Bridging the Gap between Business Problems and AI Capability

I completed a consulting project for a large business that had invested a huge amount of money building a data-science capability – and then didn’t know what to do with it. They had made the investment partly because they thought they ought to. After all, ‘big-data’ was constantly in the business press and popular media and their competitors were making similar investments and crowing about their successes in that domain (note to self – don’t believe everything you read on the Internet!).

They had been sold the idea of building the capability by a charismatic technologist who laid out a plan with a frankly astonishing promise for Return on Investment. I saw that plan for the first time some 18 months later (after it was all turning sour and, of course, with the clarity of hindsight) but the numbers were simply fantasy. Most importantly there was no real strategic argument; no ‘line-of-site’ between the investment and that promised ROI.

I got involved at the periphery of this mess whilst working on another project. By this time the data-lake had been built; statistical workbench tools had been commissioned and there had been huge investment in recruiting talented and skilled data-scientists. But this is where the trouble began. There appeared to be a complete mismatch between the, admittedly impressive, data capabilities of the organisation and the operational needs of the business. The following is somewhat of a caricature, but captures I think the spirit of what was going on:

Data Scientist: What do you want to know?

Operations: What can you tell us?

Data Scientist: We can give you ‘insights’ based on data – predictive and explanatory models expressed as beautiful and impressive charts!

Operations: OK. The thing is, I am 25 staff-members short this week and I am fighting fires on 7 different fronts including regulatory and audit issues. Can you help with that?

Data Scientist: Can you tell us what questions you need to answer from the data – we can do that!

Operations: Sorry – I am too busy for this. Call me when you come up with something good. I need to get back to fighting fires!

This was interesting to me as somebody who straddles both sides of this divide. I teach and provide consulting in Machine Learning (AI) on the one-hand and in tools for improving business effectiveness and efficiency on the other .. particular Lean Six Sigma (LSS), Design Thinking and Systems Thinking. In this specific context LSS expertise and methods provide a very natural partner for the capabilities of data science. After all, Green Belts and Black Belts in Lean Six Sigma are trained to execute high-ROI projects linking business problems to data:

• Identify business problems and opportunities

• Form hypothesis

• Collect then analyse data

• Refine hypotheses towards root causes

• Execute change to address root cause problems

• Institute on-going metrics that monitor and control processes in the longer term

Lean Six Sigma pracitioners at all levels are trained in statistics; Black Belts to a significant degree. And the type of statistical tools they learn to use are very focussed on business problems and opportunities..

So it seems to me that the skills of LSS are a natural bridge between the capabilities of data science and operational business needs. LSS practitioners are have the skills and training to identify high-value business problems, to collect data and phrase hypotheses and ideas in way that is hugely helpful to data-scientist doing their work.

I am delighted to be supporting Sent-AI-ance in an advisory and capability-building capacity. Their mission is specifically to bridge the gap between business needs and AI capabilities. The organisation brings together experts in Lean Six Sigma, Design Thinking and Systems Thinking with specialists in data-science and technology development. The absolute focus for Sent-AI-ance is to support businesses through the identification of opportunities and problems and then to execute technology projects to solve problems and capitalise on opportunities.

Over the next few months I will be writing further blog posts about the Four Thoughts Framework – the tool that brings together Agile Technology Projects with Lean Six Sigma, Design Thinking and Systems Thinking and which forms the core the Sent-AI-ance delivery model.