Succeeding Through Failing Faster

Christian Blumhoff

In this blog series, we’re discussing the 3rd Industrial Revolution (see last week’s blog for the introduction) and its impact on information practices. Today, we’ll be looking at what role data sciences play in “failing faster.”

There is something classic about playing chess with your children. In fact any board game will executive plays chessdo, but somehow there is something special  when it comes to chess. I have no idea why. I know for a fact my children prefer computer games, but I am an old romantic. What was memorable from this weekend’s game against my son was how completely and utterly I was wiped off the board. I am happy if I can see two or three moves ahead, but my son saw through everything I did.

So I asked him what he had changed from the last time we played. The answer was simple and something that gives me hope. He started looking at his Christmas present from four years ago, Mate in 3’s: 460 Chess Puzzles from Historic and Modern Games. He identified that my weakest part of the game was not the opening or end-game, but the middle bit. Apparently, instead of playing full games against computers, he trained himself to look for the best moves.

The basis of game theory is of course well over 70 years old (John von Newman 1944 Theory of Games and Human Behaviour), but these were macro economic models. More modern “fail faster” methods in business are purely designed to deliver a known business model faster. What has changed recently (and it continues to change) is the level of data and information available. What has become tricky is not optimization, but paradigm shifts.

How do you understand a game you have to invent? How do you understand the rules it is governed by? Most of us focus on trial and error or simple denial. It’s much simpler to ignore an issue until it is actively blocking us; however, most companies find that that is too late.

What does fail faster imply in business terms?

It’s becoming obvious that younger mindsets have a clear advantage when it comes to change. The young entrepreneurs don’t mind if they’re measured. Every game they play is bottom-line driven. Lap time, points earned, land captured. It’s worth asking ourselves what this implies in business terms.

Here are the top five paradigm shifts that underpin the need for a “fail faster outcome” world that is focused on innovation:

  • Measurable results beat myth (The truth about population video is well worth the 60 minutes of your time. Try asking “prove it” in the next hype conversation)
  • Communicate, communicate, communicate (Tell a business case story with visualizations)
  • Prioritize enjoying what you are doing more than money (Economist)

How should you structure your analytical approach to innovation?

The final point in particular is somewhat at odds with the baby boomer generation that is running most companies at the moment. Experience in this area has taught me that there is something to all of these points, but in a business context they are particularly effective when put into a process. Imagine you are reflecting on your telecommunication business and that monetization of data traffic is something that has not yet taken off in your business.

The following process should help you structure your analytical approach to innovation:

  1. Hypothesise. Start with multiple hypotheses. Much like statistics, take a bit of time to think things through and come up with a set of options to test and evaluate. This is a little bit more than an idea phase as causality and rigour should be applied to develop some target models.
  1. Collect data. Once you have a set ofroutes to success established, take the time and see if you have thedata to start modelling. This stage is critical as it allows you to reach out and find if proofs exist that you have something potentially successful.
    • Do you have potential customers?
    • How do you identify them?
    • Can you link actions to outcomes?
    • Do you know what you won’t measure/ can’t measure and how important this is?
  1. Analyze. Now you can start to model how you want to make your process effective. Always be prepared to go back a stage or even two. Analysis should be mature and heavy on maths. Be as smart as you can be. This should not add weeks of effort but simply require more mathematically-minded individuals. The outcome is a clear understanding of the new information feedback loop.
  1. Strategize. You have done the basic planning and you’re ready to think more realistically. Don’t just think positive, but learn from the past. Focus on a clear understanding of what success looks like, when to stop trying, and what to do when. A good strategy means you can concentrate on doing and learning rather than redefining what you want to do in the first place.
  1. Implement. Implementing is all about realizing your models. Try, try, and try again. Keep evaluating and keep measuring and improving the information flow. Once the model is producing measurable and beneficial results, go straight to stage 6.
  1. Optimize. The earlier you think about optimization (making things work better) the more benefits you generate. Optimization options should be part of your strategy and at this stage you should simply look at implementing these.
  1. Automate. The ultimate step of optimization is to automate as much of it as possible to keep costs low and results consistent. Automating learning and automating change would be the ultimate expression of this step.

Admittedly, this is all rather theoretical. So in the next blog post we’ll take a critical look at a practical implementation of this process in a telecom organization. We will also have a look at the analytical models underpinning this and how this process can be sped up. (See complete blog series schedule below).

For now, here is one of those chess problems: My son used to teach me that the best way to get better at chess is not to play the whole game a lot but to focus on your weakest part and find the fastest way to improve it. (Solution courtesy of

Still to come in the blog series:

  • What role data sciences play in “fail faster” (Part 1 of 2)
  • What role data sciences play in “fail faster” (Part 2 of 2)
  • Considerations for developing an Innovation Factory (In 3 parts)
  • Simplify our concepts and measures of success
  • Why and how an organization improves its memory


Christian Blumhoff

About Christian Blumhoff

Christian Blumhoff is the COE BI at SAP responsible for providing globally Analytics and BI Business Development and Sales support. His specialties include Solution Design, Pre-Sales, Project Management and Customer Management and Consultative Sales.