Charted: Lessons from Machine Learning Fast Learners

Michael S. Goldberg, Christopher Koch and Dan Wellers

Companies that are benefiting soonest from their investments in machine learning have gained by going all in, according to a recent global survey by the Economist Intelligence Unit (EIU) and SAP.

Among 360 executives at companies in a range of industries, 21% report that they have already seen tangible gains from implementing machine learning applications. Labeled “Fast Learners,” these companies are more likely than others that are experimenting with machine learning to have an enterprise-wide strategy, high-level leadership support, and acceptance of the technology throughout the organization.

Fast Learners are increasing their profits. Meanwhile, their experience points to potentially profound changes in how companies will be organized in the future. These companies are shifting away from outsourcing to far-flung, low-cost regions and are tapping more in-house and local people.

“One of the most interesting characteristics of the Fast Learners was their big-picture approach to machine learning,” says Kevin Plumberg, EIU managing editor. “They were more likely than other kinds of machine learners to apply its applications across the entire enterprise. This was the case whether it was a big or small company.”

“That doesn’t mean that the Fast Learners didn’t start to introduce machine learning into their business in a limited way. They scaled it out, certainly. But they were thinking about it in a more holistic way,” he added.

Strategy Comes First

As researchers such as Erik Brynjolfsson and Andrew McAfee point out, machine learning systems, in play since the 1950s, have seen their fortunes rise recently thanks to the proliferation of data, order-of-magnitude increases in algorithm quality, and robust gains in computer processing.

The EIU survey found that Fast Learners are more likely to take an enterprise-wide approach to implementing machine learning systems compared to firms that have deployed the systems but have yet to realize benefits.

The findings indicate that among the Fast Learners, machine learning is part of a larger strategy in which organizations are rethinking their business models and value propositions to customers. Fast Learners cited the following challenges less frequently than the machine learning users that have not yet seen benefits: a lack of strategic clarity, a lack of organizational leadership, and the need to counter organizational resistance to changes involved in implementing machine learning systems (see Figure 1).

According to Plumberg, these results suggest that Fast Learners have been carrying out their machine learning programs with the necessary forethought and care to manage both business process changes and employees’ adaptations to them.

About one-third (31%) of Fast Learners also credited machine learning implementations as having helped them pursue innovations in their business processes or business model. While this was not among the most frequently cited benefits, Plumberg says the association of machine learning with innovation suggests that Fast Learners recognize the strategic value of machine learning systems and how they positively influence their business.

Poised to Profit

Almost half (48%) of Fast Learners cite increased profitability as the most significant benefit from applying machine learning to their business processes, compared to 32% of executives who have not seen the same benefit but expect to by 2020. Meanwhile, cost savings, cited by 34% of Fast Learners, was the most anticipated benefit among the other executives (44%).

Another key difference between Fast Learners and other firms that have deployed machine learning is their expectations for the future. Nearly half (48%) of Fast Learners said they anticipate revenue growth of more than 6% through 2019, compared with 30% of other machine learning users (see Figure 2).

Machine learning can be a means to reshape a business model. Top executives at Pinsent Masons, a global law firm based in London, are examining how machine learning applications can restructure their client relationships by changing the traditional services-based model. The firm aims to become a provider of knowledge-based systems and turn hourly rates into licensing fees.

The Machine Learning Organization

As they apply machine learning, Fast Learners are rethinking how they source their business processes—decisions that Plumberg suggests signal potentially far-reaching changes in how companies function. Fast Learners are more likely to rely on in-house and locally sourced resources instead of outsourcing those tasks to low-cost regions around the globe, compared to other machine learning users (see Figure 3).

Overall, 74% of machine learning users expect to increase their use of in-house and locally sourced resources by 2020. Fast Learners appear to have accelerated these sourcing decisions, spending more on locally sourced resources today than others.

The correlation between deployment of machine learning and local sourcing suggests that companies want to keep a close eye on their newly automated business processes. Consider an automated customer service system that replaces a traditional call center. Customers interact with an application enabled by artificial intelligence to get help on a question or request service instead of calling into a service office to talk to a person. This is the kind of system that a company would want to keep nearby instead of shipping it overseas, says Plumberg, because the customer experience is viewed as key to competitive advantage.

The shift in sourcing priorities probably does not predict a sudden surge in onshoring, says Plumberg. Rather, it points to a slow but steady change in how companies evaluate the benefits of outsourcing: from a traditional cost-based focus to one based on business relevance and customer value.

An example at Intel illustrates the potential to use machine learning to bolster core functions like sales and marketing, Plumberg adds. Faced with limited resources for its sales and marketing organization, Intel could have considered outsourcing to support its efforts. Instead, Intel built a machine learning platform to enhance an internal process, helping its sales and marketing teams identify which resellers their customers should work with in specific vertical industries.

Moving Forward with Machine Learning

Even Fast Learners have challenges; in particular, as illustrated in Figure 1, they identified a lack of machine learning expertise both within and outside of their companies.

This is a common complaint among companies seeking data-savvy talent. To address it successfully, the capabilities needed for machine learning must be included in strategic discussions in the C-suite, Plumberg says. For companies still early in their machine learning efforts, Fast Learners engage in several behaviors worth emulating:

  • Think beyond your current business. As with any major technology initiative, you’ll pilot one function or process first. But view the project through a wide-angle lens. The effectiveness of machine learning relies partly on its ability to analyze data from, and coordinate actions across, several different enterprise functions. Machine learning–enabled improvements can lead to new business models.
  • Don’t be cowed by tech giants. The biggest technology companies may be leading the machine learning charge today, but firms of all sizes have unprecedented access to online machine learning innovators and cloud-based computing power. Small companies’ classic advantages of speed and entrepreneurialism may count for more. In fact, companies with less than US$750 million in annual revenue accounted for 69% of the Fast Learners identified in the EIU survey.
  • Don’t wait. Machine learning is moving out of the science lab and finding its way into everyday business. Companies that began testing the waters a few years ago are now moving ahead. By 2020, the gap between Fast Learners and the rest will have widened. Delaying the implementation of a machine learning strategy will likely mean falling behind. D!

Michael S. Goldberg

About Michael S. Goldberg

Michael S. Goldberg is an independent writer and editor focusing on management and technology issues.

About Christopher Koch

Christopher Koch is the Editorial Director of the SAP Center for Business Insight. He is an experienced publishing professional, researcher, editor, and writer in business, technology, and B2B marketing.

About Dan Wellers

Dan Wellers is the Digital Futures Global Lead and Senior Analyst at SAP Insights.