Thursday, August 13, 2015

On Platforms and Ecosystems

This stuff is a decade old for me and I can barely drag myself to repeat it. But, I've read lots of daft stuff recently on platforms and ecosystems normally out of the mouths of half witted strategy consultants, so I will. 

The reason why you build a platform is to enable an ecosystem. A platform is simply those components (ideally expressed through APIs) that your ecosystem exploits.

The reason why you build an ecosystem is for componentisation effects and to exploit others through data mining on consumption. 

If you create a platform of commodity components (ideally as utility services) then you not only enable the ecosystem to quickly build (increase agility) but reduce costs of failures. By mining what they build (by looking at consumption of your components) you can use this to identify patterns useful for others. Hence, you can leverage your ecosystem to spot useful patterns which you can then commoditise to new components in the platform to help grow the ecosystem. This is a model known as Innovate - Leverage - Commoditise and it's so old, it's dull. You can call it network effects if you must.

Effective exploitation of an ecosystem depends upon you actively detecting those new patterns, the speed at which you can detect new patterns and the size of the ecosystem. 

IF effectively exploited then your apparent rate of innovation, customer focus and economies of scale all increase with the size of the ecosystem. 

A few basic pointers.

1) If you don't focus on user needs and reducing friction (i.e. making it easy to use) then you lose. No-one will turn up or those that do will quickly leave for something else or build their own out of desperation.

2) If you limit your platform to internal (i.e. company only) then your ecosystem will be smaller than a company which exposes their platform to the public. Their rate of apparent innovation, customer focus and efficiency will massively outstrip yours as their ecosystem becomes larger. You lose.

3) If you fail to data mine consumption then you won't be able to leverage the ecosystem to spot new patterns that are useful to the ecosystem. Your ecosystem and platform will stagnate compared to a competitor that does this. You lose.

4) If you do mine your ecosystem and aggressively harvest without giving the ecosystem reasons to stay ... everyone will run away. You lose.

5) If you build a platform based upon product concepts then the cost of data mining consumption becomes high and the speed low compared to a platform providing utility components through an API. If you're trying to build a platform of products against a competitor who is providing a utility then - you can guess. You lose.

6) If you build a platform with components that are not industrialised (i.e. commodity like) then the interfaces will continuously change and your ecosystem will not be able to rely on you. If you're up against someone who industrialises those components then ... you lose.

7) If you have little to no ecosystem and decide to take on a large ecosystem in the same space without co-opting then assuming they are public, provide industrialised components through an API  as a utility, focus on removing friction & user needs whilst data mining effectively then ... you lose. You never get a chance to catch up.

8) If you build new components on a platform and fail to implement a mechanism of evolving those components to industrialised services then you build up technical debt. Over time, you build new upon new and this becomes spaghetti junction. Your platform creaks and collapses. The fastest way I know to do this is to have one team building new stuff, one team taking care of the platform and no-one in between. This creates almost an internal war of them vs us exacerbating the problems of technical debt. Against anyone with a faintest clue of what they're doing ... you lose.

9) If I say phrases like ILC, two factor market, supplier ecosystem and you go "eh?" ... you'll probably lose. There are many forms of ecosystems with many different models and mechanisms of exploitation. Try to learn the different types.

10) If you think platforms are all about marketing ... you lose.

11) If you think platforms are all about engineering ... you lose.

12) If you think platforms are easy (ps. I built the first platform as a service in 2005 and ran a large single sign on and imaging platform back between 2001-2006 with many millions of users) then don't even bother. You'll lose.

13) If you think the secret is to build an API specification, call it a standard, even an open standard and vendors will all come and build against it creating your ecosystem in the sheer delight of your wonderful gesture ... oh dear, you're in so much trouble. Cheaper to open your wallet to others and say "help yourself".

There's more but I'd rather gnaw my leg off than talk about platforms and ecosystems again. This is enough to begin with.