Big Decisions and Small: When Analytical Pioneers Go Awry
The corporate news of late features a couple of analytical pioneers, Caesars Entertainment and Tesco PLC, that have fallen on hard times. Caesars did a complicated restructuring to reduce debt, and some of its casino properties filed for Chapter 11. Tesco, the UK-based retailer, has been suffering for years because of overexpansion, the wrong mix of store types for the economy, and ill-advised ventures into banking and technology.
Both companies were analytical pioneers, using customer data to understand buying behaviors intimately and target promotions. Both had data-driven leaders—Gary Loveman, the former Harvard Business School professor who still leads Caesars, and Sir Terry Leahy (I wrote about him earlier), who was Tesco CEO in 2011, largely before things started to fall apart. Why didn’t their analytical decision-making save them?
My friend and “fellow fellow” at the MIT Center for Digital Business, Michael Schrage, wrote a (perhaps overly) dramatic blog post suggesting that Tesco’s analytics were at the heart of its problem:
How damning; how daunting; how disturbing for any and every serious data-driven enterprise and marketer. If true, Tesco’s decline present a clear and unambiguous warning that even rich and data-rich loyalty programs and analytics capabilities can’t stave off the competitive advantage of slightly lower prices and a simpler shopping experience. Better insights, loyalty and promotion may not be worthless, but they are demonstrably worth less in this retail environment.
I’m surprised that Mr. Schrage hasn’t written about Caesars yet in the same tone.
But analytics aren’t really the problem at Caesars and Tesco. It’s still useful to know your customers, reward the most loyal ones, and target promotions based on what customers actually want. Caesars and Tesco still do a good job of getting those small decisions right. It’s the big decisions that don’t involve data that have caused them problems.
Big decisions are typically one-off calls that don’t involve data because they involve choices that companies haven’t made before. They’re decisions about whether to develop completely new products, whether to enter completely new markets, or whether to take on a completely new financial structure. A.G. Lafley of Procter & Gamble calls such decisions “big swings,” and he says you only get a few of them during your tenure as CEO.
Both Caesars and Tesco have had problems with their big swings over the last several years. Caesars swung and missed on two big ones—the decision to take on about $30 billion in debt in a leveraged buyout, and the decision not to buy a property in Macau. When the economy turned down the company was unable to pay down the debt fast enough, and the Chinese gambling center turned out to be the growth engine for the entire industry—except for Caesars.
Tesco decided wrongly on several big swings as well—entering the U.S. market with its Fresh & Easy chain (which, like Caesars, declared bankruptcy), building hypermarkets outside of city centers in the UK, and jumping into non-grocery businesses that taxed management attention.
None of these decisions had been taken in the past by these companies, so no data was available on how they would turn out. I am sure they all involved some analysis, but it is a bit difficult for anyone to get good data about the future. At Caesars, in particular, I know there was lots of analysis about the Macau decision. Mr. Loveman and his colleagues tried to figure out a way to justify the $900 million license for Macau, but their analyses yielded only about $650M in potential value. They didn’t have any way to anticipate the fantastic growth in the Macau market. Mr. Loveman now estimates that the license was probably worth $2 billion.
Both the leveraged buyout at Caesars and the big expansion at Tesco were probably a bit risky from the start. Some LBOs have worked out well, but the track record on the big ones is not that great. Mr. Loveman once told me that his analysis team had done the numbers on the basis of the worst possible economic conditions they could imagine, but then reality turned out to be much worse. That doesn’t seem unreasonable, but borrowing $30B just before a downturn doesn’t seem like the best decision ever. Nor does expanding into a highly competitive market like the U.S. with a relatively new grocery concept—small but expensive local markets in the wide-open Western U.S.
One could argue that these “big swing” decisions are as much a matter of luck as anything else. Perhaps some leaders have a better ability to anticipate the future than others, but I have never personally consulted fortune tellers, and don’t really believe in that ability. Someone who has a history of making big decisions well might stumble on the next one.
Ideally, leaders would be able to inspire their organizations to use analytics on the small decisions, and know when to depart from the analytics on the big decisions. But if I invested in individual companies (which I don’t, because I can’t anticipate their specific futures) I would back leaders who I know will make the small decisions well. Making small decisions well is based on science; making big decisions well is based on gambling.
Thomas H. Davenport is a Distinguished Professor at Babson College, a Research Fellow at the Center for Digital Business, Director of Research at the International Institute for Analytics, and a Senior Advisor to Deloitte Analytics.
Monument Strategy Group
9 年Tom, very interesting piece. Adding to prior comments, my experience working with Fortune 500 companies (as well as emerging companies) is that analytics frequently is bifurcated into the strategic and the tactical, and often is not viewed holistically. Analytics is valuable in both spheres, obviously, but where things often go awry is that there is not enough consideration given to how the analytically-driven insights will actually be applied in practice. I goes without saying, but this is a crucial piece of the analytics value chain. Our advice is to include up front in the planning stage those people and departments that will have the task of implementing the new applications and practices. That can help you get more value out of the analytics and help you avoid fatal breakdowns. Secondly, in many cases as well there isn't enough regard for future uncertainties. Markets are changing much more rapidly and historical relationships may simply not reflect prevailing or emerging conditions. Managers don't often ask What should we do if future conditions change? What is the best that can happen? What is the worst that can happen? This kind of contingency planning (looking at minimax, maximin, decision frameworks) adds time and expense, but can save a company from much larger costs, catastrophic failures, and other unhappy consequences.
Given the problems with big bets, it seems the challenge is to figure out ways if you can to make them smaller. Could Tesco have made a small investment to test the market in Southern California? Could Caesars have made a joint venture with a small stake in a Macau opportunity with an option to grow it?
Executive Director @ EverythingALS | Senior Leader of Patient Advocacy and Patient Engagement | Consortium & Alliance Management
9 年After a night of poker with some friends, I have to say, the gambling analogy is particularly resonant as I read this piece. Yes, luck happens. But gambling can be just as much about about situational awareness, and competitor intelligence to mitigate risks related to timing - the greatest of market unknowns. Macau was a market made by competitors' moves in a compressed timeframe, coupled with a market downturn. While tactical analytics might not have been as important in defining their fate, perhaps the issue was not enough investment in other inputs to the strategic dashboard, like watching what their competitors were doing economic indicators, and whether they were benchmarking their positions to maintain alignment in case of the need for re-leveraging? After all, in poker, you're not just playing your hand against the others Ina single round of betting on known odds (tactical data). You're also managing your cash against others' cash management patterns and leverage positions across hands. So if you're gambling, play poker - not roulette. As for Tesco, based on what's presented, it seems a case of the Story being lost despite the attention to getting certain facts right. You can tell a story with the same facts many ways. It's up to the leadership to be mindful of whether the facts are leading to an internally consistent (coherent and linear) AND externally plausible (attractive and aligned with market behavior/ expectations, say of getting new customer segments and how executable the plans are) story. Thanks for yet another awesome article, your classes at Babson were among my favorites!
Data delivery should be a corporate mandate if I were setting expectations for contractual professional services. That' what Having a short or long term information/data strategy is key. Strategy and execution is where business and the human capital we invest aren't in the same room when setting a plan. Simpler to think small but scalable. Include your team players before the last qtr of the game ... Unfortunately execution often has many failures come when its accountability for agreeing to deliver priority program or projects and most often gets delusional when no program is in place. If your not sure how to start just ask. I have repeatedly heard several industries and even the competitive intelligence shows almost all of us want to believe this or that "data" is different. I want to always say tell me how? Standards and accuracy will drive productive change management and participation! One note: your employees need to start caring to want to understanding the why?
Leading authority on Digital Decisioning and delivering business impact from AI and machine learning
9 年Great points Tom. The difficulty in applying analytics to strategic decision - big swings as you put it - is real. I think that getting companies and executives to recognize that the best place to apply analytics is often at the operational level - day to day decision making - is critical. There's a sense that I have to make big, expensive decisions with data to show a return on analytics but the reality is that many operational decisions - millions or tens of millions of them - multiply up to big returns when analytically improved.