Dunzo, a Bengaluru-based on-demand concierge service, is pretty small. Tiny, really. Rivals UrbanClap and HouseJoy are backed by millions in venture funding, while Dunzo runs a lean operation with a $650,000 funding it received in March.

Yet, Dunzo is off the charts. It is growing at a fast clip with operations doubling every six weeks or so. It currently services around 25,000 transactions a month, with anywhere between 60 and 100 tasks running simultaneously, powered by phenomenal word-of-mouth goodwill and a following that can only be described as cult. Customers rave about it on social media and heartily recommend it to friends — that holy grail for customer loyalty measured by the now-fashionable net promoter score that brands such as Flipkart covet. (Read how gruelling it is for Flipkart here and here.)

There are tweets and Facebook posts saying things like, ‘These guys will get you anything from anywhere at anytime. Seriously . Just try it.’ and ‘The other night we ran out of cigarettes during a party and didn’t feel like stepping out. We texted Dunzo and one of their guys got us a full pack within 20 minutes. So awesome!’

All this without spending a penny on marketing and advertising.

What’s working for Dunzo — internet-age slang for ‘Done, over, finished! — which we at FactorDaily are going to stick our necks out and call a potential biggie? And what can other companies in the same space learn from it? That’s what we set out to find in this story.

Concierge Extremus

It is 11.30 pm in Indiranagar, an upmarket neighbourhood in east-central Bengaluru. Purushotthaman K, one of Dunzo’s runners, is out looking for a 24-hour medical store. A female customer wants a pack of sanitary napkins and he needs to deliver it within 20 minutes. The bike-riding Purushotthaman checks a few well-known medical stores in the locality to see if they are still open, and finally finds one. He checks out the available options, takes a few photos of different packs and sends them to the customer via the Dunzo app to see if the brand and pack-size are correct. Once the customer has given a go-ahead via a text message (yes, there are no phone calls exchanged; all the communication happens via text), Purushotthaman pays for the product and rides off to the delivery location, also in Indiranagar. It is nearly the end of his 10-hour-long work-day during which he’d have completed about a dozen tasks. He says he usually makes Rs 1200-Rs 1400 a day, with a base salary and incentives.

Dunzo started as a “tiny beta” in 2015, when Kabeer Biswas, former co-founder of Hoppr, a Gurgaon-based hyperlocal deals company acquired by Hike Messenger in 2014, moved to Bengaluru and started running tasks for a user group he created on WhatsApp. Soon, he was joined by Mukund Jha and Ankur Aggarwal, who were running a similar service called Wingman, as well as Dalvir Suri, an applications developer and security expert. All four are co-founders.

“We ran the first 5,000 tasks this way to understand the problem. It was a tiny beta,” says Biswas, sitting in Dunzo’s office in Indiranagar, a two-storey independent villa converted into a workspace — primarily divided between the 20-member operations team and the 14-member tech team. It’s a pretty chilled out office, with people sitting in small alcoves and balconies on bean bags. The buzz is infectious, and one member of the ops team has brought his dog to the office.

“The internet today is about fulfilment; information is already done, it’s out there. When I moved to the city, I figured the information layer was already there. You needed to connect the user to the information and help them fulfil their tasks,” says Biswas. Despite the fact that there were others operating in the same space, he and the other founders didn’t want to rely on suppositions and guesswork; they wanted to find out exactly what the pain-points were, what kind of problems users wanted solved; and the challenges of running this business. And they discovered that for most people, there are tasks that need to get done but are either so boring, time-taking, or require going out of one’s way, that they never get done.

And most people would outsource such tasks if they could, because there is no special joy or sense of fulfilment in doing them on one’s own. “One of our early users was a mom who wanted her kids’ school books covered with brown paper. She did not even think it was possible to outsource it, but messaged us anyway. We said sure, we’ll get it done. And we actually found someone in Shivaji Nagar who did this,” says Biswas, laughing. His joy at solving these kind of problems for users is obvious.

Another mom wanted a birthday party for her daughter, but started late on the planning. Though Dunzo does not usually take on complicated and vague tasks such as party planning, because breaking down the requirements into specific items is too complex, the mom convinced them to take it on by breaking down the planning herself into 10 different tasks, which she then ran separately and simultaneously.

At FactorDaily, we tested Dunzo several times while doing this story. Someone ordered a packet of cigarettes on a rainy evening at 10 pm, while someone else asked for a document to be sent by Speed Post. I needed a lamp, that had been lying at home for months, to be fixed, and a carton full of children’s books to be delivered to a government school (needless to say, there was never enough time to get these things done despite the best of intentions).

All the tasks were fulfilled efficiently and seamlessly, requiring very little intervention from us, the users, and the unnecessary phone calls that lead to a bad user experience. Biswas says that at Dunzo, they were always conscious about avoiding phone calls — right from the WhatsApp days — and solving this problem was built into the product as well as into training the runners.

“We look at chat as an offline medium,” says Biswas. “It’s asynchronous, and you can place your request at the pace that you want. We are very conscious of not making phone calls, and this has been built into the app that we provide our runners.” Not only is there a robust map layer and a chat interface between the runner and Dunzo’s operations team, which can provide the runner real-time inputs on location accuracy, runners can actually see the customer’s phone number only when he is within 250 metres of the client’s address, and can call for directions only if strictly needed. “Even that is provided because addresses are sometimes so vague in India, and streets are not numbered properly. We are trying to eliminate phone calls altogether,” he says.

Runners use the chat feature robustly, and take the help of images when there is scope for confusion. For instance, Dunzo gets many requests from users to pick up grocery — sometimes at gourmet stores such as Foodhall, where many of the products are unfamiliar to the runner. In that case, the runner is instructed to share images with the users via the Dunzo app to clarify any doubts.

Of course, the aim is to keep human intervention at a minimum, and to create processes that will reduce as much ambiguity from tasks as possible. The tech team at Dunzo, headed by co-founders Jha and Aggarwal, is completely focused on this and uses artificial intelligence to automate processes. While currently, the 20-member operations team handles around 70% of user requests and bots handle 30%, the aim is to get this to a 60-40 ratio by the end of next month, says Jha.

Under the hood at Dunzo

But what exactly happens when a query comes in from a user? Well, to begin with, the query is slotted into five big buckets: buying, shipping, repairs, home and miscellaneous. A lot of ‘buying’ examples include stuff that people want from specific store: get me bread from Thoms Bakery (an iconic bakery and supermarket in the Frazer Town area of the city), or pork sausages from Bamburies (A Richmond Road deli famous for its cold cuts and meats), or a bottle of vodka from the liquor store (incidentally, liquor buying requests are very common, especially as the week wears on).

Customers pay anywhere between Rs 30 and Rs 100 for each task, depending on the distance involved as well as complexity.

A lot of ‘shipping’ requests fall under the ‘get me this from there’ category, and sometimes they are slightly odd-ball: ‘I forgot my jacket at home and it’s cold in the office, please get it for me’; ‘My mother made pickles and I left it at her place, can you fetch it?’; ‘I left my keys at work and can’t get inside my home, I need someone to pick it up from my co-worker who found it’…

Repairs are pretty straightforward — most requests are for getting broken cellphones fixed; and ‘home’ requests include ‘find me a plumber’, ‘get me an electrician’, ‘want to get my microwave fixed’ etc.

The miscellaneous or ‘Others’ category sometimes features what Biswas calls the truly “#kuchhbhi” requests. These can be somewhat unusual. For instance, there is a niche category of users who are in the process of getting independent houses made while living elsewhere in the city, and create tasks every week to get someone to go to their plot and take pictures of the progress. Then there are those who want to know if a certain store is open, and ask a runner to check and take pictures for proof. And then there are parents who need help with their kids’ school projects. At least a couple have sent Dunzo runners to take pictures of Cubbon Park, the sprawling 200-acre park in the heart of Bengaluru.

The tasks are further divided into as many as 75 sub-categories under the five big columns, to thin-slice the task description as much as possible. This is a huge step towards automation, because not only does this make it easier for human operations people to work faster, it allows bots to take over more and more of the work eventually because the directions are clear, precise and based on keywords.

Although Repairs, Home and Buying account for routine, unglamorous tasks that don’t make for great stories, they are key to revenue generation at the company because they are predictable, repetitive and need collaboration with fixed service providers. A few months ago, Dunzo started tying up with merchants. They realised this was a good revenue model when they figured there were many merchants to whom they were driving a substantial amount of business.

“There are some places where our runners keep going again and again for repairs and other services. For instance, a very common task raised is to buy USB chargers or charging cables, and for runners in a certain area, it’s natural to find one good electronics store and do 90% of the business with them. One such merchant on Brigade Road earned 6 lakh worth of business driven by us in one month. We have now tied up with the store and have a revenue-sharing arrangement with the owners,” explains Biswas.

The revenue earned from customers who pay directly for each task through an in-built payment system within the app, goes directly to the runners. There are around 200 riders or runners contracted to the company, many of whom work part-time. Several are college students, who come online for 4-5 hours a day to earn some solid pocket-money.

What Dunzo does differently

“While other on-demand service providers get a bunch of merchants and vendors on to their platform, and then ask the customer to figure out how they want to use these services, our focus is different: you tell us what you want done and we will figure out how to do it,” says Biswas.

Like others, Dunzo founders started with rolling up their sleeves and doing deliveries. Aggarwal and Jha, who owned and ran Wingman, had connected with Biswas, who had then just exited his previous Gurgaon-based company, Hoppr, which had been bought over by Kavin Bharti Mittal’s Hike and relocated to Bengaluru with a somewhat casual aim of starting a text-based concierge service.

Both sides were looking for funding, says Jha, and decided to merge. There were more than 50 chat-based concierge services at that time — some examples include Instano, DudeGenie, Helpr, Magic, Goodservice, Tars, Haptik, Lookup, Timesaverz etc. Of these, only a handful still remain operational. (See table)

“Owning our own logistics layer was crucial to improving efficiency,” says Jha, pointing that several hyperlocal deliveries like Opinio, Roadrunnr and Grofers have contracted delivery personnel through third-party vendors.

That said, admits Jha, “The scope for hiccups is very large because there are so many moving parts. It’s a sector where it is easy to start but scaling is a challenge. With third-party logistics, you cannot guarantee quality and time. You have to control this layer if you want to provide exceptional user experience.”

The other thing that Dunzo has done differently is let its work speak for itself. The founders have spent nothing on marketing and advertising and the company’s customer acquisition costs have been negligible.

Next, for a broad-based business, Dunzo had to be ready to do everything. “For example, an app that just gets your shoes polished won’t work. In this sector, you will only be successful if you run a business that has a high frequency of users and repeats,” says Biswas. Around 40% of Dunzo customers are repeat customers; the top 10% place requests more than 40 times a month. “Volumes have to be really high for revenues and profit. You have to be broad enough and narrow enough,” adds Biswas.

Dunzo is agile, the business evolves very fast; for instance, when they saw that a large percentage of customers were basically scheduling the same tasks again and again, they added a ‘repeat’ button to the app so that the whole transaction became a single-click one. They are also adding a feature which will allow users to add someone else — a friend, roommate or family member — to coordinate on particular tasks.

They are also focused on one principle: Not doing anything they are not good — in fact, the very best — at doing.

The user interface is clean, robust, yet simple. It is also very intuitive, with features like a ‘To-do’ list where users can create multiple tasks; a ‘Track your tasks’ feature where you can see how far from completion your request is in real time; one-click pay via PayU Wallet; and a strong support mechanism whereby you can reach out to support staff even after the task has been completed.

There are several factors that pile up, making it difficult to crack the hyperlocal delivery (HLD) space, says Kabir Rustogi, principal data scientist at Delhivery. And if even one of these factors works against you, it can signal a death-warrant for your business. “Order arrival rate is unpredictable: Even if the overall order volume is high for a particular HLD service, the distribution of orders is likely to be very erratic. This makes it very hard to allocate resources as even the slightest of deviations from predicted volume can result in long queues of unfinished orders (if we underestimate), or several underutilized drivers (if we overestimate). This has an impact on service times and delivery costs, respectively. To get the balance right is critical for the business and is a big challenge for HLD companies.”

The thing is, even if a company is able to do a good job at Point 1 and all other random factors (such as traffic conditions, wait times at service locations) work in their favour, the cost for fulfilling an order within a reasonable turnaround time is quite high. “This cost can vary between Rs 65 -100 per order, but most hyperlocal businesses are not able to charge such amounts from the end customer, so the losses pile up every time an order is completed,” says Rustogi. Dunzo, as mentioned earlier, does charge customers for each order.

What Dunzo is doing to us (and why that is big news for the company)

Biswas talks about a particular user who set up a task that very day to get someone to pick up a phone charger from his home — when he was barely 500 meters from home. He himself has had runners pick up his laundry, courier a visa form to his mom, and pick up jeans given for alteration.

Do services like Dunzo encourage laziness and an entitled attitude? Biswas denies this vehemently. “Most internet businesses start off as toys, or super luxury products. For instance, Uber started as a service for the richest people to get a limo after a party. Many of these products start as baller products,” he says. According to him, the network effect exists even out of this tiny 1%, and he is fond of paraphrasing a much-shared post on LinkedIn that says ‘Forget the Poor, Build it for the 1%: The Apple Watch Strategy’: “You should build for the top 1%. Everyone wants to move to the top 1 to 10%.” He compares services like Dunzo to products like the iPhone, (“in theory, it should not exist at all”) and the trend of the 1% network effect to smartphone adoption. “Smartphones started as products for busy, rich executives who wanted to be able to answer their email all day long. And look where we are now.”

“We want to promote efficiency of time. We are saying ‘let us take care of the things that you don’t want to do but have to do, so that you can do something productive’. We are in the business of giving back time to people,” he adds.

And they have a clear roadmap of the future: “Get Bengaluru business profitable first. Expand to two more Indian cities and get them to profitability. Get to the top six Indian cities where we feel we have a market. And then, if we continue to save time for users, we believe users in the top 100 cities of the world, and we will get Dunzo in their hands, by all means.” The company could not share numbers on revenues and cash burn.

The challenges along the way? Firstly, it is highly localized — Dunzo is not even fully operational in all areas of Bengaluru. It’s only the Indiranagar, Koramangala and CBD areas that are really well-served. Scaling up could be a huge challenge, and the risk is, they get so overwhelmed by the increasing volume of tasks coming in from these areas — let’s admit it, Indians love nothing better than a reliable Man Friday substitute; Do It Yourself is simply not in our culture — that the expansion plans take a backseat.

However, according to Rustogi, this may not be a bad strategy. “Growth in other cities will have to be done cautiously. It is essential for hyperlocal delivery (HLD) companies to identify tight catchment areas where the frequency of orders is very high. I am not surprised that currently the company operates in only a few selected neighbourhoods in Bengaluru. This enables HLD operators to club orders together and ensure that a driver is not required to travel for long distances to fulfil an order, thereby bringing down the cost per order,” says Rustogi.

Well, it remains to be seen whether the company stays small and narrow, delivering Starbucks Coffee in Indiranagar, or goes forth to conquer the world. We will be watching.

Update: Edited for spellings of two names at 11.50 am on August 30, 2016. 

---