Almost every software entrepreneur nowadays builds some kind of a ?platform?. Yet very few will confidently answer a seemingly simple, but very important question: What kind of platform do you build?
All platforms are not created equal. Google Search, Facebook, Amazon Web Services, Amazon Marketplace, Android, Uber, AirBnB, Waze, WeWork, Twilio and even Bitcoin are all platforms. At the same time, these platforms are very different in how they create network effects, interactions they enable, approaches to solving ?chicken and egg? problems (do you build the demand side first or the supply side?), openness levels, growth dynamics, subsidies, competitive strategies and monetisation methods.
Building a successful platform is more about making the right trade-offs than it is about best technology. To understand these tradeoffs you must have a good grasp on what kind of platform you are building. We decided to help platform entrepreneurs understand these tradeoffs, find their role models and learn from numerous examples of other platforms.
We took a data-driven approach and analyzed 170+ platform businesses created by Internet giants, traditional companies and startups. Several iterations on the data produced nine distinct platform types that we introduce in this post:
- Technology Platforms
- Computing Platforms
- Utility Platforms
- Interaction Networks
- Marketplaces
- On-demand Service Platforms
- Content Crowdsourcing Platforms
- Data Harvesting Platforms
- Content Distribution Platforms
Platform Hunt is an open initiative aimed to help entrepreneurs build successful platform businesses. The list of 170+ (and growing) platform examples is open and is stored in a public Trello board at data.platform-hunt.com.
We hope to make Platform Hunt even more useful to entrepreneurs by tapping into the wisdom of the crowds. Anyone can comment in the Trello board. Anyone can submit a platform example by filling a simple form at add.platform-hunt.com. Each new platform example will help entrepreneurs to learn from the collective experience of other platform businesses.
Here?s how you can help: Spread the word about Platform Hunt on Twitter, Facebook and LinkedIn to reach more people interested in platform business models and startups!
Technology Platforms
Amazon Web Services, Microsoft Azure, and Twilio are examples of Technology Platforms. Technology Platforms provide building blocks or services that are reused in a large number of products. Through permission-less innovation 3rd party developers embed these building blocks and services in their products, driving more adoption of the platform.
The reason we included Technology Platforms in Platform Hunt is that platforms like AWS are routinely lumped together with multi-sided platforms like Uber or AirBnB.
Note that Technology Platforms are not two-sided markets. They are not designed to connect platform participants (for example, producers and consumers, or people in a social network). Instead, Technology Platforms monetize by selling their services to developers and are typically invisible to the end users. For example, while Netflix runs its video streaming services on top of Amazon Web Services platform (AWS), end-users interact solely with Netflix. AWS is the plumbing that enables the service.
There are no inherent networks effects in Technology Platforms. They grow linearly with adoption by developers and do not rely on the interaction between a demand side and a supply side. As a result, Technology Platforms are much easier to launch because there is no need to solve the chicken and egg problem seen in multi-sided or peer-to-peer platforms.
Computing Platforms
Computing Platforms, in stark contrast with Technology Platforms, enable interactions between platform users and 3rd party developers. In Technology Platforms the developer ?owns? the user. In Computing Platforms, the platform ?owns? the user. Computing Platforms, like Apple iOS, Google Android or Microsoft Windows, allow developers extend the platform with new use cases, making the platform more valuable to users.
In modern Computing Platforms the connection between users and developers is through an app store/marketplace, which streamlines discovery, recommendations, activation and monetisation of apps/bots/extensions.
Computing Platforms develop strong bi-directional network effects once the platform reaches critical mass. Users attract developers, developers make apps, apps attract users, users attract developers and so on. Successfully launching a Computing Platform requires a solution to the difficult and all too well-known chicken and egg problem. How to attract users when there are no developers? How to attract developers when there are no users?
To solve this problem, computing platforms first build critical mass of one side of the platform and then open it up to the second type of participants. For example, Apple iOS was launched without any 3rd party apps or developer ecosystem. iPhones only had Apple?s own apps and services (Safari, Mail, iTunes) or those developed through direct partnerships. Once their user base (demand side) reached critical mass, Apple opened the platform to 3rd party developers (supply side) and the rest is history. The network effect between iPhone users and iOS developers then fueled explosive growth of the platform.
Google, on the other hand, took a different approach when launching their Android platform. Google first focused on attracting an initial population of developers (supply side) by rallying them with open source slogans and running competitions with substantial prizes. Developers started to make apps even before physical devices were available, using an Android emulator. When HTC G1, the first Android device, was launched, it was already seeded with an initial app portfolio.
Utility Platforms
Google Search, Kayak and Zenefits are examples of Utility Platforms. Utility Platforms attract users by providing a useful, typically free service. Once there is critical mass of users using the service, the platform opens to the second type of participants, advertisers in the case of Google Search, airlines in the case of Kayak or insurance companies in the case of Zenefits.
There is no network effect in the useful service itself. Users attract businesses, but businesses on the platform do not necessarily attract users. We go to Google Search looking for information, not to see ads.
Launching Utility Platforms is fairly straightforward ? Make sure you have a useful service that generates repeated use and has negligible marginal cost. Once you have an asset created by a critical mass of users (e.g. data, targeted engagement, etc.), open your service to businesses to monetise the platform (for example, through advertising, commissions or anonymized data).
Interaction Networks
Facebook, WeChat and Bitcoin are examples of Interaction Networks. The common element is that this type of platform facilitates interactions between specific participants (people and/or businesses). The digital interactions can take form of a message, voice call, image, or money transfer.
Identity is the foundational characteristic of Interaction Networks. All interactions on the platform are anchored on specific accounts.
Users join the platform to interact with other users, and therefore the primary network effect is between the users of the platform. Users attract users, who attract more users. In that sense, the platform is a one-sided platform connecting participants of the same type.
Launch strategy for Interaction Networks is quite well understood. The platform owner would typically target groups of people who already interact with each other, create a critical mass of interactions on the platform and build up the network effect adding more and more users to the platform. For example, Facebook initially launched a network exclusive to Harvard University students before moving on to other colleges and then finally, opening it up to all users.
Marketplaces
Marketplaces like eBay, Amazon Marketplace, AirBnB, Kickstarter or UpWork are two-sided platforms connecting supply with demand. Marketplaces enable transactions between demand-side participants (buyers) and supply-side participants (sellers). Prices of goods and services offered on the platform are set by the supply-side participants. Not less important, there is high sensitivity for variety of services/products ? generally, the more variety offered on the platform, the better.
The network effect in Marketplaces is between buyers and sellers. Sellers attract buyers, who attract more sellers, and so on.
Identity plays a secondary role in the platform. Buyers look for a specific product or service, not a specific seller. The product/service can be offered by multiple sellers who compete on price, reputation and experience.
Launching a marketplace and solving the chicken and egg problem is a difficult balancing act. Typically, a nascent platform begins with platform owners bringing small number of sellers catering to a niche audience. The platform then grows from there with most efforts devoted to bring buyers to the platform. Amazon Marketplace used a different launch strategy. Amazon already had substantial number of buyers on its online retail service, when the company allowed 3rd party sellers to sell to Amazon?s buyers.
On-demand Service Platforms
Uber, Munchery and Heal are examples of On-demand Service Platforms. This type of platform is designed to deliver end-to-end services fulfilled by a network of independent service providers/contractors. Its tradeoffs are very different from those of Marketplaces.
On-demand Service Platforms integrate discovery, order, payment, fulfilment, certification and confirmation of the service under one roof. Price, quality standards and the fulfillment processes are all set by the platform. The user/buyer typically has very little freedom, if at all, in selecting how the service will be delivered and by whom.
Availability and predictability of the service are essential quality metrics of On-demand Service Platforms. Contrary to Marketplaces, high variety of services is actually damaging for On-demand Service Platforms. The higher the variety, the less control the platform owner has over how the service is delivered, leading to a poor user experience and lower user retention.
The network effect of On-demand Service Platforms manifests itself in service availability. That is users are not directly attracted by the number or variety of service providers. Instead, the more service providers there are on the platform, the better the service availability, and as a result more users will be attracted to the platform.
Uber measures availability in minutes. Other On-demand Service Platforms can measure availability in hours, days or even weeks. For example, marketing on-demand service Doz delivers the service in weeks. The availability needs to be predictable and aligned with customer expectations for the particular type of service.
Launching an On-demand Service Platform typically involves signing up just enough service providers to ensure service availability to the first users of the platform. As the number of users grows, the platform owner must also grow the number of service providers on the platform to guarantee service availability.
Given the paramount importance of service availability in On-demand Platforms, it is clear that Uber?s Surge pricing was not designed to increase revenues, but to maintain service availability by balancing supply and demand.
Content Crowdsourcing Platforms
YouTube, Yelp and TripAdvisor are examples of Content Crowdsourcing Platforms. This platform type is about collecting content from a subset of users (video, blog posts, reviews, ratings, etc.) and sharing this content with a wide user base of the platform.
As opposed to Interaction Networks, where interaction is anchored on specific accounts, in Content Crowdsourcing Platforms users interact with the platform and the interaction is anchored on the content.
The network effect is between content contributors and content consumers of the platform. The more content there is on the platform, the more content consumers will join the platform making it more valuable for contributors, who in turn generate more content.
Launching Content Crowdsourcing Platforms is fairly straightforward. The platform owner will typically seed the platform with the initial content, then work to acquire users and motivate some of them to contribute more content.
Data Harvesting Platforms
Waze, OpenSignal and InsideSales.com are examples of Data Harvesting Platforms. Such platforms offer a useful service to the users and generate data through usage of the platform service. In fact, the agreement to contribute data is a requirement to join the platform. The data collected from all users of the platform is fed back to the service making it more useful for users.
The network effect in these kind of platforms is based on data rather than users. Usage generates data, which in turn makes the platform more valuable for users, which attract more users, whose usage generates more data, and so on.
Launching Data Harvesting Platforms typically requires building an initial user base without having the data network effect. For example, Waze started as Freemap app for Nokia phones at the time when navigation services were expensive. Freemap built its initial user base by offering free navigation using crowdsourced maps. Once there were enough users on the platform, the service value proposition shifted from free maps to traffic prediction based on the data collected from all users of the service.
Content Distribution Platforms
Google AdSense, Outbrain, Smaato and Millennial Media are examples of Content Distribution Platforms. Such platforms connect owners of user touch-points (web sites, mobile apps, devices) with content owners wishing to deliver the content (or ads) to the users.
The network effect in Content Distribution Platforms is between owners of the user touch-points and the content owners. The more touch-points the platform aggregates, the more attractive it becomes to the content owners. The more content is available on the platform, the more attractive the platform is for the owners of the touch-points.User reach and accuracy of content matching are the main quality metrics.
In summary
If you?ve successfully read through this long post, you now should be able to confidently navigate the complex landscape of platform business models. Don?t stop here. To get inspiration for your business, understand the trade-offs and find your role models, dig into the platform examples on Platform Hunt. Of course, don?t hesitate to comment and add new examples.
Happy Platform Hunting!
Michael and Sameer
- Updated in December 2016 adding Content Distribution Platforms
