With the biggest party of 2016 just on the horizon, we’ve got a brand new feature to help our wonderful users organize the evening better – online pre-ordering! That’s right, you can order your New Year’s Eve dinner right now, and it will be delivered to you fresh on the 31st. Here are the steps to a hassle-free year-end celebration.

Pre-ordering will be available from midnight on the 28th, and will be active up to midnight on the 30th. We are partnering with a select few restaurants to make sure your pre-orders are delivered on time.

In case you don’t place a pre-order (we all have those “last-minute-impromptu parties”), you can still use the regular Zomato online ordering feature for your last meal of the year. However, given the surge in demand, these last-minute orders have a high chance of taking an unreasonably long time to reach you. It’s also important to note that our online ordering service will go offline at 9:30 PM on the 31st. You’ll still be able to call restaurants to order, but given our experience over the last couple of years, restaurants have a tough time efficiently dealing with orders coming in post 9:30 PM on New Year’s Eve.

Still not convinced? Here’s a little more motivation for the party planner in you – most restaurants are offering huge discounts on pre-orders! So help the small business owners (and us) with a bit of prior notice, and we promise to make it an amazing experience for you.

P.S. You’ll only be able to pre-order for New Year’s Eve for now. But in the coming weeks, we’ll be making pre-ordering a permanent feature on the Zomato app.

Today marks 4 years of our international presence. On the 1st of September in 2012, we launched our very first market outside of India – Dubai.

Four years on, we claim to be the #1 player in our space in 18 countries (of the 23 we’re in) in terms of monthly unique users. We say ‘claim’ because it’s hard to substantiate, given the limited availability of data about our competitors in these countries.

Given the data we do have access to, this post is a reflection on how our international foray has worked for us, and where we’re headed as a global consumer internet brand born in India.

To start with, here’s some data on some of our cities for you to binge on –

What’s our biggest learning after all these years of running a multi-country business, you ask?
It is that everybody should find their own way to determine the real market size for their business. Nobody knows your business as well as you do, so don’t believe the press, the VCs, or market research companies. Listen to everyone, but then decide for yourself.

Why do we say that?
We often get asked – “Auckland? Seriously? Isn’t the market size too small?”

Since we’re on the topic – we love Auckland. Not just for the people and the city, but also for the market size. Before we go into detail on how the market sizes of our various cities stack up against each other, here are some definitions you need to know:

• Market Size = # of listings in the city x average size of a restaurant x utilisation rates x average ticket size of a meal
• Indexed Market Size: Assuming the average size of a restaurant in a city and utilisation rates are the same across the world (and they are, more or less), the relative market size can be defined simply as # of listings in a city x average ticket size of a meal
• We measure Relative Market Size using Delhi NCR as the benchmark (where Delhi NCR = 1). Why Delhi NCR? It’s where we have the richest and most complete restaurant information, making it a fair yardstick for us to compare other cities and their importance as a market.

Now, here’s how the cities we saw in the infographic compare to one another in terms of relative market size. The Strong/Medium labels indicate how we view our position in that market, in terms of a multitude of factors – including, but not limited to – our user base, client penetration, revenue, brand recall, and competition.

So you see, despite being physically smaller (you can fit ~42 Aucklands into one Delhi NCR), and having fewer restaurant listings (~6k vs ~16k), Auckland is actually stronger than Delhi NCR in terms of market size.

Does loving Auckland as much as we do make more sense now? :-)

–

The road ahead

Over the coming months, we’ll be focusing on 21 cities across these countries, improving our presence and position in newer markets like Kuala Lumpur (where we still consider our position of strength to be comparatively 'Low’), and high-potential markets where we’re yet to move into a position of strength, such as Sydney and Istanbul. A lot of this will be driven by operational improvements on the ground, backed by laser focus on our consumer and business products. 

For those wondering – it’s unlikely we’ll look at establishing ourselves in markets like the UK and the Americas using the same approaches we’ve taken in the past. These call for a slightly different strategy, and can only realistically be solved for with very strong and smart tech. There’s a lot of that in the pipeline, so watch this space.

–

Looking back, are we glad we decided to take the leap and go international when we did? Certainly. Has it been a roller coaster ride? We’d be lying if we said it hasn’t. Have we learned a lot from it? Way more than we would have learned from sitting in India, reading the Expansion 101 manual.

Most importantly, though – has it worked?

Given that we’re traffic leaders in all our 21 focus cities and are already delivering USD 3bn worth of value to restaurants – we’ll dare say it has.

P.S. The data in the infographic is sometimes from various unreliable sources; we have made an effort to be borderline pessimistic/realistic. Please consider these numbers as guesstimates.

One of the most important things we’ve been working on over the past year is making the process of online food ordering and delivery as simple and hassle-free as possible. While there are some aspects that we think we’ve done well at – bringing customers the widest choice of restaurants in the country, and building a personalised and incredibly easy-to-use app – there are some areas that have immense room for improvement, the most significant one being delivery tracking.

Today, we’re excited to announce that we’re taking a big leap forward in bolstering our delivery-tech capabilities with the acquisition of Sparse Labs, a logistics-tech startup that helps restaurants track and optimise their in-house delivery fleet.

Sparse Labs has developed a lightweight Android-based app, which transmits a rider’s location to both the restaurant and the user in real time. The rider can use the app to receive route information, which helps ensure quicker and more efficient deliveries. The system also allocates orders to riders based on their proximity to the restaurant, and uses machine learning to identify a rider’s familiarity with a neighbourhood to further optimise delivery efficiency. Restaurants also have the option of using a proprietary GPS tracker developed by Sparse, that can be fitted onto bikes.

Pankaj Batra, the founder of Sparse Labs has bootstrapped his product for over a year now, and is much loved by all his customers. He will continue developing his vision at Zomato, and the rest of the Zomato team is very excited to lend all their support to Pankaj going forward.

Joining forces with Sparse Labs will allow us to significantly improve the food ordering experience on the app, by giving users real-time GPS-based status updates on their order. While we were already working on making this feature available for deliveries handled by our logistics partners, Sparse Labs will now help us enable delivery tracking for restaurant-owned fleets as well.

At the restaurant end, this technology will help make deliveries highly cost- and time-efficient, allowing them to optimise delivery routes and ensuring minimal wait time for riders. We’ve always maintained that the most cost-efficient delivery fleet is the restaurant’s own, where they can utilise the same staff during off-peak hours for back-of-house and marketing activities.

Sparse Labs will be renamed as Zomato Trace, and we will be rolling out the service free of cost to our restaurant partners very soon.

The number of food orders being placed via Zomato is growing by the day, and we’re constantly thinking of ways to improve and simplify the user experience. Zomato Trace is a big step in that direction, and we’re looking forward to get this up and running to scale in the upcoming weeks.

For the past few days, the internet has been awash with outrage over the incident at the legendary Mocambo restaurant in Kolkata. For those not in the know, a guest was told that Manish, her chauffeur, couldn’t be seated along with her at the restaurant. The restaurant’s reasons for this supposedly swayed from Manish being ‘inappropriately dressed’, to them alleging he was drunk, and then saying a ‘roadsider’ like him could not be seen dining amongst the city’s ‘elite’.

The incident has gone massively viral on social media, and everyone is up in arms against the restaurant. Since then, there has been a flood of reviews for the restaurant on Zomato – as many as 5000 in a day – and their rating has dropped from a very respectable 4.3 to a disastrous 1.8.

Now, many (if not most) of these reviews have been written by users who admit they haven’t been to the restaurant, and probably never will. Our policies and algorithms are tuned in a way that detects and deletes these reviews automatically.

But there’s a catch - whenever there is an outpouring of negative reviews for a particular restaurant, our anti-spam systems take a pause, and flags the issue to our moderators as a potential “social issue”. And a social issue this is.

In these exceptional cases, we have to deeply analyse the situation at hand, and then take a call on whether or not we should delete these reviews by marking them as junk/spam, and for not reflecting a real, first-hand experience with the restaurant. In this case, it was a very hard call for us.

Here’s why – while we agree that Zomato is a technology platform that shouldn’t take sides, and should stick to the rule book that we have created for ourselves, in such social cases, we have to look deep within us and ask ourselves the question – “what is the right thing to do?”.

So what, according to us, is the right thing to do?

We think every hospitality business should have basic ethical guidelines and principles. If the customer’s side of the story is true, it is deeply upsetting for us. We agree that the management reserves the right to refuse admission to someone, but there’s a polite way to turn a customer down, and then there’s an absolutely unacceptable way to turn someone away.

This incident, if indeed it happened verbatim, highlights the elitism of some hospitality businesses in India. Of all the countries that we operate in, India is the only one in which we hear of such incidents on an ongoing basis. It’s taken us centuries to supposedly move away from the class system, but traces of that still remain within our society. Such incidents is when things blow up and the chinks in the armour of our modern society become visible.

However, theoretically, the event may not have panned out exactly the way we were told it did, and the details posted by the customer in the original Facebook post could have been exaggerated. The customer could have just had a long and tiring day, and the fact that the restaurant exercised its right to admission – politely at first, and then assertively – led to her venting her sentiments on social media. And that, in turn, sparked a massive backlash against the restaurant.

We all know the social media mob attacks really well. And that’s what is happening here. There’s a mob attacking and voicing their opinion on an issue based on one side of the story.

Our goal is to enable people to make informed decisions on whether or not to dine at a place, based on the real and personal experiences of people who’ve been there. Therefore, an individual’s opinion/review of a restaurant should be reflective of their own experience there. We don’t want Zomato to be used as a tool for mob attacks – there are enough platforms being used for that already.

So again, what is the right thing to do?

The customer is entitled to her opinion, and has left a review on Zomato – that review will count, and stay on Zomato forever, for potential customers to read and decide whether they want to visit the restaurant or not. The business owner also has the means to reply to that review to present their side of the story. At the end of the day, any potential customer can decide whether to dine at that place or not.

Every other review, as a result of the mob behaviour that some of us show without understanding the real issue at hand, will be deleted very soon, and we will re-enable our automated anti-spam on this restaurant’s page again.

That, according to us, is the right thing to do – and this is what our anti-spam policies were originally designed for.

———

Note: Mocambo is not an advertiser on Zomato. However, our review policy is – and will always be – completely neutral. We do not treat advertisers differently from non-advertisers in any way when it comes to reviews and ratings. This has been, and shall forever be, our guiding principle as we continue building Zomato into the future.

This post isn’t for the easily offended. If you’re the type of person who puts 2 and 2 together and gets 5, I suggest closing this tab and reading no further.

–

I was chatting with Oytun, our Turkey country manager, this past weekend, because I’ve been quite concerned about what’s going on in Turkey. I was worried about our team’s safety and well-being (in addition to being concerned about the well-being of the world in general). Oytun reassured me that Turkey is a resilient country, that it always bounces back in about 2 weeks, and that I shouldn’t be worried.

That thought stayed with me, and I figured that our traffic data in Turkey should very easily be able to prove the hypothesis that Turkey bounces back in two weeks. After all, the frequency of eating out is a fair indicator of how a country is reacting to a particular terror attack or situation.

For those not in the know, here’s some context on Turkey. The country has seen more than its fair share of unrest in recent times. Since the start of 2016, they’ve suffered a series of attacks and events that have put the country on high alert. Istanbul alone has seen three violent attacks, and been at the centre of an attempted military coup.

Every time one of these events occurs, we see a country that reacts the way any country in its place would: businesses pull down their shutters, people stay indoors, security measures are beefed up, and day-to-day life is disrupted. Sitting helpless thousands of miles away, it’s terrifying – all we can think of is the safety of our colleagues, and pray that normalcy will be restored soon.

–

Something that’s amazed us is how – every time, and without exception – our team carries on with business as usual (while taking all necessary precautions, obviously). We can put this down to their incredible resilience and work ethic.

But what about the rest of the city? Does it actually take Istanbul as a whole just two weeks to get back up on its feet and get back to life?

One rarely gets to see that from an outsider’s perspective (and we were curious), so we started digging through Istanbul’s traffic patterns on Zomato. Here’s what we found:

Our website traffic in Turkey drops by as much as ~50% in the immediate aftermath of an event, but recovers to its original point within ~10-15 days.

Here’s some data to illustrate this better.

• Attacks in Ankara (March 13th) and Istanbul (March 19th): The attacks in Ankara on the 13th of March had a strong effect on traffic in Istanbul, causing a sharp dip in traffic. Just six days later, on the 19th of March, Istanbul was hit by attacks. However, the effect it had on traffic was nowhere nearly as dramatic. By the 27th of March – just two weeks after the attack in Ankara – traffic in Istanbul had returned to the point it was at just before the attacks.
  

• Attempted military coup across Turkey, July 15th-16th: Between military-imposed curfews, widespread protests, and the declaration of a three-month state of emergency, Turkey seemed to be in complete disarray. By the end of the 16th of July, our traffic in Istanbul had dropped to almost nothing. But by the 30th, it had recovered the gap and then gone on to grow to 6% higher than it was on the 14th of July.
  

Interestingly – but not entirely surprisingly – delivery restaurants gained more share of the recovering traffic (compared to dine-out restaurants) after these events. Dine-out restaurants’ traffic was also near normal within two weeks. What does that tell us? That slightly fewer people want to head outdoors to eat in the wake of any unrest, simply because it’s safer to stay indoors.

–

The hypothesis was right: two weeks.

It takes Istanbul two weeks to bounce back from events that have rocked the city. It speaks volumes of the fortitude of the Turkish people, and their ability to get on with their lives.

When I mentioned this to someone yesterday, they argued with “Yeah, but many cities bounce back from adversity. What makes Istanbul unique?” To be honest, we don’t have a benchmark to compare Istanbul to – and we hope we never do. No city, and nobody, should have to suffer what Istanbul has in recent times.

This post is simply a salute to a city with immense courage; a city that refuses to take things sitting down. But more than that, it’s a tribute to my colleagues in Istanbul, whose unyielding and eternally positive spirit is an inspiration to us all.

This milestone achieved by our online food ordering team is dedicated to our mates in Turkey. Our thoughts are with our team and everyone in Turkey  – the people who show an undying spirit in the greatest moments of adversity.

–

“Will we get there? Will we not?” – Everyone at Zomato

Before the start of July, we knew that if we did our job well, we could grow our online food ordering business to a million orders a month in July. We kicked our planning into gear, and made sure that we did enough to get there. We came up with a plan – a day-by-day prediction of our order volumes for the entire month.

The first time we circulated a projection chart within the team was at end of day on the 3rd of July. This is what that chart looked like:

A few days later, we were pleasantly surprised to see that things were going according to plan:

I’ll be honest – such things don’t really happen too often at Zomato. We are usually over-ambitious, and often fall short of our own expectations. This time, us meeting our expectations was a result of two things – 1) setting realistic expectations (even though the expectation would need us to stretch ourselves), and 2) the team doing a great job and smashing out results one day after the other.

In general, my predictions have never been spot-on. So much so, even when I wanted just one kid, I got twins :-)

The fact that July had five weekends certainly helped our cause. Finally, at 8:23 pm on the 31st of July, we hit the million-order mark for July – and then we racked up another 20,000 orders before the night ended. This is what the chart looked like at the end of last night: 

The average order value (AOV) of our orders is holding well at ~Rs 480, and our unit economics have improved by 10% since the last time we wrote about it. We now make Rs 23 per order as contribution margin compared to Rs 21 two months ago. Our customer retention has also improved sharply; in fact, our cohorts are now looking even more like a smile – a shape that lets me sleep well at night.

Note that this cohort is for returning users – not orders from returning users. Orders from returning users cohort looks even more exciting because of the ever-increasing frequency from retained users.

Congratulations to our entire team, which has worked super hard to get here. But this is just the start, and we’re going to power on keeping in mind – what got us here, will not take us there.

We have a confession to make. Over the last two years, we’ve been so busy with our international expansion and our manic focus on winning delivery, that we let the basic USP of our core product – search and discovery – suffer a fair bit. Our own app feels a bit outdated to us, and we’re now beefing up (no pun intended!) our engineering and product teams to get our app back to being one that’s delightfully enjoyable to use and explore.

We’re starting with something our users have complained about for so long – “You know so much about me now. Why do I still see the same restaurants and recommendations as everyone else? Show me stuff you know I will like. Don’t be a dumb f**king app.” (Yup, users don’t mince their words when they care about you so much).

Listening is a very important trait. So we’ve made personalisation one of the top priorities for our product and engineering teams, and we are now working on building personalisation into various aspects of Zomato. Needless to say, we are starting with the online food ordering product, but we will very soon bring personalisation to our main search and discovery product as well.

To start with, we are drawing on data and behavioural patterns to personalise your ordering experience on Zomato, and making crafting your meal from all those extensive fifteen-page menus waaay easier.

First, we’re tackling the big question:

Where should I order from?

Earlier, restaurants in the online ordering flow used to appear based on factors such as popularity, and distance from your physical location. The problem was, this made the list of restaurants static, and you’d likely have to scroll endlessly to find that one place you really love ordering from. To overcome this, we’ve added a layer of personalisation that helps us put the restaurants you are most likely to order from right at the top of the stack. How do we know which ones to show you first? Your search, browsing, and order history on Zomato offer some very strong indicators, such as your cuisine preferences, how much you spend on orders on an average, and what you typically order at a given time of day. The algorithm also factors in how you’ve been using Zomato in general, and shows places you may not have ordered from yet, but might like to try (for example, your bookmarks).

This has improved our DAU ⇒ Checkout conversion on the Order app by a whopping 2.5% (and made us feel like we’ve been living under a rock all this while).

And then, we have the bigger question:

What should I order?

We have done two things to solve this problem.

First, we simply reorganised the dishes on each menu page for a restaurant, and put the ones you are most likely to order at the top. For example, if we notice a pattern of someone consistently ordering non-vegetarian dishes, we push non-vegetarian dishes to the top of every menu page on all restaurants that person views, so there’s less scrolling to get to dishes they might want to order.

Secondly, and most importantly, we made the Recommended tab on menus super smart by putting dishes we know you are most likely to order in it. These dishes are picked on basis of your past order history, other users’ order history, and corresponding order ratings received for those dishes. Our aim while designing this tab was to ensure that you are able to build an entire meal for two from that one tab, without having to swipe at all. It literally takes seconds now. I, for one, have placed my last six orders this week using just the Recommended tab. There’s some simple machine learning behind these recommendations, and we hope you’ll love this nifty new feature.

As expected, both of these changes have significantly reduced the time it takes to build an order – as of now, we’re seeing orders being placed 21% quicker, and we will further improve that number.

Like we said earlier, these are just baby steps towards making Zomato more personal and loved. We are going to get smarter, and will make up for all the work that we didn’t do for you over the last couple odd years.

There’s still a lot more to come, and we can’t wait for you to see it.

1% done.

For years, our top priority at Zomato has been to ensure that we remain a trusted resource for the millions of people who use Zomato every single day.

An important part of this priority is to keep biased content out of reach of our users. There are a number of automated checks, which learn and get smarter over time. In addition to that, our neutrality team ensures that we are constantly watching for new ways in which people try to game the system; this knowledge is fed back to our engineering team, which makes sure that we always stay one step ahead of business owners who write good reviews for their own business, and bad reviews for their competition.

Over the past year or so, with the growth of popularity of Zomato as a restaurant discovery product, we’ve seen a rise in the number of people creating biased content – a few agencies spread across the world are now offering services to artificially boost restaurant ratings on Zomato, and individuals are offering to write positive reviews for restaurants ‘as a service’. And business owners are happily buying into this, because of the lure of a higher rating on Zomato. For the most part, this doesn’t work. When it does, it means that someone somewhere has figured out a way to get smarter than the system that we have created.

We are constantly working on making sure that we are smarter than the folks who want to game the ratings on Zomato to make a quick buck. Over the last few months, we have been working to completely overhaul our bias-detection algorithms. On Thursday, 9th June 2016, we are rolling out a strong new anti-bias algorithm that will help clean up biased reviews retroactively, and also put sophisticated new bias checks in place for the future. While we can’t divulge too much (for obvious reasons), we’d like to highlight some of the key things that will change.

This is our Panda update, so to speak. Here are the details of what’s changing –

• Less deletions, more hiding. One might assume that deleting biased content from Zomato is the easiest thing to do. It is, technically, but anyone smart enough to try and game the system in the first place, will also be smart enough to identify patterns in what gets deleted and what doesn’t. So from now onwards, we will be deleting fewer biased reviews (but we will still delete the obviously biased ones), and algorithmically hide such reviews where they won’t be seen by most users. This will ensure that the user experience doesn’t get hurt, and spammers don’t have a clue.
  
  
• User credibility scores. The new algorithm takes a fundamentally new approach to user credibility, and significantly increases the confidence level with which we can predict a user’s bias in their reviews. The new credibility scores assigned to users increase or decrease their ability to affect the restaurant’s overall rating. Credibility now factors in users’ behavioural patterns on Zomato over a period of time, as well as the quality of a user’s content. There’s enough NLP and manually curated historical data in place for us to cluster users into various ‘bias’ categories.
  
  
• Moderation history. For ease of explanation, we’re going to use one of the oldest (and slightly more dramatic) cliches in the book – it takes many good deeds to build a good reputation, and only one bad one to lose it. What this means is, if a user has a history of having their content flagged and moderated, their ratings will automatically carry far less weightage in a restaurant’s overall rating. While Zomato is an open platform where we encourage people to write honest and unbiased reviews, we take abuse and bias very seriously, and will do what it takes to keep Zomato free of it.
  
  
• Recency and decay. Everybody makes mistakes, and consistency is not always a given. Business owners have often complained that they still get “punished” for a bad review which they received 5 years ago, and their streak over the last 2 years has been very good. Fair point. Going forward, the effect of older ratings and reviews on a restaurant’s aggregate rating will taper off over time, giving users a better idea of what they could expect at a restaurant if they were to visit today.

Those are, in a nutshell, some parts of the new anti-bias algorithm. If we tell you more, or give you more details, we’d be doing a disservice to our users by disclosing more than we should to the people we are fighting to keep Zomato bias-free.

Starting tomorrow, some restaurants may benefit from having a lot of biased, low-value content hidden from plain sight, while some may see their rating reduce due to lower ratings from (in)credible users. We hope users and restaurant partners alike will understand and appreciate that this is being done to improve the overall quality and credibility of ratings and user-generated content on Zomato for the long term.

We’ve always told every business owner we have ever gotten in touch with – “improve your business, delight your customers, and the ratings will take care of themselves”.

However, since you got to this point in this post, there’s one important point to remember. Zomato ratings are not simple averages. You cannot calculate the average of the ratings and reviews that you get and say “this rating doesn’t make sense”. And then there’s normalisation. Or classroom ranking as some people call it. For a city, we forcibly fit all restaurants and their ratings on a normal distribution curve. In short, there’s a lot that goes on under the hood to make sure that you get a true sense of what you can expect.

There’s plenty more to come that will help make Zomato an even better and more useful product. Over the years, we’ve kept working on ways to keep the bias out, and it’s something we will continue to do. Folks who try to beat the system will always try and find new ways to do it, so it’s important that we evolve faster – and this is a strong step in that direction.

Let’s face it – some things just go really well together. Shahrukh and Kajol. Rainy days and a great book. Or Pepsi and a delicious meal. We can’t really make the first two happen on demand, but for the third, we’ve got it covered for you.

We met the Pepsi team a couple of months ago to discuss interesting ways to partner. The very first few ideas were simple and dumb – for example, we went as far as asking ourselves “should we give a free Pepsi to customers who place an order above Rs 500”? We’re tired of “free”, so we decided that we need to dig into our creativity to find something more appealing for the user, and target a product integration towards customers who would value the product at the price it sells for. Not for free. At Zomato, we don’t love people who love only free.

Finally, we found the answer. And today, we’re super excited to announce the new Pepsi integration into the Zomato app, which makes it ridiculously easy to add a Pepsi to your online orders on Zomato when you need your cola fix.

The next time you’re ordering online on Zomato (from a restaurant that serves Pepsi, of course), you’ll see the little Pepsi icon and an Add button in two places – in the ‘Recommended’ tab on online ordering menus, and inside your cart. From here, you can just go ahead and add as many bottles as you’d like to order.

For a customer placing an order, having an easy way to add a drink from within the cart means less effort swiping to the ‘Beverages’ section, which usually sits at the very end of the menu.

As far as brand integration goes, this is massive for both Zomato and Pepsi. For one, the context in which the Pepsi brand is appearing couldn’t be more perfect. People open Zomato when they’re hungry or thirsty, and this puts the product right in front of a large, targeted audience specifically looking for things to eat and drink. Moreover, adding a drink to a food order is almost second nature – in the past month alone, with the summer at its peak, 72% of all online orders placed on Zomato included a drink.

–

Your app will need an update for you to see this integration. If you use the Zomato app for ordering, both iOS and Android apps (download now at www.zomato.com/mobile) already have this feature; it will be rolled out for our standalone Order apps over this week.

In terms of traditional advertising metrics, this integration gives an average of 40k targeted eyeballs to Pepsi everyday. And the Zomato restaurant finder apps (where this is live already), have seen a 140% increase in Pepsi orders. We’ll have more concrete results to share on what this has done for us, and for Pepsi, in a few weeks’ time.

Until then, here’s to making your orders on Zomato a little bit sweeter!

In earlier posts, we spoke about what our online food ordering business has taught us, and how we built the transactions DNA at Zomato over the past year despite having been a content-centric business so far. In this post, we will zoom in and look at the unit economics of our online ordering business in India, and how things look for us in the future.

Before we go on, let’s show you how order volumes stack up during various times of the day, and days of the week.

We look at our unit economics on a monthly basis. Now, if you slice out the peak volume on Sunday, those four hours obviously have extraordinarily great unit economics. But the trick of the trade is to make your unit economics work over an entire month/week, which factors in multiple peaks and troughs.

In the month of May, we processed a total of ~750,000 orders, continuing to grow at an average of 30% month-on-month. We’ve grown our online ordering business in a sustainable manner, focusing on high average order values, consciously avoiding the discounting route, and putting a great customer experience at the core of everything.

Of the ~750k orders last month, 80% of the orders were fulfilled by restaurants themselves (where we don’t do the delivery for them) – we call these Type A orders. 20% of the orders were fulfilled by us, through our last-mile logistics partners – we call these Type B orders. Keep this nomenclature in mind as you read on; this blog post analyses both Type A and Type B orders separately.

Here’s how everything stacks up for us.

The tldr; version of unit economics is in this chart below (all numbers in INR):

Note: In an analyst call a few days ago, we had said that the loss per order for Type B orders was Rs 2. That was because until last week, we used to average out our total processing/support cost over all received orders. However, now we have now started calculating processing/support cost separately for Type A vs Type B orders, and the loss per order for Type B orders comes out to be Rs 9.1

If you want to know more about the thinking behind each of these numbers, and how these are calculated, please read on.

Average Order Value (AOV)

AOV (sometimes also referred to as average ticket size or basket size) is the average value of an order during a period of time. AOV can be pre-discount, or post-discount, depending on how you look at it. Having said that, you can only gauge the real value of the business you are building if you consider the AOV post-discounting. Customers acquired through discounts are almost always going to be discount seekers, and not building that into your long-term unit economics means you are not being honest to yourself.

Our AOVs (post all possible ways to consider discounting) are Rs 480 for Type A orders, and Rs 375 for Type B orders.

Zomato doesn’t do much discounting by itself. Less than 2% of our orders are discounted by us, while 27% of our orders are discounted by the restaurants themselves. When restaurants discount something, we don’t call it discounting, we call it ‘pricing’. Just like airlines, or hotel prices, because restaurants push discounts on Zomato themselves to drive demand, using the Zomato for Business app (which they use frequently for review notifications, etc.).

Even this real-time pricing is built into the average order values for Zomato. Effectively, this formula is what holds true (and should hold true) if you are honest to yourself about your average order values –

(average order value) * (number of orders) = (total amount of real money transacted on the platform)

Take-rate/Commission/Gross Margin

This is the percentage of the total bill amount that the restaurant pays us for bringing them the delivery orders through our online ordering service. We have a variable commission rate for most of the restaurants on our network; the commission rate depends on the delivery experience for each order as rated by the customer – the better the experience, the lower the commission. While it might seem counterintuitive to do this as against charging a flat commission rate, it incentivises restaurants to provide better experiences, which leads to better customer retention.

For us, this is averaging out to 8.2% in the most recent month for Type A orders, and 18.2% for Type B orders (which includes an extra 10% for delivery).

Here’s the math for this.

Type A orders, commission – AOV * Take Rate = Rs 480 * 8.2% = ~Rs 40

Type B orders, commission – AOV * Take Rate = Rs 375 * (8.2% + 10%) = ~Rs 68

This means that we have Rs 40 to play with for Type A orders, and Rs 68 to play with for Type B orders. That’s our gross margin on each order, on an average.

For type A orders, we need to make sure that the total order processing cost – including payment gateway fee, order processing fee, support cost – is less than Rs 40 per order.

For type B orders, there’s an extra cost we incur per order which is the delivery cost, and we need to make sure that the sum of order processing cost and delivery cost is less than Rs 68.

Now, let’s talk about Delivery cost.

Delivery cost

Applicable to us only for Type B orders, this is the cost incurred to deliver a single order. This is a loaded cost, which means it includes salaries for delivery staff, the equipment and vehicles they use, as well as training and administrative costs. We pay Rs 50 to our delivery partners per order on an average (more on that later).

But it costs our delivery partners Rs 62 to fulfil an order they receive from us. An important point to note here – despite our business being extremely spiky, our delivery partners (Delhivery and Grab) are able to work in a relatively cost-efficient way because they can do e-commerce and grocery deliveries during our lean times (when food ordering is not at its peak). So Rs 62 is extremely good already. There’s room to make it better, and we are together working on predictive tech to make route planning better for delivery personnel so that this cost can come down to around Rs 50 per order.

Another important point to note – Rs 50 will eventually only be possible if you are able to keep delivery personnel busy during lean periods for food orders (which they do by serving e-commerce and grocery). Oh, and by the way, this math doesn’t factor in the very high recruitment cost of delivery personnel because of sky-high attrition.

We are thankful to our delivery partners to work with us to grow this part of the business. If we were to do this ourselves (and we’ve had the urge to do it multiple times), we would end up ruining the unit economics for the entire company in one shot. Because of the spiky nature of the business, the delivery cost would work out to ~Rs 105 per delivery. Of course, in some dense areas, during peak times, this can be much lower. But over a month, over all the markets you serve, it makes no sense at all, and you will lose more money than you can ever imagine recovering in any way possible.

And we haven’t yet gotten to the order processing and support costs yet, which is next.

Processing and Support Cost

Processing cost includes the telco data fee we incur when we automatically transmit a new order over to the restaurant. However, sometimes the device at the restaurant’s end has a bad signal, and we have to then make an SOS call to them to get the order processed. Our network is designed to have 95% automation, but it ends at 80% automation because of data/battery issues at the restaurant’s premises. Obviously, automation is way cheaper, easier, and quicker than making a call to relay an order.

Then there’s the support cost, which is what we incur when something goes wrong, and the customer raises an issue with our support team.

For Type A orders, we only have a two-way conversation to handle – between the customer and the restaurant. For Type B orders, we have a three-way conversation to work with – the customer, the restaurant, and the delivery partner’s person who is on the road.

Our total processing and support cost for Type A orders adds up to Rs 18.4 per order. For Type B orders, it adds up to Rs 27.4 (primarily because of the three-way calling, which adds more support agent time, and tariff to the cost stack). Three-ways aren’t always a good thing.

Six months ago, our support cost was about Rs 50 per order – we’ve managed to bring this down to current levels with automation, and by making hundreds of micro improvements. There is no silver bullet to bring this cost down – you have to keep chipping away at this cost bit by bit. And that’s what makes this business exciting and endless. Save a rupee per order, and the unit economics start looking way better than they did before.

Another thing this cost includes is the payment gateway fee, which is not borne by the restaurant separately (35% of all our orders are now paid for online). The number of restaurants accepting online payments on Zomato has gone up drastically in the past six months. In addition to a smoother product experience for customers, it also significantly drives down the cost associated with cash collections, and prevents pilferage – something that is very hard to control with a large delivery fleet, even if you own it.

A founder of a now-defunct local grocery startup once told me that their entire net revenue got wiped out due to cash pilferage at the hands of delivery personnel. Interestingly, he wasn’t taking that into account when calculating unit economics, and was adding these costs in the corporate overhead line item of “Miscellaneous”.

Net Contribution Margin

Net contribution margin = (AOV * Take Rate - Delivery Cost - Processing Cost)

This is what contributes towards the overall profitability of the business, and has to, over time, offset the fixed cost of the business.

Customer acquisition cost (CAC)

This is the marketing cost. This is beyond unit economics and fixed costs. How much CAC you incur, depends on how much money you are willing to spend to acquire each customer into your transaction funnel, and at what pace.

India’s CAC to LTV (life time value for a customer) ratios are very bad. Some companies are even acquiring a transacting user for as high as Rs 1200. Why’s that so high? We pay extremely high prices to various marketing channels in addition to discounting for customer acquisition.

Also, most e-commerce/transaction businesses have a traffic problem. They need to spend large amounts of capital to re-engage customers i.e. getting them to transact again on the platform. On that note, the Priceline group spent $2.8bn last year on online marketing, and Amazon spent $3.8bn. Unless customers visit the platform for something else (e.g. content, reviews, photos, and sharing) and then naturally move into the transaction flow, it is very hard to drive high re-engagement rates for commerce platforms. And that’s exactly what makes it easy for us at Zomato – our classifieds business, where we have 8.5 million monthly uniques in India alone. And so far, less than 3% of them are ordering from us. There’s massive room to grow before we think of paying big $$$s for marketing.

Fixed Cost

Sales teams, engineering teams, analytics and business teams. This cost stays fairly the same over time if you are adequately staffed here. Over time, the total net contribution margin should become more than the fixed cost, which will give you…

Profit.

This might have needed explaining in 2015. Not anymore.