This is the third of three blog posts looking at how the turmoil of 2020 has created divides between successfully digitally transformed enterprises and those who are falling farther behind.
Three Kinds of Digital Transformation
In prior installments, we noted how successful organizations have some common attributes, while less successful groups tend to have the opposite attributes. That pattern holds true in the adoption of big data.
Big Data Adoption
Even before the pandemic, Harvard Business Review reported that most companies were not “data-driven” and a decline in the number who claimed they were.
Before Covid-19, big data had plenty of problems. Among those was conflating big data with AI adoption. The truth is there are many good uses of massive data sets without AI, and AI has many uses without truly big data. Other challenges are institutional problems described in Part 1 of this series.
But the epidemic and recession have added three new divides, which create a tale of two cities: those who are accelerating their use of data and those falling farther behind.
Divide 1: Does your big data describe a world which no longer exists?
For many organizations, this is a crushing problem. Travel and entertainment are a couple of examples. Years of travel data and patterns driven by seasonal trends, business travelers, meetings and conventions, and holidays are all out the window. The futility of looking at this old data is one reason why there have been massive layoffs in the data departments in airlines and T&E service providers.
While airlines and hotels are easy examples, many others in organizations are very busy. Some retailers like Walmart and Amazon have seen huge increases in online sales. Costco also saw significant increases, but perhaps only 30-50% of the gains at Walmart. Costco struggled to scale online when they needed it most. Part of the problem appears to have been the use of pre-Covid data patterns.
In aviation, Maintenance, Overhaul, and Repair (MRO) providers supported airlines and were hit with two shocks, the Boeing 737 Max grounding and the epidemic. Changes in airliner usage make much of their historical data useless. In contrast, some segments of the private jet fleet saw an uptick, and a confusing problem for MRO operators was whether this would drive more overhauls when supply chains were struggling.
A risk in pure, data-driven AI is that your data is history; it tells you about the past. It’s possible to drive your car by only looking in the rearview mirror for a while. Covid-19 is showing that for some firms, this approach led them to run a few red lights. They didn’t know it until something terrible happened.
As the pandemic fades with the introduction of vaccines, yet another “new world” will emerge. Pre-Covid data and Trans-Covid data may not be useful in the Post-Covid world.
This is why Lone Star generally prefers Hybrid AI, with a blend of data-driven and cause-effect mathematics, for better prediction and prescription.
Divide 2: Do you need to be right, or just less wrong?
In 2020, big data has been more successful for organizations that needed to be “less wrong.”
In transportation and logistics, the estimated arrival time is much more complicated than estimating the drive time on a morning commute. It is a challenge in the best of times. But during the epidemic, estimates have been nearly impossible for some shippers and some supply lines.
Big data derived from historical patterns were obviously useless. But another limitation emerged. Most bulk transports lacked any asset visibility. Shipping containers and railcars simply disappear into the transportation network, with a hope they will return. Prior to the epidemic, estimates of arrival were an effort to be “less wrong.” No one took them seriously. Now it’s clear we need to be right about asset insights across our transport and logistics networks. That’s led to system-wide changes like Rail Pulse, which Lone Star is supporting.
Divide 3: Top-line vs. Bottom line – can your data and analytics support both?
Many retailers saw their online sales exploding. They didn’t need product recommendation AI to sell more stuff. But they did need to understand their need for call center customer support and other capabilities, which didn’t scale by simply turning on more cloud capability.
Retailers, shippers, and airlines all saw dramatic changes, which changed their operating structures and the relationship between their top and bottom lines. In most cases, basic big data was simply useless.
Most big data sets aren’t analyzed in the context of business rules and operating constraints. It can’t show your limiting factors, and it doesn’t know when your limiting factors have changed because your younger employees are home with kids who can’t go to school.
Government agencies have a different kind of “top line” and “bottom line.” Usually, they think in terms of their approved budget (their top line) and the benefits, value, or mission goals they pursue (their bottom line). Like any other organization built on humans, these agencies found 2020 difficult. They struggled with employees who needed to take care of vulnerable loved ones or who could not find childcare. Their goals changed much faster than their budgets in most cases. And their “top line” became disconnected from their “bottom line,” much like corporate organizations.
What Big Data and Analytics Actions Make Sense for 2021?
First, realize the world is changing again, and we all hope 2021 won’t be described by 2020 data. Let 2020 be a lesson about the dangers of driving only with the rearview mirror. An investment in the analysis must include prediction and prescription, not history.
Second, be thoughtful about analytic goals. What does your organization really need? Is being less wrong good enough? Being right may require more sophistication than big data analytics and visualization can provide.
Third, refine your goals in terms of both top-line growth and bottom-line performance. Tie both the objectives and the structural design of your strategy to the top and bottom lines.