This is part 1 of a two part Q&A with Sangeet Paul Choudary about platforms and open business models.
Sangeet Paul Choudary writes the blog Platform Thinking which has been featured on leading publications and research labs including WSJ, HBR, WIRED, FastCompany and the MIT Centre for Digital Business. He is an innovation analyst and consultant who has lectured at leading academic institutions like MIT Media Labs and INSEAD and consults leading startups as well as traditional firms making the transition to digital platforms. He is a mentor at leading accelerators like 500Startups and serves on the advisory board of leading platform startups.
There isn’t a day that goes by that one doesn’t read about another startup that is trying to create a marketplace. What are the key things they should not do to succeed in their platform approach?
The Platform approach is very different from the traditional approach which I call the Pipe approach because in the pipe approach your business works like a pipe-it’s creating stuff and delivering it to the end consumer. Think of a television channel, sourcing and delivering content to the consumer. It’s very linear.
In contrast, a platform allows users to create and exchange content and information and allow them to interact with each other. So, YouTube would be an example of a platform while the counterpart pipe is Television. Fundamentally, Platform approach it’s very different from the pipe approach, in the sense that to start with there is no value on the platform because users create value for each other. So, think of Twitter without any user, think of AirBnB without any hosts or without any travelers. There is no value in the platform of itself until the users start coming on board.
A traditional taxi company can serve the consumers directly but Uber cannot serve the consumers until the taxi drivers come on board. So, there are different users who use the platform, who provide value to different types of users. So, we typically think of these as producers who add value to the platform and the consumers who consume this value. And the platform facilitates the interaction between the two.
With that distinction in mind, the key things that a start –up should not do to succeed in their platform approach are:
1) Technology is not the end game, Interactions are the end game: It should not think of technology as the end game because the platform approach is not just about setting software. It’s about gaining adoption and building interactions on top of software. And so, the goal of the company is not to create great technology and see technology as the end offering. But it is to create technology that facilitates such kinds of interactions. It is to create rules that allow these different kinds of users to come on board and interact with each other.
2) Pricing models change in case of network effects: In general the way one goes about business in the traditional sense where you’re building a product and selling it is that you just find a good model of pricing- maybe it’s based on the value of the product or it’s based on the cost that it took to create the product. You find a reasonable method of pricing it and you sell it to everybody. On platforms, you can’t do that because your users create value. So, there are going to be some users who, if you start charging them for using your technology, are going to actually not be able to create any value and that’s going to diminish the value of the platform for the other users. So, the key thing over here is to understand who to charge and who not to charge. In many cases, you actually have to incentivize some of the roles on the platform to participate.
3) Who uses the platform defines the platform: Finally, more than the technology, what determines success on the platform is who’s using it, because the values is created by the users. So, curation becomes very important, determining who can use the platform and who cannot use the platform becomes an extremely important part of platform management.
This is very different from products-where the product is given to an end user and the end user uses the product. In such a case, you know, one person’s usage does not affect the experience of other people. Whereas in case of platforms, there are network effects-one person’s usage does affect the experience of other people. In general, more people using it helps but if the wrong people use it, it hurts everyone esle. For example, think of ChatRoulette, a video chatting platform which randomly connected with anybody on the internet to have a video chat. And as, more and more undesirable users (read: naked hairy men) started coming on the platform, the others started leaving the platform. Pretty soon, ChatRoulette became a place where you would end up being randomly paired with a naked guy in some other part of the world and that’s not what you wanted.
So, users actually stopped continuing to use the platform because of the kind of people that were there on the platform. More recently, we’ve been hearing complaints about how Quora is losing quality because the best people have left.
But in general, because of network effects, these three things are very important-
- There’s no value on the platform when you start. You need to understand that it’s not about just building the right technology. It’s about creating values through enabling the right interactions.
- Because network effects are important, because users create value, you cannot charge everyone and you might even have to incentivize some people to come on board.
- Because network effects create value because the value is not in the technology, who uses the platform is often more important than the technology itself. So, curating who gets access to the platform itself becomes very important.
If I’m an entrepreneur who has an idea for a product or service, and am not aware that I might be creating a platform, what should I be thinking about first?
If you do not know whether you are creating a platform, the first thing you should be doing is see where the value is being created. Do you own the value, are you providing value to the end user or are users providing value to each other and you’re managing that interaction? That is the single most important test of the platform.
The second test is-Is your platform going to be more useful for other users as users join in?
And the third test is-What can the first user do on the platform? Can the first user do anything at all? If the first user cannot do anything at all then you rely on network effects.
So, all of these three things are important in their own ways and you don’t need all three things to know that you’re building a platform but even if one of these is true-even if your first user can do something on the platform, but users have greater value when more users join in and their network effects, you’re still building a platform and you should keep the principles of building platforms in mind.
What are the main challenges in scaling a platform business?
There are three main challenges in scaling a platform business.
First of all, there’s a chicken and egg problem. How do you get people to use the platform when there is no value when nobody is using it? And getting the first user is extremely difficult, when Air BnB has lots of users it’s very interesting but when the first user comes on AirBnB, there’s nothing over there. Facebook with a lot of users is great but with the first user, it’s nothing. Twitter without users is just a place to write 140 characters, it has no value. And so, getting that chicken and egg problem solved. How do you get users when nobody else is there? How do you get producers when consumers are not there and consumers when producers are not there? It’s one of the most important challenges in scaling a platform business.
The second challenge is-How do you scale quality on the platform? Very often a platform may get initial traction but as it grows, in its hurry to scale, it might allow everyone to come on board and then it may become a bit messy. That’s what happened with ChatRoulette in the example I gave earlier. It happened with dating websites for a long time. And that’s why curation becomes very important. You need to scale a platform in such a way that the right people get access to the platform and you prevent the wrong access from happening-any access that could lead users to stop using the platform should be prevented. So, that’s a key challenge. My Space is a great example, a company that scaled very well but beyond a point they allowed the users to do too many things-they allowed users to start tampering with the user interface and the whole navigation on My Space became a nightmare.
And finally, platform monetization is very non- trivial because you cannot charge everyone. Very often, you can get traction with free users but when you start charging, that traction starts falling by the wayside. That’s an important thing to think about-How do you ensure that you can scale a platform business, not just platform usage but how can you get people to pay for something and yet continue to grow usage?
In a platform world, how can companies begin thinking through the scenarios of different monetization models?
There are four ways platforms make money.
First, producers and consumers transact with each other and the platform takes a cut. So, any market place-think of HC, think of AirBnB, there’s a transaction that happens and the platform takes the cut. There’s a seller, there’s a buyer, producer and consumer and the platform takes the cut. So, the transaction model is the first model.
Second model is that one side pays the platform to gain access to the other side and this is usually the lead generation model. The platform gets consumers in but it asks the producers to pay to get access to the consumers. So, Open Table is an example of that. The platform asks the restaurants to pay for the consumers who book seating on the restaurant. Dating websites are another example. You can sign up but if you want to access the other side and you’re a guy and you want to contact the girl who’s profile you’ve seen, you need to pay the platform by becoming a paid member.
Facebook promoted post is another great example because you can put a post generally on Facebook but if you want to promote it, you need to pay the platform. So, the platform is essentially saying-‘’ If you want to access more consumers, you gotta pay me.’’ So, that’s the second monetization model. How can you create a wall between producers and consumers and ask for payment to get the wall down?
The third model is what we often think of in terms of advertising. But let’s think of it in this way-you have a producer and consumer, they’re interacting with each other but there’s no money being exchanged. Say, we have a third person in to show the consumer something in exchange for which the platform makes money. So, an example of this is advertising of course in the case of Facebook but even in LinkenIn it’s recruiters-you have users talking to each other on LinkedIn and then the recruiter comes in and shows you a job. On Dribble, the recruiter comes in and shows you a job. It may not be traditional advertising but it’s not too different from that. It’s essentially, if something is not happening between the producer and consumer, if money is not being exchanged, you need somebody else to come in who can bring money into the picture.
Finally, the fourth way of monetizing platforms is to actually ask users to pay for technology and this is what happens very often in a Freemium model if you’re using Vimeo as a producer, you need a better video creator who’s a producer, you need better infrastructure, you need better hosting, you pay. If you’re using Flickr as a producer, you pay because you need better infrastructure. So, broadly these are the four monetization models for existing platforms.
Big Data is a huge topic right now, and when so much data exists about customers, it is easy for companies to fall into the trap of focusing just on numbers. Where do you draw the line between understanding customers, not just through data, but also through relationships?
So, this is actually a great question because there is an element of design and there is an element of analytics when it comes to understanding customers. Data can help you clarify hypotheses, it can’t build hypotheses.
The way I see data split into business is that you need qualitative information, you need to understand more from a design perspective how consumers interact with your business-what are the points at which their engagement is falling, what are the points at which they’re having friction and then the hypotheses on why that’s happening and then you stay to ascertain why that’s happening.
So, one model that I think of, where data fits into business is figuring out where in the design of the business are issues creeping up, problems cropping up. Design hypotheses for why this is happening and confirm that with data.
The other is where you take the data first approach and I take the example of Facebook over here. Facebook wanted to figure out what separates the highly engaged users from the less engaged users, the frequently returning users from the ones who don’t. And they realized that when they connected users to ten friends in seven days, those users were more engaged than those who were not connected in that fashion. And that insight came out of data because they had to do a cluster analysis and figure out the factors that were separating the frequent users from the less engaged users. This was done by Zynga as well-they figured that users who returned within a day of signing up, were more likely to keep returning and that’s how they set up the notification strategy, to get users back-once they are signed up, get them back within a day.
So, there is a method of using data starting from data and there is a method of using data to clarify a hypothesis that you get from a design-centric approach and I think there’s a place for both.
At this point, web monetization models are well known. What emerging models do you see being experimented with that may be more valuable to customers, partners and providers?
Web monetization models are well known. But what I see increasingly becoming more important is how you think of it in terms of a platform? For a long time we looked at web monetization from the perspective of-‘’I have a website. I am running something. How do I make money out of it from either the customer or a third party.’’
Now, the way I’d like to think of it is if you’re a platform, if you’re allowing two people interact, what are the ways in which you can make money from that interaction and that’s where the transaction economics become important. For example, if you have two people exchanging something-if there’s a buyer and a seller on the platform, who do you charge and how much do you charge because there’s a point beyond which pricing doesn’t just drive away the user, it kills the network effect and drives other users away too even if they aren’t charged. It’s not just the seller who leaves but the buyer as well.
And there’s also the question of who gets charged? For example, in an education marketplace, the teacher gets charged because you’re clearly helping him get business and he pays a cut for that. So, it’s the supplier who gets charged. Whereas, in something like Sitter City, which is a marketplace for babysitters and customers – parents who need babysitters – the parents have to pay for it. So, the demand side pays for it, not the supply side. So, there’s no silver bullet exactly on who pays. What’s important is to understand which side is less price sensitive and is more likely to take the brunt of a charge.
And the other important model that I see emerging is that when you have platforms you can do this transaction charge if you can control the transaction on the platform but not when you can’t control the transaction on the platform. Task Rabbit faced this problem because Task Rabbit being a local marketplace-once you have the first transaction being done, the parties can take the transaction offline and they can just avoid the cut.
So, how do you monetize such platforms?-it’s an interesting question. And that is where certain other things come into the picture. In those cases you build a wall, you let one side get certain access to certain things about the other side but if they want to get full access, they need to pay before they want to get full access so that before the users take the transaction off the platform, you already got them to pay. And that’s what happens with dating websites-you can look around at profiles and see-who’s hot and who’s not but if you want to contact the person, you need to pay the platform.
Another example is Open Table as a lead generation platform-you don’t own the actual transaction which is happening at the restaurant but you charge a lead generation fee towards that ultimate transaction.
As companies begin to adopt a platform mindset, how will they distinguish themselves?
As companies adopt a platform mindset, there are typically three different ways they distinguish themselves. They can distinguish themselves on the basis of the network, on the basis of technology or on the basis of data.
Let’s take an example over here-there’s You-Tube and there’s Vimeo. You-Tube competes with Vimeo because of a superior network. It has a bigger range of people actually adding videos to You-Tube and has a bigger group of consumers consuming those videos. Vimeo instead, focuses on the technology side of things and making the experience better for the producer. So, it focuses on creating high quality HD-video hosting facility. So, these two companies compete with each other, even though, technologically, they are very similar.
Another example is distinguishing on data. Think of CraigsList and AirBnB. If you want to rent out a house or if you want to rent out a room for a day or two, you can use Craig’s List or AirBnB. But the way they compete is that AirBnB has better data- it has better data about hosts and travelers, about who should be trusted, who should not be trusted. These are all missing on Craigslist and since trust is such an important factor in this interaction, AirBnB wins over Craigslist. CraigsList in turn wins over every other start up that competes with it partly because its network is so big.
So, I typically think of the network way of differentiating is to build a better network-who you have on the network helps you to differentiate yourself. There are dating websites that allow anybody to come in and there are dating websites that curate the kind of men who come in like Cupid curated and that curation becomes a way of differentiating yourself. So, the way you build the network differentiates you, the kind of technology you have differentiates you or the data that you capture and how it enables users to interact with each other, that differentiates you.
On the startup side, traditionally startups that go global raise huge amounts of capital to fund the sales and marketing engine. In a networked world, how does this change the game? Will we see more startups going global without the need of a lot of capital?
The answer to this varies from platform to platform. Even in a networked world, you still need a lot of capital but more than the capital, you need to be smart.
You need more capital when a network scales because when a network scales, it scales really fast and community management as well as the technology need to scale well and that’s what needs capital.
What’s different is that if you’re starting a network business, if you’re starting a platform, you just can’t start it by throwing sales and marketing money and throwing in more headcount. You need to be smart about how you get the first bit of community going. If you think of Facebook, they’d focused on Harvard and that’s how it was useful on the first day it was useful. If, on the other hand, they had opened it to the whole world, and got thousand users from around the world who did not know each other, there would not be anything for them to create interactions on. And so, that becomes a very important aspect.
So, while I won’t necessarily comment on whether the need for the capital will go or not, the places where the capital is spent will change. And Groupon is also a platform- it requires a different level of sales and marketing because the end users, the merchants are not directly using the platform. And so, there’s an interfacing part required where you need to throw in resources. But in general, when the end users use the platform directly, I don’t see sales and marketing being required in the same way as it was required in the past. Think of how advertising worked pre-Google and hows AdWords’ self-serve platform works.
One of the main issue with platforms is stickiness. How can a company think through keeping buyers and producers engaged?
Stickiness is a huge problem for platforms, especially when new behaviors are being created. When you create a new behavior, the problem is not just stickiness but it’s getting people started with the behavior. For example, never in our lives, while visiting a store, have we taken a gadget out and scanned a bar code. And now, there are Apps which want us to do that. So, how do you create this new behavior?
The platforms that really went in creating a new behavior do two or three things. They figure out why users would want to do something (motivations) and what would incentivize them. The second thing that is important is to understand that you cannot change behavior in one day and this goes back to how habits are created. When we try to train a dog, the dog doesn’t change its habit in a day and I don’t mean to use that analogy pejoratively. What I mean is that habit change requires time and usually, it requires a reward schedule. Remember Pavlov’s dog? You can change habits, you can change reactions based on changing the reward schedule. And this is where the incentive comes in. You make the user do small things by which he creates new behavior and through that he becomes more engaged on the platform.
Finally, one very important thing for stickiness is that the more the user uses the platform, the more useful the platform should become for him. And if you think of AirBnB, the more of host uses the platform, the more reputation he gains over there and the more unlikely is he to shift to another platform doing the same thing because he’ll have to again gain reputation over there which prevents him from showing up at the top of the search results. The more the user uses Twitter, the more followers he gains so if he goes somewhere else, he’ll have to start from zero. So, these are some of the things where the more the user uses the platform, more useful it becomes to him. If you can build something like that, you’ve got stickiness.