Our client had a strong legacy of improving therapeutic outcomes for patients by perfecting the mechanics of medical devices. But this avenue for improvement had been exhausted. To continue to drive innovation, its leadership team looked to the healthcare arm of its business. It asked us to consider how it could use data, analytics and AI to personalise or adapt therapies to improve patient outcomes.
The organisation wasn’t short on data. It had a wealth of historical data made up of machine data and electronic health record information. Yet this was scattered across an entire multinational organisation. It was difficult to access, bound by legal regulation and never meant for AI use.
We also quickly realised this was more than a data gathering exercise. The client’s culture, organisational structure and IT systems needed a drastic shift to be more suited to fast, market-facing innovation. How could we take its existing capabilities and create an organisation where analytics and AI could flourish?
Building an innovation unit for data and beyond
By the time the project was finalised, we’d created a team of seven people, established a way of working and built trust throughout the organisation. The team was up and running quickly, taking its first few projects from proof-of-concept through to implementation. A method of attracting talent had been established. And the organisation owned a new culture it would continue to shape.
Mapping processes, gathering data and shaping an innovative culture
The project would see us focus on three areas: 1) tech and data infrastructure, 2) people, and 3) culture. In order to achieve the dramatic shift we wished for, we needed to show people why it was important to introduce new applications of technology. To accompany new technology, people and culture would be key.
We began by gathering an understanding of the organisation’s processes around the globe – in the US, Spain, Italy and China. We identified the people involved in important tasks, as well as their skills and roles in collecting, curating and analysing data. We then focussed on shaping the organisation to support its long-term ambitions and created a method for finding and hiring the brightest talent.
We plotted four future scenarios to help us consider how the organisation created data, what data might be needed in the future and what data assets already existed. We also conducted a rapid exploration to uncover where we had data quality deficiencies. Once we had our findings, we set out to create a team and the necessary processes to give us a grasp of data ownership and achieve success.
No one gets excited about tech unless they know what it’s going to help the organisation achieve. Based on this, we built a network of senior stakeholders and trust within the company. This unlocked cooperation from key people with deep knowledge and experience who’d become advocates for the project. It was the making a new culture centred around data and innovation.
What made it successful?
Do not hire data scientists when getting into data science (at first)
If your organisation is new to data/AI, it will take a very long time to ensure data availability and data access. It can easily take 12-24 months. This kind of work also requires only minimal data science know-how and resources, so isn’t suited to the expertise of data scientists.
Instead, the focus should be on overcoming legal, political and cultural barriers. Having data scientists being forced into project management or consultancy roles will frustrate them. You’ll go through at least one generation of hires (making it more difficult to hire the next one!).
For projects like this, from the outset you’re much better off hiring an experienced, technically-minded project manager who not only understands Data Science but is also experienced in cutting through company politics. Then, set up your data science team when you’re ready
Do not start with the data when looking for opportunities, start with business opportunities aligned to overall strategy
The ‘data is the new oil’ metaphor is highly misleading as data is not valuable per se. Even though it’s useful to have an overview of what is there in terms of data at the moment, to help generate ideas and allow for a first assessment of feasibility, data should not be the focal point of formulating a strategy
This is, in large parts, because it’s likely the data you may need for the most relevant opportunities is currently not there. Even more valuable than the data you currently have is your position in the value chain that allows you to capture data.