Authored by Kim Nilsson, data science advisor, consultant, thought leader and speaker
I started my first data science business in 2013. At this time, AI was barely on the map and ‘data science’ was cutting edge stuff. Then, around five years ago, suddenly AI was all the rage. Everything that was cool was AI, and as a business exec if you were not talking about your AI projects you were a dinosaur. Conferences popped up everywhere discussing everything from how AI was going to change humanity to how it would impact their businesses (positively). As the owner and proprietor of a data business, we saw the boom coming and were gleeful that the market had finally woken up to the value of using data better.
Fast forward five years, have we seen this revolution happen? Have we seen hugely positive impacts on data from AI? Nope. We have not seen negative impact either, I hasten to add, but there has been very little ROI on that AI hype. Why is that?
I think there is a mix of reasons why companies have not yet truly taken advantage of their data assets. To start, all those execs jet setting around the world to preach their AI strategies? They were simply not aligned with the realities of their employees. There is typically a disconnect between what happens at strategy level and what gets done on the shop floor. Top level execs want progress and innovation, ideally as cheaply and quickly as possible. Teams and mid-level execs are often enthusiastic and keen, but lack resources to deliver. This leads to disenchantment all around, and can set teams back by years.
The lack of resources can come in different flavours. It could be a lack of sufficient data to work on, either because access is a problem, or because there is not enough data, or not enough clean and useful data. It could be a lack of understanding of 'where to start', or what is feasible, ie what is easy and what is hard to do. And it could be a lack of skills and experience to start and deliver data projects. The bad news is that this plethora of challenges can paralyse an organisation. The good news is that there are easy-ish fixes to all of these problems!
I think all of us would do good to turn the conversation from AI to ROI. Does it matter if it is true AI, if it delivers a tangible return for the organisation? Surely the opposite is not true? In my experience supporting more than 150 organisations on their data journeys, there is always ROI to be found in better utilisation of data assets. So how do we move on to the data ROI question?
First thing to do is to recognise the constraints of the organisation. Is data available? How much? What type? Do we have enough people with the right skills? And do we know what question to work on first? The latter is one of the key steps to unlocking your ROI; it is critical to identify what questions you could solve with data, and to prioritise them based on which is the quickest win. This way you build skills and motivation in the team, while already delivering ROI.
Dr Duncan Shaw, Assistant Professor in Information Systems at the Nottingham University Business School, and fellow Corndel Data Board member, explains more. ‘AI technologies are qualitatively different to other technologies because they take management decision-making to a completely new level – of precision, of speed, of scale and even the type of decisions. Trade-offs like quick versus accurate choices, or selective/personalised versus common/standardised will soon not be applied to many management decisions. And completely new types of decisions are rewriting industry structures and questioning the roles of all stakeholders. For example, decisions about purpose, joining digital ecosystems and entering completely new markets.
In the corporate boardroom, these have profound implications for an organisation’s vision, mission and values. Not just at the level of how new capabilities affect industry structures and business models but because of the questions they generate about ethics, accountability, transparency and liability.’*
Second, you need a data strategy that includes both a people and project roadmap. Identify individuals within the team with interest and motivation, upskill and enable them, and complement them with new hires or external consultants to plug any skills gaps. And work to build and maintain a roadmap of projects, including their expected benefits, their cost, and their risks. This way you have options once resources free up, and can work on increasing ROI over time.
Finally, target a positive reinforcement loop. The loop typically starts with ideation and update of the roadmap, build and deliver your first proofs of concepts or small projects, and prove the value of your data work. This will grow commitment among your team and the execs, leading to more resources and ideas for new projects. Ideally include all your stakeholders in the process, especially any employees of the organisation who will be affected by any outcomes, for maximum benefit.
I have seen many teams bang their head against the constraints mentioned above, and I have seen others flourish by patiently going step by step and building support and excitement at every stage by proving the ROI on data science work. Some of the most exciting work we did in my previous company were small and unsexy projects, such as when we delivered a 6% profit growth to a mid-sized reseller of car parts by tweaking their pricing strategy based on sales data. It was not AI, it was ‘just’ using data to price products better. A good data strategy (not necessarily an AI strategy) can deliver incredible ROI for companies and teams, both monetary and otherwise. So let’s stop talking about AI, and start talking about ROI!
This article is brought you by the Corndel/University of Nottingham partnership
Corndel and the University of Nottingham offer a streamlined suite of professional development opportunities for employees with different needs of data. The partnership brings together Corndel’s expertise in high quality professional training and the University of Nottingham’s educational excellence.
Kim Nilsson is an Advisor, Consultant, Thought Leader and Speaker. Formerly a Hubble astrophysicist and Chief Executive of Pivigo, Kim is a member of the Corndel Data Advisory Board. She helps to ensure the Corndel programme bridges the gap between academia and business, using her experience of building data science capabilities for the likes of KPMG, Barclays and British Gas.
Unlock the potential of your most promising talent with training, executive education and degree apprenticeships at the University of Nottingham. They offer a range of specialist higher-level degree apprenticeships in the areas of Data Science, Scientific Careers, Engineering and Healthcare.
The University of Nottingham is a pioneering university providing an exceptional research-led education. A member of the UK’s prestigious Russell Group, we are ranked eighth in the UK for research power with 97% of our research recognised internationally (REF 2014). The University is one of the UK’s most successful institutions in the fields of innovation entrepreneurship and commercialisation, and is a major industry partner both locally and globally. We are 103 in the world (18 in the UK) out of more than 1,000 universities in the QS World University Rankings 2022, with award-winning campuses and unrivalled facilities in both the UK and Asia.
Corndel delivers brilliant training in Leadership and Technology to the UK’s largest companies. Our training is tailored to each company’s bespoke objectives and delivers measurable benefits. We offer training in Leadership and Management, Project Management, Data Analytics, Data Essentials, Software Engineering, DevOps Engineering and Fundraising.