Skip to Content

Building Scalable Platforms – Part 1

Gopikrishna Aravindan
August 31, 2022

According to a 2015 Forbes report[1], ride-sharing app Uber’s valuation was greater than the 100-year-old General Motors. It took Uber less than 6 years to reach this number without owning a single car!

When travel and hospitality industries were worst hit at the peak of the pandemic, Airbnb was still worth more than 3 largest hotel chains put together. It took Airbnb little over 10 years to achieve this growth without owning a single room!

The Economist dubbed China’s Alibaba as the world’s greatest bazaar[2]. In little over two decades of its existence, Alibaba’s Net Income of $75B in FY2022 surpassed Walmart’s $17B by almost five times. (Source: WSJ)

So, what’s common between Uber, Airbnb and Alibaba that led to their phenomenal growth in a relatively short span?

Platforms have disrupted our lives in ways we could not have imagined few years back. We use iPhone and Android apps enabled by platforms, buy things from merchants on Amazon, read content off Wikipedia published by authors, view posts/ images on Instagram from content creators and … even this article you are reading, enabled by the SogetiLabs platform.

In this blog series, we will look at various elements of successful platforms and what goes behind the scenes to make them successful.

Design Principles:

Years after its release, Steve Ballmer quoted Windows Vista as his biggest mistake[3]. Although, when Microsoft introduced Vista in 2007 it was termed as the biggest launch in the product of Microsoft’s history.
Vista had a bunch of cool features and forward-looking Graphical User Interface, but it was soon becoming unpopular because of its tendency to hog resources. The problem lay in its design which had components needed to maintain reverse compatibility along with components for next generation systems – all part of core OS. This made the system complex & slow while at the same time developers were finding it difficult to write code. Also, PC users used to blazingly fast load times of WindowsXP could not help noticing how much Windows Vista was slowing them down. The only way to get the better performance was to upgrade the hardware which meant throwing more money. Soon users were reverting to Windows XP. Years after its release, Vista was still trailing behind XP in terms of market share.

Compare this to Mac OS X architecture[4], which consists of several layers with a core Unix system and additional subsystems. The application subsystem had a classic environment which enabled OSX to run applications run for previous versions without any modifications. User could run previous generation apps in the classic sub-system without slowing down the OS. While the Cocoa environment allowed developers to take advantage of latest application development environment and was built ground up to make most of the Mac OS X capabilities.

The Key take away from Mac OSX design story is that application specific features should built at the periphery layer, so that core functions do not degrade. Platforms should consider building features that cuts across multiple apps at its core. This would also allow platforms to scale rapidly as users demand more features without impacting performance.

Ease of Onboarding:

Products that were built on a pipeline model – a model which comprised of product developer at one end and consumer at the other – needed gatekeepers and checks to ensure that the product created was valuable to the customers. This introduced bureaucracy in process.

Book publishing is a good example – Until Kindle, authors were heavily dependent on publishers to sell their books. It’s easy to imagine how this model would easily favor authors well established in the industry. Kindle changed that completely by allowing anyone to publish their book and make it instantly available to thousands on readers. All an authors needed do was upload the book and set the prices, which could be changed any time after the book is published. The easy onboarding process makes it effortless to sell books on Amazon for anyone. Subsequently, the Kindle Book Publishing platform saw its model scaled rapidly – till date there are over 6 million titles[5] available to select from US Kindle store.

Making it easy for content creators to onboard is one of the key factors in success of a platform. While some review process may be required, any emerging platform will benefit from minimizing the checks and allowing content creators to principally self-serve.

What design principles do you think are key to successful platforms?


[1] https://www.forbes.com/sites/liyanchen/2015/12/04/at-68-billion-valuation-uber-will-be-bigger-than-gm-ford-and-honda/?sh=2ca5adc432e3

[2] https://www.economist.com/briefing/2013/03/23/the-worlds-greatest-bazaar

[3] https://www.businessinsider.com/steve-ballmers-biggest-regret-2013-8

[4]http://etutorials.org/Mac+OS/using+mac+os+x+v10.3+panther/Part+I+Mac+OS+X+Exploring+the+Core/Chapter+1.+Mac+OS+X+Foundations/Mac+OS+X+Architecture+and+Terminology/

[5] https://www.amazon.com/Kindle-eBooks/b?node=154606011

About the author

Gopikrishna Aravindan is an experienced professional services leader who has a passion for everything technology starting from ideating concepts to delivering large-scale technology transformation solutions while taking ownership of everything in between. He has a Masters degree in Information Systems from Carnegie Mellon University. He started his consulting career at Deloitte and has served seve

    Comments

    Leave a Reply

    Your email address will not be published. Required fields are marked *