The oddest venue where I've ever delivered a talk was the QEII. In 2000, a conference group had hired out the ship, which carried an audience of CIOs east for a day and a half out of New York, then did a U-turn and sailed back. The group hired me to give two talks on how the CIOs could regain control of corporate data processing following nearly two decades of decentralization caused by the spread of personal computers and local area networks.
A key concern was the spreadsheet. Seemingly every person had one, containing data that might or might not be current and that might or might not come from a source that was recognized as authoritative throughout an organization. Basically, every person could have their own version of truth, and big organizations have thousands or even tens of thousands of people.
That problem persists. But the sort of theoretical solution I laid out 20-plus years ago is becoming practical today, because the massive improvement in computing power and connectivity means we can rethink what a spreadsheet is and how it should be used.
The basic idea is to centralize control of data even if you decentralize its expression.
So, you make sure that all important data comes from a corporate source or can be vetted centrally after someone pops it into a spreadsheet. That's not only key for data used in underwriting, claims processing or some other business process. It's key, too, for data on the size of potential markets, the economic outlook, etc. Everybody needs to be on the same page.
You also make sure that there are live links from a spreadsheet back to its source, so the data gets updated in real time. That way, you don't make decisions based on spreadsheets that have been frozen in time, like a snapshot from months ago. The spreadsheets reflect reality, as perceived by the corporate as a whole, at any moment.
Users can still apply analysis and expertise to the spreadsheets, but they become the means for displaying data, rather than a separate version of truth about the data itself.
Our old friend Marty Ellingsworth wrote a very smart piece recently that takes my old premise further, arguing that many spreadsheet jockeys are so tied to their individual expertise that they resist adopting cutting-edge analytical tools that can make organizations even smarter. He writes:
"The fact is, much of the underlying actuarial modeling that still drives the industry today is built on the good ol’ Excel spreadsheet. That can make integrating new high-powered analytics a bit like installing Formula One parts on an Oldsmobile.... The most tenured and credentialed practitioners work harder at becoming experts in spreadsheets than they do adopting cutting-edge tools that can handle new data.
"Shocking as that observation may be for those who’ve been following the evolution of the InsurTech space during the past several years, it makes sense. Excel has become the primary means of communication between executives on matters of importance over the course of four decades—which is the entirety of most executives’ careers. Suddenly upending that workflow is not something that’s going to happen overnight, but the industry is going to need to evolve before the cutting edge can go truly mainstream."
Marty also notes the need for syncing up all the data in spreadsheets, especially in these turbulent times. For instance, many spreadsheets related to auto claims contain historical assumptions about the cost of replacement parts and about depreciation, even though costs for parts have been soaring and cars have been appreciating in value when they're driven off a dealer lot, not depreciating.
Now, I've always loved the origin story of the electronic spreadsheet -- a student at Harvard Business School in the late 1970s got tired of having to recalculate by hand all the values for a paper spreadsheet once he altered even one assumption even slightly. Talk about necessity being the mother of invention. That student -- Dan Bricklin, a lovely fellow who deserves to be famous even outside of computer circles -- happened to have a degree in computer science from MIT and to be acquainted with the then-new Apple II. He and his friend Bob Frankston wrote the code for a spreadsheet, and the rest is history.
But 40 years is a long time even for a seminal software tool. It's time to rethink our reliance on the spreadsheet and handle data better while clearing the way for even more advanced tools.
Cheers,
Paul