On building a platform business
“We are a data analytics platform startup.” Depending on the person I’m talking to, this statement could mean something or nothing at all.
Often, people have a vague idea of what data analytics and platform are on their own. However, most have different understanding of ‘data analytics platform’: What does it do? How does it work? Who would use it? Why would people use it?
I’ve considered not writing this article and just letting the market see for itself when we launch our Human Managed platform. To be honest, writing about data analytics platform seemed too complex, too daunting of a prospect, especially while we are still building it.
This is part 1 of a 2-part blog, where I explore the most influential principles about ‘platform’. In part 2, I will do the same for ‘data analytics’. This is by no means a comprehensive account of the two concepts, merely an attempt to share the lessons that are shaping our journey to become the data analytics platform we envision.
What is a platform?
A platform is a business model that enables value exchange between two or more entities, typically producers (supply) and consumers (demand). Platform can be both physical (eg. shopping centre) and digital (eg. Shopify), but today’s largest platforms are mostly digital, growing from network effect and economies of scale.
First, crystallise your core value unit.
A critical feature of a platform is it has a core value unit that can be exchanged. For example, Shopify is an e-commerce platform for creating (production) and buying from (consumption) of online shops (core value unit). YouTube is a platform for the hosting (production) and viewing (consumption) of videos.
It was not easy defining our core value unit. Starting out, we knew the problems that we wanted to solve:
- Decision Making problem: Despite abundance of data and technological leaps in fields like data analytics and AI, decision making in business environments is not evidently better and is often characterised by human bias, knowledge gaps and errors.
- Data problem: have you heard of the saying, ‘garbage in, garbage out?’ This is a very painful and real problem that is caused by the challenge of data cleaning and data transformation. The lack of resources and integrated data analytics solutions leads to companies adopting standardised data analysis and use cases (eg. Customer acquisition, fraud prediction), merely scratching the surface of the real potential of their data.
We had to dig deep to ask ourselves what is our core value unit that solves these problems.
In terms of our value transaction, customers will subscribe to Cyber, Risk and Digital use cases designed to solve specific problems. However, use cases are the format of our service, not the core value unit itself. What our customers will be looking for (and what our data models and algorithms will be designed for) is layers of analysis across multiple data sources and data sets. We call this: depths of analysis. The more complex our customers’ problem, the deeper our analysis.
Human Managed is a data analytics platform for conducting (production) and applying (consumption) depths of analysis (core value unit).
Once the core value unit was clear, we mapped all our solution design to it, including our use cases, Cyber, Risk, Digital data analytics framework and all the way down to specific data model mapping.
Then, design for an entire ecosystem, not just your customers.
So why do we want to deliver depths of analysis through a platform, and not as a solution through managed services or consulting?
One reason for this is because our solution is an end-to-end data analytics solution, with the capability to collect batch and stream data from any data source, process data through automated frameworks and visualise information and recommendations through dynamic dashboards. Platform is the best medium through which customers can interact with our dynamic analysis and services, real time.
Another reason is much larger than our service scope; it stems from our mission to find answers through collective intelligence. Today, we are delivering depths of analysis for business users, but beyond this, we are building a platform that can host frameworks, use cases and learnings for all the players in our community: customers, partners, suppliers, employees, future hires… really, just about anyone who wants to take part in the discourse we are creating about pushing boundaries and making better decisions from big data analytics.
Here’s another description from the Platform Design Toolkit that I really like: a platform’s goal is to mobilise an ecosystem that creates value in interaction. The defining characteristics of an ecosystem is open, complex, diverse and prone to change. A platform must be able to deliver its core value unit to its stakeholders, despite changing processes and evolving businesses.
Finally, have a framework for platform architecture thinking and stick to it.
Now this is the toughest part — and it’s all about mindset and deliberate execution.
To an end user, a platform has to be like a light switch: It simply has to work. At Human Managed, our platform is everything that happens in between the second data enters our platform and the second it gets displayed as useful information on our portal. Behind the effortless user interaction lies many layers that must work together in synchronicity. In the digital world, most of these layers are abstract: Cloud infrastructure, Hypervisors, IaaS, OS, Image, Code, Container, Microservices… and we haven’t even reached the Application layer yet. Indeed, each layer is a ‘micro-platform’ in itself and they must be integrated to talk to each other through APIs.
What’s more, a platform that has multiple services and products like ours has to be business agnostic and use-case driven; which means, different users can consume different services through our platform to serve their particular use case. This is enabled by microservices: applications designed to function independently or combined together.
When so many things are getting built at the same time and especially during times of high pressure, it is easy to take the quicker or familiar route. However, we take conscious decisions to come back to our framework for platform architecture thinking:
- If you cannot reuse your work, don’t build it
- Make decisions according to the Human Managed Platform Reference Architecture
- Scale according to the volume of data that will be consumed to realise use cases
- Design to be resilient in a multi-site configuration
- Design of data flow to be influenced by use cases, not the other way around
Conclusion
What I keep learning from building a platform is that it’s much, much easier said than done to be a platform business. Because a platform is not a single solution for a single use case, there are so many moving parts all the time. And because a platform is never a finished product, this way of work will certainly not stop.
Building and growing something so fluid and dynamic requires clear, unwavering principles, especially when the going gets tough.
As with many things, being a platform company starts with identifying the core. Then it needs to be designed for an ecosystem to deliver value beyond the immediate users, because a platform not only solves the problem of today for few, but also the unknown problems of tomorrow for many. Finally, the platform architecture thinking has to be the driving mindset going forward, by everyone in the company. It takes the entire team to keep each other accountable whenever we diverge from the platform thinking.
We have made some painful mistakes and have to keep unlearning. But progressing towards a platform and a vision so much bigger than ourselves is what keeps us going and we wouldn’t have it any other way.
Part 2 on data analytics, to be continued…
If you found this content interesting, check out the Human Managed’s blog Human Thoughts, where we cover a wide range of topics that we are passionate about.
You can also follow me on Twitter and follow Human Managed on LinkedIn and Twitter.
Originally published at https://www.linkedin.com.
