Whitepaper: Why do analytics and AI projects fail?
A new report from Melbourne Business School’s Centre for Business Analytics reveals why more than 80 per cent of data science projects fail and how to avoid the common pitfalls.
The rise of data science has marked a significant shift in how organisations operate and make decisions.
As businesses and institutions accumulate vast amounts of data, the ability to analyse and interpret this information has become crucial to gaining a competitive advantage.
Yet despite the hype, the level of interest from executives and the number of analytics and AI frameworks available, research shows that more than 80 per cent of all data science projects fail.
A new report from the Centre for Business Analytics seeks to end this endemic failure by identifying the recurring themes in failed projects from around the world and providing practical recommendations on how to overcome them.
“There are hundreds of reasons as to why this is, so knowing where to start can often feel overwhelming,” says Dr Evan Shellshear, lead author on the report.
“By focusing on the primary issues which were causing most of these failures, we were able to distil them down into four thematic areas, and then provide key recommendations to address each.”
Analytically mature vs analytically immature
A key distinction the paper makes is between organisations that are analytically mature or analytically immature.
An analytically immature organisation lacks most of the requirements to deliver on analytics projects successfully – things such as in-house capabilities, high-quality data or buy-in from senior stakeholders.
Mature organisations, on the other hand, tend to have most of these elements.
“Analytically mature organisations generally have the required human and technological resources, and have experience applying analytics and realising at least some commercial benefits from these activities,” says Dr Shellshear.
“They are not ones who typically suffer from the failures we discuss in the whitepaper.
“It is the analytically immature organisations, which are the majority, that will benefit greatest from our report.”
One alarming finding to come out of the project is that the failure rate of projects for analytically immature organisations is likely to be far greater than the documented 80 per cent.
“There are numerous reports and peer reviewed studies which place the failure rate of projects at 80 per cent,” Dr Shellshear explains.
“However, these studies don’t differentiate between mature and immature organisations.”
Through a detailed analysis, the authors found that the failure rate for analytically immature organisations is more likely to be around 90 per cent – with mature organisations sitting at approximately 40 per cent.
Distilling the data
To understand the forces behind these failure rates, the authors analysed more than 100 pieces of content from blog posts, whitepapers, podcasts and videos, more than 2000 peer-reviewed articles, and spoke with dozens of world-leading practitioners.
They were then able to group various failures into four key areas, under the themes of People, Process, Technology and Strategy.
This is where the distinction between the two types of organisations is essential, because we begin to see that a lot of the common failures are due to these immature organisations lacking the fundamental requirements to successfully deliver their projects,” Dr Shellshear says.
Providing the solution
As well as providing clarification and definition around common problems, the paper also provides practical solutions to address each thematic area.
Ultimately, we want organisations to succeed,” Dr Shellshear says.
"If this high failure rate persists, it will significantly damage the analytics industry.
Therefore, it is essential that the groundwork be laid in a way that is palatable and maximises the chances for organisations to improve their experiences with data science.”
Professor Michael Smith, Chair of Management in Econometrics at Melbourne Business School, highlighted the importance of collaboration between industry and academia to address these failures."Academics bring deep analytical expertise, while industry provides the contextual insights and practical challenges,” he said.
"By working together, both sides can bridge the gap between cutting-edge theory and real-world application, improving the success rate of data science initiatives.”
He also emphasized that these collaborations often lead to innovative solutions that neither party could achieve alone.
Dr Evan Shellshear is the Managing Director and Group CEO of Ubidy, an innovative global recruitment marketplace that leverages AI to connect employers with specialist agencies. He holds a Bachelor of Arts and a Bachelor of from the University of Queensland and PhD in Mathematical Economics (Game Theory) from the Institute of Mathematical Economics at the University of Bielefeld.
Dr Shellshear is currently an Adjunct Professor at the University of Queensland and Queensland University of Technology, where he teaches business analytics and AI strategy
Download the full whitepaper Why Do Analytics and AI Projects Fail? to learn how to reduce the chance of failure for your organisation’s data science projects.
For more analytics information and research, contact the Centre for Business Analytics if we can help guide the success of your organisation's analytics and AI projects.
To find out more about studying at Melbourne Business School, visit our Degree Programs and Short Courses pages, or learn about our range of services For Organisations
