Machine learning (ML), business intelligence (BI) and artificial intelligence (AI); we have got beyond these being mere buzzwords. A whole industry has grown up around them, worth billions. And the number of success stories are growing across a diverse range of sectors. This technology is proven.
But at the same time, businesses can still be forgiven for not understanding what and how they can benefit from machine learning. And wondering if that is connected to other words and phrases that are kicked around?
Machine learning (ML) is one way of describing a similar type of solution to the same problem. Data. Data is the problem. Businesses now have too much of it. Businesses know this needs a solution, and many want to unlock new opportunities that are hidden in vast quantities of data.
So, ML is one potential solution to that problem. BI is another, as is AI; although that is at the higher end of the solutions spectrum and one that often requires considerably more computing power than machine learning. When it comes to ML, the answer usually involves an algorithm, data scientists, some serious work involving databases, and to bring it all together, software to provide an accessible interface.
Are businesses already using ML in practice?
Yes, in many different ways.
In some countries, test programs are already unlocking how to use ML to increase profits and pricing. Known as customer-driven pricing. One UK-based startup is deploying a complex algorithm to analyze 3-years of customer purchase history, to help companies increase profits 19% in only 9 weeks.
GE is already leveraging predictive analytics and connected devices to make maintenance preventative, thereby reducing downtime and improving efficiencies for industrial companies. Plus, this same technology is keeping their own Industry 4.0 systems operational across multiple countries and divisions. Amazon, always a pioneer of technology, uses machine learning and AI in many ways, from predicting what customers want to buy next, to improving how cloud-services run (AWS), to using AI-powered Kiva warehouse robots to pack orders more efficiently.
However, in many ways, this is only the begging. McKinsey found, in a study of over 3,000 executives, and 160 case across 10 sectors in 14 countries, that only 20 percent have active ML/AI projects running beyond the trial project stage. A further 40 percent are at that stage. Early adopters of AI and ML-powered technology are more likely to be confident that these projects can “grow their profit margin by up to five points more than industry peers”, according to the McKinsey study.
Larger companies have an easier time implementing these sorts of projects successfully, although big data-generated projects are starting to take hold in mid-size and smaller companies too. Access to the technology and skills that make all of this possible is getting easier as adoption rates increase, more data scientists graduates, and prices for off-the-shelf solutions go down.
One thing is clear; more companies than ever are putting time and effort into Investment in ML and data related solutions is expected to increase from $35.8 billion in 2019 to $79.2 billion in 2022, according to the International Data Corporation.
Telecoms and financial services are the two sectors currently investing the most and leading the way. Healthcare is also taking a keen interest. Compared to others, these sectors are sitting on more data and the upside of unlocking new opportunities could be enormous.
How can ML unlock new opportunities?
Machine learning and other data-related projects can unlock huge potential. You can uncover actionable insights. Create new products and services around what your customers really need (maybe what they are buying from a competitor), and even enter new markets with more innovative solutions.
However, as with any technology project, you need to get the fundamentals in place first.
Firstly, let’s make sure there is support at senior levels, and even the board. Companies that do well when it comes to ML/AI adoption have C suite, business owner and board level support. When that support doesn't exist, or isn't strong enough, projects can falter and fail.
Secondly, when thinking about how to implement a project: do you have the right skills in-place? Many companies bring skills and solutions in from trusted third-party providers. This is one of the reasons Google bought DeepMind, as there weren’t enough data scientists and engineers internally. So don't worry if you can’t do this internally. Get a trusted technology partner on-board to start analyzing what you need and how to implement a solution.
One thing to avoid is to put internal tech teams in charge of AI/ML-related projects. Yes, there is technology involved. But your IT manager or even CIO doesn't necessarily have the skills, time or experience to project manage something completely different. One of the ways these projects fall down is when the wrong person or team is tasked internally with managing an external partner.
Instead, to unlock the potential of the data you have, ML projects should be managed by a cross-disciplined team of business and technical leaders. It’s equally important to ensure any third-party delivering such a solution has the business analyst skills to ask the right questions and can help a big data team come up with the right answers.
How to implement ML internally?
In the short-term, start with something simple that has an off-the-shelf solution. Find a way to make a process more efficient, with the help of a proven piece of ML/AI kit, and implement at scale to demonstrate effectiveness.
Medium-term: Bring a technology partner on-board to analyze the datasets and databases you have. From these, gain an understanding of the questions that need asking and therefore algorithms that need creating to deep dive into the data and come up with the answers. As a result of this work, you should have a clearer idea how ML can make a real difference within your business, at scale.
In the long-term, either look at extending or expanding a pilot project, or identify other areas where solutions are needed and ML can demonstrate enhanced value. Every mid to large business is sitting on vast quantities of data. It used to be a byproduct of operations. Now data is an opportunity and growth area itself, and with the right team and tools, you can unlock this value and drive forward future growth.