates on what to name this new baby! We wanted the name to embody the exact characteristics of this collection, which is enriched with the goodness of emotional design on top of a data-backed AI layer. Soon we concluded our brainstorming, and this progeny had an apt name — “Your go-to items”.

How it’s going:

At Swiggy, we believe in the value of “Do more with less” — we were able to envision, condition, capture and execute using this mantra. We managed to achieve a big impact by focusing on knowing our repeat users and what we could do to make their ordering experience more efficient.

Below are screenshots of various real and repeat users’ go-to items, personalized just for them. It provides a fascinating insight into how different people have different ordering patterns and preferences.

What this means for the FBI collection’s fate. The FBI widget still stands tall for our new users who come to our platform. After they have shopped a while from Instamart, we present them with the go-to items collection, assimilating all the relevant information we need to power this engine for them.

Never Settle:

Another Swiggy value that we take pride in, is constantly pushing the limits. So what else could we do to cater to our users?

It turns out that a reasonable chunk of the orders is made for a different address or city, for their parents or friends. We added location as a parameter in our “never settle” release. The screenshots below illustrate a user’s go-to items in the city of Bangalore (home city) and in Chennai (parent’s address).

Left: Bangalore. Right: Chennai

The extra mile:

So what does this collection show and not show. Simple, it does not show up anything that you might not have bought, ever. However, as part of our tobacco guardrails, we barred tobacco-related items to be populated here. A topic for another day :)

What did we achieve?

Apart from earning our repeat users’ appreciation, we also moved the needle quite a bit.

We improved the add-to-cart from this collection by more than 2X

Which is a pretty good start as a validation that our customers were finding the suggestions relevant.

We reduced cart building time by more than 15%

This means that our users can utilise their time doing things that are important.

Always curious, always learning:

So what’s next for this collection, you ask? As we continually learn more about our users and their needs, we are training this smart engine to incorporate more patterns. So that individuals coming to our platform can feel that we design systems for them — one that is functional on the outside, but belies emotion on the inside.

The people who have been an integral part of this journey are: Mithun TM, Sumit Gupta, Nitin Tanwar, Sai Aravind, Somasekhar Hy, Bhavi Chawla, Anirudh Sharma and a big shoutout to Raunak Singh for making this into a reality.

This project is one of many to accelerate the entire checkout experience (from pre-order to post-order) for our users. After all, Swiggy’s motto is to offer unparalleled convenience.

Illustration Courtesy: Shivam Thapliyal

--

--

Amrita Ganguly
Swiggy Design

I usually write a lot of personal and professional stories here just to save them as drafts, and procrastinate over grammar. Design Architect @Swiggy