Why Digital Transformation Depends on a Machine-First Approach to Analytics, and How to Achieve It

Many companies are not retrieving the value they expect from their digital initiatives and they don’t fully understand why.

In general, enterprises have been slow to make progress on their digital transformations and, in particular, their analytics-led efforts. A recent Forrester Research survey of 1,600 business and IT decision-makers found that while 56% of firms are working on digital transformation, the scope of their work and the size of their  investments are surprisingly modest. For example, only 34% of banks and insurers are targeting marketing, and only 45% are focusing their initiatives on customer care.1 A 2018 Gartner’s survey of 196 organizations found that 91% had not yet reached a “transformational” level of maturity in data and analytics, despite that being the top investment priority for CIOs in recent years.2 Indeed, 60% rated themselves average or below average in data and analytics maturity.

Are digital transformations driven by advanced analytics complex, time-consuming and hard? Yes. But there is more to the relatively slow progress being made by the majority of these initiatives – and their self-limiting scope – than a lack of  commitment from the top. Instead, there is a fundamental misunderstanding of what an analytics-based transformation can do for the enterprise.

Today, businesses need to raise both their expectations and aspirations or risk falling behind in the race to data and analytics mastery. And the way to do that is by taking what we call a machine-first approach, in which whatever can be automated, will be, freeing the business and its employees from rote processes to create true enterprise value.

Why Most Transformations Don’t Deliver

A major reason that companies are not getting enough value from their digital transformations is that their leaders are failing to grasp the opportunities that today’s data analytics and related technologies present to increase operational efficiency by an order of magnitude through business process change. Instead, companies implement analytics systems to create efficiencies and cut costs without attempting to transform a business process such as demand creation (marketing and sales), order fulfillment, or concept-to-market (product development, marketing, sales, customer service).

For example, to reduce inventory, a manufacturer might create an analytics-driven view of its stock, and then move materials among various warehouses depending on which site has a surplus and which needs more. This will generate an incremental improvement in costs, not a dramatic one.

By contrast, optimizing analytics could improve forecasting – leading to both top-line gains (by preventing stock-outs) and cost reductions (by improving turns and reducing stale and costly inventory), while better inventory classification can enhance cross-department communications (leading to improved sale efficiency).3

Why do enterprises set their sights so low? For one thing, many managers are trapped in mindsets that formed when companies operated in an environment of data scarcity, as well as less advanced technologies for capturing, processing, and analyzing data, as well as using it to make decisions and execute. They had little data on critical aspects of their business such as customer opinions, or the performance of their products in the field. And even if the software to analyze such data was available, it was too complex for leaders and managers to use. Added to that, computing power was costly, and thus limited to processing  customer order data and other operational nuts and bolts.

But four fundamental changes in this decade – in data, analytics, computing power and automation – now enable businesses to capture, process, analyze and use data to operate far more effectively and efficiently:

  • Data collection. Companies today can collect data from outside their enterprise systems – opening them up to the larger world – and in many more forms, including audio and video, social media, IoT outputs from sensors embedded in their products, sensors and video installed in the places they do business, and from their customers’ mobile apps.
  • Compute power. Cloud providers now enable even small companies to rent affordable, immense computing power to process and analyze data about their business conditions and make rapid and necessary changes to the way their business is operating, from marketing campaigns to manufacturing and distribution processes.
  • Analytics and AI. Analytics today are more user-friendly and deliver high-quality data visualizations allowing executives to make better, data-based decisions. Even more importantly, artificial intelligence can help companies automate the decisions and actions they take on changing business conditions, such as automatically rerouting a delivery truck or tweaking marketing campaign messages.
  • Automation. In recent years it has become possible to have computers take autonomous actions through robots and AI systems that can process voice commands and help customers answer questions and solve problems.

These advances rise in importance as data flows continue to grow, as the number of communications channels expand, as consumer populations rise and the products they use multiply—all leaving less time for analysis and decision-making. It’s all too much for a typical business user to manage. In such circumstances, it is important to shift the workload of operational decisions to machines that are capable of handling the load.

Companies that pursue “digital transformations” without harnessing the benefits of these advances will still see incremental improvements in cycle times for order fulfillment and customer problem resolutions. But they won’t get the dramatic gains they increasingly need. They will continue to perform much more manual work than necessary or advisable. Their tweaks to customer service and marketing and sales campaigns will neither be large nor swift. In short, they won’t truly transform. Without leveraging advanced technologies, they will be like an automaker designing a new car with 20-year-old tools, unaware that sensors now can alert drivers to road hazards and maintenance issues and enable consumers to navigate traffic jams while reserving a table at a restaurant.

The Promise of a Machine-First Approach

To take full advantage of the opportunities advanced technologies offer means taking what we call a machine-first approach to digital transformation. The machine first approach uses automation and AI to complement human capability to enhance the potential of the enterprise.  The approach changes course from traditional ways of working. Instead of a business process redesign influencing the use of data and analytics, machine-first promotes automation that enables companies to rethink their business processes, eliminate superfluous work and lay a robust data foundation to leverage analytics.

Powered by analytics and AI, a machine-first approach drives greater efficiencies to create exponential value improvements for enterprises that execute. As a company deploys intelligent software to automate actions, a machine-first model relies on agile approaches to business, emphasizing rapid trials and applying lessons learned from those trials in an iterative fashion.

While describing this approach, it’s important to note what machine-first is not: a headcount reducer. Rather, it is a way to free people from routine and repetitive work.

Analytics is integral to a machine-first approach. The companies that are best at analyzing digital data are leaping ahead of their competitors because they can generate unique insights in what to produce, who to sell to, and what and when to promote. They are also better at understanding and managing their finances, supply chains, productivity challenges and human resource needs.

Integrating analytics with a machine-first approach produces benefits on an exponential scale. In retail, data scientists and analytics experts are helping Amazon and Walmart out-compete many established grocery and general merchandising companies.  Amazon’s share of all retail sales is predicted to nearly double, from 5% in 20184 to nearly 10% in 2020.5 Though it got a relatively late start, Walmart’s e-commerce business, with a 3.7% share of the U.S. ecommerce market, is rapidly growing thanks to analytics, and is expected to increase 40% this year.6

Adopting the machine-first approach for analytics represents a departure for most enterprises in that it’s not the sole responsibility of the CIO or IT leaders to drive adoption. Rather, it’s a critical enabler for all C-suite stakeholders. In a machine-first world, automated systems can enable faster and more precise decision-making. For example:

  • For the CEO and heads of strategy, analytics can help them craft business strategy, decide where to pursue new market opportunities, whether the current business model is about to become obsolete, and how and in what ways it needs to change.
  • The CMO needs to prepare for the next wave of digital transformation: mass personalization leading to event-driven marketing (EDM), in which companies provide timely, relevant offerings to show they understand customers’ changing preferences. And, with EDM, a retailer no longer delivers a tailored promotion when it wants to push a product, but when the customer demonstrates a need. That requires becoming alert to the customer’s need at the moment it arises and fulfilling it at the optimal time: before he or she begins to research a solution.7
  • For supply chain executives, analytics present opportunities to save money and increase productivity. Oil and gas companies have reduced maverick spending by analyzing procurement data and finding waste and fraud; others have used analytics to identify less costly suppliers. Automotive companies are using analytics to identify potential supply chain weaknesses, enabling proactive countermeasures before costly problems emerge.8

The benefits of a machine-first approach to analytics cuts across industries. In the pharmaceutical sector, it enables the automation of certain aspects of clinical trials and can automate processes for getting drugs into the hands of patients that need them. In insurance, automation can speed the processing of claims while reducing errors and improving the customer experience. For manufacturers, it can automate iterative, heavy tasks, thereby improving productivity and increasing worker safety. In a machine-first environment, workers support the machines rather than performing the tasks.

Five Key Steps in A Machine-First Approach

The machine-first approach requires identifying the work that machines should do – both physical and analytical – to give the machines the “right of first refusal.” That means using analytics in combination with AI to enable the automation of many operational tasks. To do this, one must:

1. Identify activities that are voluminous, repetitive and prone to mistakes and therefore ripe for automation.

For example, target areas where workloads are heavy and duplicative, such as in an insurance claims and reimbursement processing center. Indeed, many enterprises have yet to automate core business processes like closing the books or legal document analysis. If machines can do the job, they should – i.e., the right of first refusal. By understanding where a company’s weaknesses lie, it can address them.

2. Assess your company’s current data and analytics maturity.

If a company tries to automate tasks based on bad or incomplete data, it won’t be effective. There are three layers to data maturity:

  • Defining the core data that drives your business (such as machine performance and maintenance data in a manufacturing plant).
  • Ensuring data is accurate and available for analysis in real-time.
  • Linking analytics with other systems that manage operations to create a platform for generating insights.

3. Build a data foundation.

A robust data foundation will enable a broad range of users to consume the products of analytics. To democratize the use of analytics, and thereby foster a data-driven culture, train your business users and give them tools like data visualizations to support their work. In addition, organize collaborative events that include ideas from outside the company, such as hackathons, to make the enterprise more aware of opportunities to improve business through analytics.

4. Apply AI to tasks ripe for automation.

Begin projects to test these activities and use agile approaches to iterate and refine implementations according to enterprise priorities..

5. Keep security and privacy top of mind.

Giving machines the right of first refusal requires understanding the relevant security and privacy risks, and the applicable regulations, as you work to automate tasks and improve operations.

The Right Time for Machine-First Analytics  

Companies today face an ever-growing number of decisions that must be made quickly. These decisions must be informed by data. A machine-first delivery model will make it possible for people to work in collaboration with AI-enabled machines to automate repetitive tasks while supporting decision-making throughout the organization. This will transform performance and allow companies to begin retrieving the benefits of advanced technologies. If they do, digital transformation will become a reality not an aspiration.


About the author(s)

Continue your journey


  • Taking a Machine-First Approach to Cloud Migration


    Many companies have put cloud computing at the center of their digital transformations. Yet a number are struggling to shift their systems from on-premises technology to public clouds. Several challenges are slowing them down, but they can be overcome with…

  • Assuring Digital Customer Experience (CX) – the role of Quality Engineering in Digital Transformation programs


    Adoption of digital technologies has led organizations to have omni channel reach to end users increasing business volumes exponentially. This has been rightly enabled by ever expanding end user ecosystem, such as increasing usage of personal devices, social integration, etc,.…