As with digital transformation some years ago, multinationals have been in the vanguard of data science transformation. It is often true that mid-sized businesses – categorized in this article as those turning over more than €100m – may not be as far down the data science transformation journey as global blue chips. That said, leaders of most mid-sized firms understand the value, power, and potential of data science and the application of AI and machine learning to enhance business decision-making; to enable them to make smarter, quicker, better-informed decisions at scale. And while they may be slow to get going, they can accelerate faster thanks to the inherent agility of a smaller organization.
Data science transformation: 3 approaches that businesses typically take
The tools-first approach almost always fails, attempting as it does to fit the square peg of a company’s specific data science needs into the round hole of an off-the-shelf data management platform. No one provider offers the entire chain of data services required and comprehensive suites of services tend to be too rigid. Simply buying a product does nothing to address the cultural change needed to become a data-driven company. Tools require team members brought on board to run them, and training to learn how to use them to best effect. Products alone cannot bring about long-term, sustainable data science transformation and enhanced decision-making.
Creating a team from scratch is also time-consuming and something of a mythical, promised land. You need to start by hiring a seasoned head of data – one with experience with a variety of different tools as well as a strong intuition of business impact – to make the right data architecture choices and develop a long-term data strategy. With a leader in place, you need a team who can execute that strategy. And yet recruiting data scientists to develop an in-house solution can result in hiring second- or third-tier quality data scientists who are required – from a blank sheet of paper – to reinvent the wheel. It’s challenging to build a team from scratch when you haven’t yet determined your data roadmap and what mix of profiles you need. It’s even harder to retain them.
Think big, act small!
Genuine data science transformation requires a business both to use a variety of tools to build a data platform – one that can ingest, store, align, and consolidate all of the data that matters in a business – and then put that platform to work in a progressive series of use cases that demonstrate value as you go. In our and our clients’ experience, it is the middle – hybrid – way that is by far the most effective, both for global multinationals and mid-sized businesses.
By working with third-party experts who have effected data science transformation many times before, businesses can build their long-term data vision. They can then assist in recruitment the right team members and at the same time start to implement a data platform and change the way business operates. Thinking big – with a company-wide vision of the data platform and organization as the overarching goal – but acting small by quickly implementing a few well-selected use cases, businesses can make measurable, tangible progress in a handful of key areas fast. There is then a flywheel, accelerating effect, as the same, transformative approach is applied to other aspects of the business according to a strategic roadmap for change. The first use cases enhance and enlarge the core data available, demonstrate the value of the platform, and serve to excite colleagues in other functions to start using data smarter.
Businesses just starting out on a data science transformation journey often identify initial use cases around data visualization and descriptive performance statistics using medium-sized data sets.
These use cases are critical for creating the foundation of reusable data that can be built upon, and simply having clean, timely, visualizable data can already have a powerful impact on how business decisions are taken. This then establishes a base that can pave the way to AI and machine learning based on complex algorithms using multiple, big data sets to predict customer behaviour and even develop new products and services. This approach immediately brings experienced, external data scientists and engineers to assess a company’s prevailing data science maturity. This, in turn, allows them to determine what foundational platforms are required to identify and to prioritize the opportunities that more rigorous data science transformation could deliver in terms of smarter, quicker decision-making at scale. This rapid evaluation of context and opportunity allows us to outline the potential for data science transformation as well as likely hazards and speedbumps along the way. Critically, this approach quickly identifies areas where the biggest impact can be made first, enabling businesses to reap the rewards and demonstrate value within as little as eight weeks. Ekimetrics’ approach pairs mid-to-long-term transformational change with the rapid demonstration of business value in two or three immediate use cases.