A fun fact, 60% content of this essay is from my essay archive - I don't put all my content in public :). The rest of the essay I wrote during Holi weekend. Frankly, I had no plan of sharing this piece of content - I am sharing it due to an external trigger.
Let me clear the air - I don’t have any obsession with Zomato. But they are one of the fascinating companies in the Indian ecosystem. Also, I can’t write about ZILAs’ processes - those are trade secrets - and hence, testing my design thinking on successful companies is fun.
Whether you hate it or love it - you can’t ignore the paradigm shift of 10 minutes grocery delivery on business models targeted at metro consumers. I have come across 10 minutes internship platform, 10 minutes ambulance platform, and 10 minutes education. I did come across tweets around the possibility of 10 minutes food delivery, but I thought that was fundamentally impossible. Because when I was writing 10 minutes grocery delivery process - for my benefit - I concluded that 10 minutes of food delivery is out of the question. But it seems Zomato’s team is utilizing their knowledge directory, and they know something that I don’t - that is TRUE as well. :D To understand how Zomato is solving this complex problem - we need to understand the 10 minutes grocery delivery process.
Over time I have learnt
I satisfy my curiosity after putting all the pieces together. And visualising A to Z in the form of a process. Therefore the external trigger forced me to go back and relook at the 10 minutes grocery delivery process from my archive.
My initial hypothesis - based on the process - The 10 to 15 minutes grocery delivery was always possible if you don’t have to prove unit economics from day 1.
When Dominos promises you a 30 minutes delivery, technically, 30 minutes is not the entire delivery time. It is actually - Order Confirmation Time + Cook Assigning Time + Food preparation time + Packaging and Invoicing time + Picking time + Delivery time. Now let’s say an average pizza cooking time is 12 minutes, and other activities, except delivery time, take one minute each (These folks are a bit lazy).
This means Delivery Time = 30 - (1+ 1+ 1+ 1+ 12) = 14 minutes. Now in many cases, you must have received your Pizza in 25 minutes or even in 23 minutes. That means the delivery time was 9 minutes or 7 minutes. Also, Dominos has mastered the art of 30 minutes delivery in tier-1 cities by having more outlets in micro markets and setting a clear radius because Dominos can’t take the risk of exceeding the delivery time - 12 to 14 minutes. We need to focus on a few key points
Dominos has a super high gross margin on Pizza - Around 50%
After giving free delivery for many years - Dominos started charging delivery costs, INR 30, from July 2020 to increase their overall profits (Not Cool!). If you look at their last 5 years' profit and loss statement - they have been making net profits of a few hundred Cr. each year.
Now tell me - 10 minutes grocery delivery is possible or not? Technically, the feasibility of 10 minutes grocery delivery was always there. It was never a front-end (From store to customer’s address) problem, it was a problem of back-end - Order receiving, Invoicing, picking, packaging etc. If you can reduce this from 4 minutes to 1 minute, and keep dark stores based on micro-communities - you can always deliver groceries in under 10 minutes. I am sure you can understand the importance of predictability and consistency that these companies shall achieve over time.
Now delivering groceries under 10 minutes consistently is operationally complex - that you can solve by applying first principles.
Let’s solve this problem by working backwards - Amazon’s classic framework to achieve impossible goals.
Statement: We must deliver groceries in under 10 minutes every time.
Apply the first principle and work backwards - we will have our solution. First, Let’s divide the problem into two parts
Front-end: From Darkstore to Customer’s address
As I said - it is more of a back-end problem, so we will go deep into back-end detail and give you a picture behind the curtains.
These companies use an order alarming device. Orders placed by customers get assigned to the nearest dark store - it is the easiest part, technically.
The device starts beeping for each order received from customers. The customer's order includes product details and the coordinates of products ( I have explained this below) inside the dark store. As the idea is to optimize every second. These companies' Darkstore racks are in matrix form - rows and columns.
You can see each shelf is marked in numbers (1, 2, 3, 4,....N) and each rack is marked Alphabetically (A, B, C, D……) - each box is named [1, A], [2, B], [5, E]........ [N, n] N = Natural Numer, n = Anphebet.
Well frankly, this method is not new. Most inventories led eCommerce companies stacks their high shelf life SKUs in their warehouse through a similar process. They have nomenclature for all the pallets and racks. That’s how they keep/track Inventories in a massive warehouse - lakh sq feet.
Apart from beep - a sound - the order device also displays the information of products and their position - [N, n]. As the size of these Dark Stores is less than 4K sq feet. The entire back-end process, receiving orders, accepting orders, and packing orders, can be completed in 1 to 2 minutes. After that, orders are ready for the delivery partners after invoicing. Let’s say invoicing takes 30 seconds to 1 minute - with an average delivery time of 4 to 5 minutes. An order can be delivered to a customer in 7 to 8 minutes. Once you solve the backend problem and keep the entire process under 1.5 to 2.5 minutes - it becomes easy to deliver under 10 minutes.
The frontend optimization can keep the experience consistence. The back-end optimization part of the business floats heavily on technology. These companies also measure order picking, packing, invoicing etc. [I can’t go deep further, sorry]
Front-End Problem: There was not much to optimize apart from keeping the Darkstore near to customers’ density and assigning fixed riders for a micro-market. The frequency and repeat rate will allow the assigned riders to memorize customers' locations overtime - high predictability and consistency. This is unlike Zomato's current method, where each time a random delivery partner get assigned for your orders.
This might sound simple but building a sustainable tech platform and keeping up the expectation is super challenging. And the 10 minutes grocery delivery have shown tremendous promises.
Without wasting much time let’s quickly understand what Zomato is promising - food delivery in 10 to 15 minutes. Initially, I thought this is a crazy idea, and it will require some fundamentally architectural redesign. To understand the 10 to 15 minutes food delivery process - we need to understand Zomato’s current food delivery process.
Zomato’s current delivery process: Customer placed an order to a restaurant - On average the distance can be 0.5KM to 8KM - A delivery partner will get assigned based on restaurant lat long and delivery partners lat long, Delivery partner reaches to restaurant, display the order through mobile and wait for food preparation, restaurant partner prepare the food, do the packaging and invoicing, delivery partner collect the food, and rush towards customer address. Because the delivery partner is assigned randomly, based on distance, he might call you a few times.
Based on this process 10 to 15 minutes of food delivery sound crazy. But Zomato is a unique company - they don’t want to repeat the mistake of Paytm (There was a time when Paytm was the market leader in online payment through Wallet. When UPI came, they ignored it initially, and the rest is history). Also, the importance of Zomato’s knowledge directory. They know a lot about their 50 million users - micro-communities, food choices, food ingredients, high demand food. So they know a lot about stuff related to food. :)
Suppose Deepindar is worried about the growth of 10 minutes grocery delivery and getting this feeling what if the same company will start 10 minutes food delivery - that will be a question of existential threat. If a different platform will deliver food in 10 to 15 minutes, no one will order from Zomato. And I am hoping Deepindar might have got all these clues from Christensen’s Books - The Innovator’s Dilemma and Competing Against Luck (Two of my favourite books). He will reach out to his team and share his concern. As the concern is genuine - they would have concluded, we need to find a method to deliver food in 10 to 15 minutes. And suppose, I am that program manager (That is a demotion from the CEO ;)) - who has been assigned with this responsibility. This would have been my thought process.
I would have not spent any time identifying stakeholders because in the case of Zomato it is clear - Restaurants, Delivery Partners, and Customers. So I would have started with DATA about
Average Dishes preparation time
Dishes demand based on the micro-communities
Gross margin on Dishes
Dishes demand from restaurants
The average time between order assigned to a Delivery Partner and reaching to restaurants [First-mile delivery]
Average time of order pickup by delivery partners and making the first call to customers for the location guidance [Middle-mile delivery]
Average time from making the first call to food delivery to the customers [Last-mile delivery]
With these data points - I would have enough understanding about which part I can optimize and make sure the food is getting delivered in under 10 to 15 minutes. Let’s call the above points key areas - to build a sustainable 10 to 15 minutes food delivery process. The key areas should be in sync. And the failure of any key area means - high costing, bad customer review, unhappy delivery partners and restaurants partners. For example, if a dish preparation time can’t be less than 6 to 7 minutes - it would be impossible to make delivery in under 10 to 15 minutes. Or When I will create a restaurant partner selection doc for this programme - not knowing the strength of the restaurant for a dish - it would be an injustice to the customers and restaurants partners. The bottom line is - everything should be in sync.
The above data would have helped me in taking the following decisions for this programme:
Selection for dishes (If you have: chopped onion and green chilli, fried mungfali, and Wet Poha mixed with Salt and sugar for 15 minutes, you can prepare the Poha dish in 2 to 3 minutes. I have learnt this a few months back - everything is a process ;))
Selection of micro-communities (Micro-communities that order those selected dishes with high frequency)
Selection of restaurants/cloud kitchen partners (Restaurant partners those have the highest demand for selected dishes - for me, it is important to have a compiling sales pitch in the benefits of restaurant partners otherwise they might not find value)
As always - I would have divided the problem into two parts
Backend (Before food handover)
Front end (From Centralized restaurants to customer address).
Let’s start with Backend: Restaurant location, food preparation, Packaging, Invoicing First-mile delivery,
If you can recall - I talked about 10 to 15 minutes of food delivery sounds crazy based on the current architecture. Because it is super difficult to make sure the selected dishes can be delivered from multiple restaurants’ locations by randomly selected delivery partners to a selected micro-community.
So based on micro-communities data, set up a kitchen warehouse for selected restaurants - In a centralized location, all the selected restaurants/cloud kitchens will have their setup. More like a restaurants/cloud kitchens warehouse.
And food ordered by customers will get delivered from the central location. Having a centralized location will eliminate the first-mile delivery need. This is a great point to consider - I am sure you have heard of the magic is in the detail. The ignorance of the details is one of the reasons companies break after a time. To explain why the elimination of First-mile delivery is a great deal. Suppose the average First-mile delivery distance is 1KM: Delivery partners drive to restaurants from their respective positions after receiving customers' orders. Zomato does 1.3 million orders a day i.e 1.3 million KM per day as the first-mile delivery. If the milage of delivery partners' bikes is 40KM/lit, it is 32,500 lit fuel per day. INR 100 per lit fuel cost = 3,250,000/per day in the first-mile delivery cost. Per month first-mile delivery cost = 9,75,00,000 ~10 Cr. Wow!! You might be thinking about how this cost is related to Zomato? But again - the magic is in the detail. This is the reason investors allow startups to burn money and achieve critical mass. These folks certainly understand the economics. And hence a small startup with hyper focus and ultra detailing ability shall always beat mediocre deep funded/incumbent.
The pricing structure shall be also different because earlier it was restaurants/cloud kitchen own space and setup. In 10 to 15 minutes of food delivery, Zomato is offering that - I might have proposed multiple models such as freemium, higher take rate, profit-sharing etc.
Since all the restaurants/cloud kitchens are at one location - I will propose central packaging and invoicing systems. And this would have saved time and money.
Great! For selected dishes, we have achieved a super-efficient system that can get the food prepared, packed, and ready to deliver under 3 to 5 minutes.
Front-end: Middle-mile and Last-mile
Here I would have followed dedicated delivery partners for micro-communities. This would have eliminated the last-mile delivery overtime - the delivery partners would have memorized locations, cutting significant middle-mile.
I would have tested the model based on the numbers of Orders a day more like a simulation and had my premortem ready to present.
I thought of giving you a simulation model to articulate the viability and breakeven in this essay. This essay is already an eight-minute read.
Even though, on paper the idea of delivering food in 10 to 15 minutes seems possible - I feel this is still a risky proposition. But that’s where Zomato is signalling to the industry that - they still believe in taking ultra risks. Good!!
Time will tell whether the 10 to 15 minutes food delivery will be a success or failure - the future of food delivery for metro consumers seems exciting. What are your thoughts on this? I would love to hear.
Published on: 21-03-2022 at 23:01