Recently, McKinsey Global published, "Solving the productivity puzzle", starting with the concern that aging economies are building projection models based on productivity gains from technology. The report observed, “Yet in an era of digitization, with technologies ranging from online marketplaces to machine learning, the disconnect between disappearing productivity growth and rapid technological change could not be more pronounced.”
What happened to the productivity gains promised by automation, AI and the augmented workforce?
These are critical issues with dollar-value implications for IT budgets in a climate of global volatility. McKinsey’s report deserves a closer reading to help guide business leaders on which actions could bring back productivity on both the macro and the micro scales.
Before we get into the specifics of the productivity puzzle, take a moment to question all of your assumptions, even those that you don’t know you have. Never forget that statistics is the art of crafting compelling stories from unstructured data. That may be bit harsh, but that’s certainly how statistics is normally practiced in the business world.
Confirmation bias is perhaps the most prevalent of countless deadly traps waiting in the murky world of data analytics. It's not hard to see why business leaders, who have achieved their own personal success by determination and beating the odds, are quick to agree with conclusions that support their intuitions -- and even quicker to ignore facts that contradict their organizing worldviews.
At the other end, data analysts who get promoted tend to be the ones who present findings that get them noticed by the C-suite. Science doesn’t happen in a political vacuum and that’s doubly true of data science.
After confirmation bias, the next set of insidious logical traps for the innocent statistician include correlation/causation attractors, the elusive significance of outliers and the slipperiness of commonly defined terms.
Any contemporary study of practical business applications for statistics deserves to run through the gauntlet of fallacies presented by Gertrude Mary Cox, an early statistical genius and founder of the journal Biometrics.
After navigating the treacherous shoals of logical disasters, there are issues that must be addressed with unreliable data sources, improperly normalized aggregates, flaws in model development, or inadequately supported interpretations.
That brings us to the world as it stands today, where warring statisticians prove opposite conclusions from the same data. Meanwhile, executives regularly plot out highly divergent action plans based on the same reports.
No one can afford to accept statistical analysis or conclusions without going back to the data themselves, accompanied by an experienced and wary statistical guide. With all that in mind, today's question is:
"What is really going on with productivity at the level of businesses, not economies, and what should you do about it?"
Let’s take a look at the unstated assumption that productivity on the national or regional scale is reflected on the micro-scale of individual firms in aggregate.
That’s a useful model for switching between micro- and macro-arguments within a limited space, but this simplified model of the world sheds too much valuable detail.
In our own case, Tangelo is living proof that productivity improvements from AI and digital transformation are very real, with measurable performance impacts. There’s no question that we are more productive now than we were a year ago. Our turnaround times have been cut by more than half. The number of pitches we've put together are at an all-time high and we have aggressively penetrated a number of original industry markets.
That’s a very different story than what has been happening in the macro-economic ocean that we swim in. The graph below shows U.S. productivity closely matching the average of select Western nations for two centuries.
Productivity growth over time for US vs average of select Western nations, 1871–2016, %
U.S. workers reached peak productivity around the end of the first World War, though there was a small uptick after computers were introduced in the 1950s. The country’s current plummet toward zero productivity began its downward turn around the time of the dot-com bubble.
In the case of Tangelo, we feel that we owe our success to expanding our facets of diversity and having majority female senior leadership. McKinsey’s own report on "Why diversity matters" concluded, “New research makes it increasingly clear that companies with more diverse workforces perform better financially.” Along the same lines, Fortune magazine reported that women-owned businesses are performing three times better than the S&P 500.
The next argument, however, is a bit more subtle and the answer more complex.
The McKinsey report predicted, “[T]he nature of digital technologies could fundamentally reshape industry structures and economics in a way that could create new obstacles to productivity growth.”
The critical word in that sentence is “could” because there is no way to argue with a contrafactual. There are many possible futures, but a budget can only be built on the balance of probability.
It’s impossible to fully tease out the economic impacts of global instability, natural disasters and historical cycles. Still, on the grandest of scales, digital technology’s staggering advances have not changed the direction of productivity trends. McKinsey suggests factors such as the time lag effects of business readiness, transformation costs and the dissolution of legacy systems are artificially pinning down productivity gains in the short term.
The big picture, however, is very different from the measurable ROI that firms have posted from specific AI, machine learning and business intelligence (BI) projects. One size certainly does not fit all. The time spent up front aligning the digitization project with company objectives will have determining impacts on the success of execution and employee adoption.
According to the latest AI research by Gartner, while it’s true that the nature of work is fundamentally changing, the financial reality is that AI will add trillions to the global economy while breaking even on job losses by 2020 and generating 2 million net-new jobs by 2025. Gartner’s report concluded that by 2021, AI augmentation will add $2.9 trillion in business value to the economy and recover 6.2 billion hours of lost productivity.
In the end, as McKinsey’s report suggested, “Furthermore, continued research will be needed to better understand and measure productivity growth in a digital age.”
Subscribe to Insights at Tangelo