Sep 18, 20234 min read

The Play Order (be data driven and flip the 'cart before the horse' problem)

Updated: Aug 3

Working backwards from the customer energises the team to find the technical solution

Visualising the customer outcome (customer announcement) before even considering the viability of the technical solution is a great way to put the horse before the cart. This is the 7 step play order that we recommend to organisations looking to get the most out of their technology and data initiatives, and enjoy the multiplier effect from having a good foundation in place.

#1 The Customer Announcement

What is the customer announcement or outcome? With the product or feature you want to release, what are you promising to your future customer regardless of whether you have a single line of working code, and whether it’s your 1st or 1 billionth customer? Perhaps best epitomised by Amazon’s philosophy. The magic of working backwards is that visualising the end outcome excites and energises the people working on it and creates the sustained engagement needed see the solution through.

#2 Ethical Posture

Is the product and how it is delivered aligned with your ‘long term’ best interests? Because in the long term, your fortunes are aligned with the fortunes of employees who have chosen to help deliver on your vision, the customers you have chosen to serve, and many other stakeholders. The term ‘Long-Term Greedy’ is perhaps the most succinct way of putting it.

#3 Security Posture

Good ethics leads to sound security. Only the rightful owners have access to their property (including data). And in a digital business it is easy to focus only on digital assets. Don’t forget physical assets and physical security

#4 Accurate data capture

A bird in the hand is worth two in the bush. We capture data from the web, mobile, tablet, b2b data providers, public & private sources, IoT devices and more. If you want to enable a seamless user experience, you better make sure that the data capture is appropriately complete (are you capturing everything you need to deliver on the customer promise?) and you are honest up front (are you being transparent about what you are capturing and why?), even if the price you pay is some initial user friction.

#5 “Thence cometh quiet to the mind” - Roger Bacon

The empirical revolution started by Aristotle in Ancient Greece, and later championed by Ibn al-Haytham in ancient Iraq and Roger Bacon in medieval England, lives on today as the data driven culture powering modern innovation. Rarely practiced well, but the staple of some of our greatest technological organisations today. Champion experimental study over reliance on authority. Don’t just ‘trust’ the data. Question it. The world is evolving fast, and what was true yesterday may not hold today.

"When the facts change, I change my mind. What do you do sir?” - John Maynard Keynes.

#6 Speak the language of Science - Mathematics

Learn to speak the language of science - which is Mathematics, or more specifically, Statistics. If it’s not expressed through mathematics, it is not science. You don’t have to have a math PhD, but learn the fundamentals so that you can make the most of the remarkable power of the mathematical tools you now have freely at your disposal over the internet. Here are some fundamentals to consider before diving straight in to the world of data science:

What problem are you trying to solve? Asking whether your model is trying to explain or predict will help you determine everything from how much data you are going to collect, modelling technique, and measuring performance.
Get clear on the basic concepts of Sampling, Confidence Intervals & Variability
Understand linear regression - the most basic kind of model. This short piece on understanding the meaning of an average (statistics is after all a glorified averaging exercise) and testing the quality of a regression model helps to clarify fundamentals.
Visualising data - this is a cool example of the magic of visualisation as popularised by the late Hans Rosling.

"Often though, the magic of the dashboard on the surface hides the nightmare under the hood"

The ugly truth - Pretty dashboard, ugly pipeline. — The ugly truth

What is AI all about? Get an appreciation of how you can easily go from understanding simple linear regression to understanding neural networks, which is the building block of all modern AI systems under the hood.
How confident are you that your explanation or prediction (inference) is true? Know the common Hypothesis Tests.
Exploit real-world capabilities. Working from a sound foundation in linear regression, the non-linear world of neural networks and AI is not so daunting. While the nuts and bolts of neural networks and other machine learning paradigms that make up the AI suite of tools is rich and deep in content, your transition to using these sophisticated capabilities can be pretty smooth if you have these easy to attain capabilities in your toolkit:
1. Knowing how to test the accuracy of models
2. Awareness of the availability of pre-trained models, models that are easy to train, and wrappers that allow you to set up and train models using powerful programming tools like python
3. Full fledged API enabled services that you can pay for (e.g. Chat GPT for Generative AI)
4. Awareness of what not to read or believe. Don’t fall for shallow advice that fails to appreciate the importance of understanding the foundations, as pointed out by Keith McNulty.

#7 People First

Focus on people, empowered by technology. Don’t be led by process. Think education, training, open culture, empowerment and greater access to technology, rather than policies, processes and permission. Yes, these things are also needed, but process is your slave, not your master.