Multi-sided business models are a unique phenomenon – unlike standard businesses which offer a product / service to a particular type of consumer, multi-sided businesses don’t offer any product / service. Rather, they provide a platform that connects buyers andsellers.
Think of Uber – it connects cab drivers and passengers, who benefit each other. E-commerce marketplaces are also examples – they connect buyers with sellers.
Such businesses face a natural chicken and egg problem. For the platform to be useful, both sides have to be present. Sellers won’t come on to your platform without buyers, and buyers won’t come either, unless there’s enough choice (i.e., sellers).
For example, people buy video game consoles only if there are games they can play. But game designers make games for a console only if there are enough people who own it. The proverbial chicken and egg problem. How do one solve this impasse?
The above article discussed a few ways in which businesses can break this deadlock. Many readers wrote in after the article, asking if I could create a framework / checklist that they could use to brainstorm ways to scale their own multi-sided businesses.
Towards that end, I recently published this presentation on SlideShare. Check it out, download it, and let me know what you think!
A few weeks ago, my wife and I were in Galle, Sri Lanka for a much-awaited vacation. We chose a villa with great reviews on TripAdvisor. It seemed a decent place. A little far from the main town, but the hosts were quite friendly.
But we couldn’t get much sleep any of the nights we stayed there, because our room had bedbugs.
After we came back from the trip, we made sure to rate the place. We left not one, but two ratings (one each from my wife and me). Both of them were 5 stars.
Did we enjoy getting bitten by bedbugs?
I was surprised too. Not just at my own rating, but at other ratings on TripAdvisor too. This place was one of the most recommended ones in Galle!
So how did this happen? How did I – and all the other guests – rate inferior customer service so highly?
Do ratings work?
The prevailing wisdom is that ratings work. That’s why they are everywhere. When you open an app on your Android phone, it asks you to rate it on the Play Store. Complete a ride on Uber, and you have to rate your driver. Order something from Amazon, same story. Open your inbox after a long vacation, and what’s the first email you see? A message from either your airline or hotel, requesting you to rate your experience.
I’ve always found the act of rating quite empowering. The equation is simple – if you can rate a service provider in public, he has every incentive to ensure that you get great service. Right?
Well, after that incident in Galle, I realized that ratings may not result in better customer service. In some situations,they may be worsening it.
Wait, how does that make sense?
I’ll explain. But first, let’s agree on two key facts about ratings.
1.Ratings have an impact on service providers.That’s one reason they’re ubiquitous. Drivers on Uber do get blacklisted for low ratings. Top-rated hotels on TripAdvisor do get ten times as many bookings as lower-rated ones.
2.Customers know ratings have an impact.This makes them capricious (this Verge article calls them – us – entitled jerks). To see this, you only need to see a few app reviews. Sample these ratings on Circa (an app that used to provide summaries of important news):
Ratings are supposed to highlight how good an app is. But no, sometimes you get a 1-star for an innocuous review request.
This customer fickleness is not just an app store phenomenon. As the Verge article says,
We rate for the routes drivers take, for price fluctuations beyond their control, for slow traffic, for refusing to speed, for talking too much or too little, for failing to perform large tasks unrealistically quickly, for the food being cold when they delivered it, for telling us that, No, we can’t bring beer in the car and put our friend in the trunk — really, for any reason at all, including subconscious biases about race or gender.
Please the customer, and hope for the best
The fact that we wield a strange amount of power and know it, turns upstanding, proud cab drivers and B&B hosts into fawning, obsequious and servile slaves. You can’t jilt or offend a customer in any way. A single misstep, and you get a 1-star rating. Not a 4- or 3-star. Your last five customers may have given you 5 stars, but this single rating could put you out of business. In New York, Uber delists drivers from the platform if they go below a 4.5 star average!
So what is a service provider to do? Provide honest-to-God great service.And hope that nothing gets screwed up.
But there’s an easier way.
The honest approach is hard, time-intensive and expensive. And it’s subject to random whims of the entitled customer.If a customer expects Hilton service at McDonald’s rates, you’re bound to get 1 star.No matter what you do.
But there is an easier, quicker and more inexpensive way. One of the oldest psychological tricks in the book.
Dr. Cialdini, the author ofInfluence, calls this trick “Liking – The Friendly Thief”. Studies show that if you spend more time with a person, you end up liking her. And if you like a person, you tend to favor her in your dealings.
At a certain level, this is obvious. But that doesn’t make it any less powerful. Malcolm Gladwell cites a great example of this in Blink. Patients don’t file lawsuits when they suffer shoddy medical care, if the doctor is polite. They only file when they feel the doctor mistreated or ignored them.
“People just don’t sue doctors they like.”
So, to get a great rating, all you need to do is: (a) smile a lot and appear likeable; and (b) talk a lot, to create a human connection and familiarity.
Tried and tested. Once you get to know the service provider, you’d be a stone-hearted reviewer to leave anything less than 5 stars.
Nice host + bad customer service = 5 star rating
That’s what happened to us in Galle. Even though I was aware of this cognitive bias, I was powerless to counteract it.
The owner received us with great cheer. He chatted with us for hours. Always smiling and laughing (even when I didn’t crack a joke. And I’m not that funny anyway). I learned a lot about his life. I commiserated on his past troubles, and lauded him on his recent turn in fortunes.
My room still had bedbugs.
But my wife and I didn’t complain. Who can tell off such a nice guy? And when he requested us to leave two ratings on TripAdvisor, how could we refuse?
Talk more. Do less. Get 5 stars. Repeat.
This is just one small episode. But it sets in motion an insidious feedback loop, which could result in worsening customer service over time.
Customer give a 5-star rating despite bad customer service.
Service provider sees this as validation of his strategy. And becomes more chatty, more fawning.
Soon, if he’s smart (our guy was), he realizes there’s no return on actual customer service. It’s much easier to smile and bluster, than it is to clean the room. Over time, he’ll become more talkative, and truecustomer service will degrade.
Woe betide the unsuspecting traveler when that happens.
Thus, ratings may have an impact that’s thepolar opposite of your intention.
How do we break this loop?
Now, I’m sure you want great customer service. So, how can we break this loop?
Just being aware of what’s happening is not enough. You’ll only feel worse, as you continue to give 5-star ratings like a powerless lab rat.
The only way to break this cycle is to have a system of multiple ratings on different attributes, instead of one single unidimensional one.
Why would that work? For three reasons:
It would force objectivity.If you’re rating your stay at a B&B separately on Cleanliness, Quality of Food and Friendliness of Staff, you’re more likely to question the halo around your host’s head, and distill your cheery feeling into its components
It would give the service provider the right feedback on how to improve.
Ratings are here to stay. Let’s make sure they actually improve customer service. Rather than slowly turning us into smiling zombies.
What comes first, the chicken or the egg? An idle question on which children (and sometimes adults) can spend hours shooting the breeze. The question is, however, not so innocuous when it comes to businesses.
Some of the most exciting ventures today have a unique characteristic – they’re multi-sided businesses. What’s a multi-sided business? It’s one which connects two or more distinct user groups that provide each other with benefits. Think of Uber – it connects cab drivers and passengers, who benefit each other. E-commerce marketplaces are also examples – they connect buyers with sellers.
Such companies, once established, have a high barrier to entry. While that’s wonderful, it also means that they’re incredibly hard to build. Users on one side of the business model find the platform useful only if the other side also exists. For example, people buy video game consoles only if there are games they can play. And game designers make games for a console only if there are enough people who own it. The proverbial chicken and egg problem. How do one solve this impasse?
I face this problem too, in the product that I’m building – connecting advertisers with consumers (launch coming soon – watch this space!). How does one break the deadlock between the two sides? Unlike the philosophical question of which comes first, here the only right answer seems to be both!
There are four ways in which successful multi-sided platforms have overcome this stalemate.
Slow and steady: Build the two sides together in lockstep
Jumpstart: Get one side up quickly, and then build the other
Fake it till you make it: Build one side gradually with a makeshift offering, and then bring in the other
Bait & switch: Start with a single-sided value proposition to build one side, and then introduce the multi-sided offering
1. Slow and steady
In such a model, your offering needs to be valuable to your very first customers on both sides. One way to achieve this is to start small – very small – at a level where it is possible to get both sides onto the platform and provide the necessary cross-network effects. Focusing on a single city, area or even a neighborhood first can help you prove the model to both sides. Once that happens, you can expand gradually, building the two sides in lockstep one neighborhood at a time.
This is what Uber did in San Francisco – getting the model going in one city, and then applying it to other cities one by one. Going small could also mean focusing on one specific customer or product segment before expanding to others. Amazon, and more recently Flipkart, started with selling just books, building a user base and brand recall before expanding to other products.
Tinder, the dating / swiping app, built initial traction in a very creative manner. In its early months, the marketing lead toured several college campuses. At each campus, she first convinced the girls to download the app. After that, when she showed the app to the male fraternities, they quickly jumped on, seeing the number of girls they knew on the app.
Growing both sides of the business slowly in sync is great, but what if you want to speed up growth? Speed is often critical initially – given the high barrier to competing in this space, multiple people with the same idea would try and hustle into pole position. Maybe you don’t need to have both sides up and running from the get-go?
There are many ways one can quickly get one side of the value proposition up, and then build the other gradually.
a. Partner with someone who already has a large user base
One way to break this stalemate is to opportunistically partner with someone who already has a large user base on one side of the platform. This could be another product with mass acceptance in your target user base, or someone who has strong existing relationships that could be leveraged. Once one side is thus engineered into being, you can then build the other.
Google hacked its way to an initial user base using partnerships. It partnered with Netscape to become its default search engine in the late 90s, and also tied up with Yahoo! to power searches on that platform.
b. Make it easy, low-cost and low-risk for one side to come on board
At the same time, you need to make it as easy as possible for one side to say ‘yes’. Drivers are much more open to trying Uber when all they need to do is accept a phone from Uber and keep it on. It’s simple, and it’s low-risk – there’s plenty of upside if any ride requests come on the app, but there’s no downside at all!
Belly, a loyalty program for small businesses, did the same. It gave retailers a very low-risk, easy to install and low-friction loyalty solution. Once a critical mass of retailers had it, localized network effects began to take shape – customers and other retailers, noticing this in some shops, started demanding it of others.
c. Subsidize initial adoption for one side
A subsidy or ‘free’ offer always helps give the initial nudge. This is what video console companies like Xbox or Sony PlayStation do. The console is sold at a subsidized rate to users, and the company takes royalty on the flurry of games that follow. Uber subsidizes both users and cabs initially, to speed up adoption.
3. Fake it till you make it
Sometimes, it’s difficult to obtain one side of the model quickly enough. In this case, you have to build one side gradually with a makeshift offering, and then get the other.
a. Be the counterparty till the real counterparty appears
Most large e-commerce marketplaces started with an inventory led model, where they stocked products themselves. Once user base was built, they found it easier to make the shift to the more lucrative marketplace model, connecting product suppliers to buyers.
b. Use existing systems or services
Sometimes, you don’t need to build the entire solution for all sides of your platform – winging one side of the platform is an option, at least until you demonstrate user traction. I’ve heard the story, possibly apocryphal, of how the Flipkart founders would actually go buy books from stores to fulfill their initial orders. Look for a repetitive, non-scalable way to fulfil one end of the bargain initially, rather than investing in building service infrastructure, supplier base, etc. for a model that is yet unproven – not only is the latter risky, it also delays your product’s launch.
4. Bait & switch
A very nifty way to build a multi-sided platform is to first offer a single-sided service, that doesn’t need a counterparty. Once a user base is built, you can layer on the multi-sided service. Sounds complicated?
a. Build a user base on one side with a focused (different) offering, then introduce the second side
Square is a payments solution for small businesses in the US. It’s really cool – a small chip-sized device that plugs into a mobile’s headphone slot and allows you to start accepting credit card payments. At least this is what it was initially. Once it built a sufficient scale of retailers, it added a second business model. Today, Square also runs a discounting app offering consumers great deals at its partner retailers, card-less transactions, etc.
LinkedIn also did something similar. It started as a pure play network for professionals. Today, a huge user base allows it to be much more – it now offers unique solutions to recruiters, job seekers, and professionals.
b. Start as an information portal
Another way to do this is to start as an information portal for one side of the platform, offering users a directory of information about the other side. Zomato started as a pure-play information source on restaurants in your neighborhood. Gradually, as user base grew, they started overlaying restaurant promotions. I bet they’ll add other services soon – allowing you to book a table, pay after your meal through the app, etc.
These are the four different approaches that companies have used to resolve their chicken and egg deadlocks. What do you think? Have you faced chicken and egg situations of your own? Would love to hear from you – mail me at firstname.lastname@example.org, tweet at @jithamithra, or comment here on this blog. And do subscribe here – I post roughly once a week, on startups, business models, consumer behavior, etc.
PS. In the actual chicken and egg problem, the egg comes first (there’s absolutely no doubt about that).