[Tribune] Why the fall of Tesco does not call into question the data-driven economy
Since the announcement of the fall of Tesco and the resignation of its chairman Sir Richard Broadbent, analyzes of the future of data-driven business have multiplied. One of them, published by the Harvard Business Review, particularly caught our attention with its alarmism.
Recently, the excellent Harvard Business Review site published a controversial article on the fall of Tesco and what it would say about the future of big data for businesses. Let us recall for the latecomers that this English institution of mass distribution that is Tesco was until now a benchmark in terms of data driven business and customer knowledge. A pioneer in loyalty programs, it launched Tesco’s ClubCard in 1995 with the aim of collecting as much information as possible on the behavior of its buyers. A revolution for the retail industry. She also plunged into the digital deep end when the competition couldn’t swim yet. Two strategic choices oriented towards “customer knowledge” which have paid off: to use the very terms of the article, “Tesco has become the market leader because it has been able to meet the demands of its buyers” .
But for the author, it is also this strategy which would have caused the current situation: a market value divided by two in 11 years.
The words of Michael Schrage are harsh: “Britain’s biggest supermarket chain has not only seen its fortunes eroded, we read, but also its reputation for competitiveness, creativity and integrity collapsed. ” And the journalist points the finger at the culprit: “Tesco’s decline is an unambiguous warning that not all loyalty program data and all analytical capabilities in the world can substitute for the competitive advantage of lower prices and a simpler customer experience. Insights, loyalty programs and targeted promotions may not be unnecessary, but they are clearly less important in the retail world. ” An analysis which finally inspires the journalist with this beautiful concise formula: “In less than a decade, the determining factor of Tesco’s success has transformed into an analytical albatross. Knowledge made him pass from power to impotence. ”
The problem with this reasoning, in our opinion, is that it confuses correlation and causation. A classic of analytical bias that bears the sweet name of endogeneity. The journalist broadly establishes a cause and effect link between “data driven” (variable x) and the fact that Tesco’s market value finally collapsed (result y). But it pretends to ignore the existence of an unobserved heterogeneity, called variable z, which could influence both the variable x and the result y. To put it another way, it is not because a “data-driven company” collapses that it is necessarily the fault of the “data-driven” variable x. There is a variable z to take into account in the equation: “company”. In the case of Tesco this variable was undoubtedly decisive.
No matter where you take the problem, the same conclusion emerges: the flagship of mass distribution has not felt the tide. As rightly pointed out this other article from the HBR, the supertanker Tesco was surprised by the sudden segmentation of a hitherto very homogeneous English market. The threat posed by Aldi and the other discounters was underestimated due to a weak start on English soil. While this low-cost model was successful in Germany, British consumers at first began to be reluctant, supporting the strategy of the dominant model represented by Tesco. But the crisis is there. And in times of crisis, nothing is worse than certainties. Due to its undivided dominance in the market, Tesco did not see (or want to see) that people’s consumption habits were changing.
As a result, as the market and consumers have segmented between the low end and the high end, Tesco has remained stuck between the two poles, in a mid-range without appeal or future. It also failed to see that the fragmentation of purchasing behavior was the de facto end of its sacrosanct loyalty cards. But by accusing data of being responsible for this immobility, Michael Schrage is wrong target: it is never the fault of data; it’s always the fault of what the company chooses to look at in its data. In other words, to come back to our story of endogeneity, it is Tesco’s hubris (variable z) that has influenced both the analysis of data (variable x) and the collapse of the value of the company (result z).
Crisis of growth, crisis of confidence
Zynga, the company behind the Farmville saga and other successful Facebook games, had a similar set of setbacks a few years ago (read this excellent article from Arstechnica about it). The first success story and the first industrial crash of the big data era, the company did not know, or wanted, to look in the right place. At the time, Zynga piled on the heads of divisions, project managers, vice-presidents… So many leading micro-entities which paradoxically struggled to look in the same direction.
Tucker Dewitt, Service Quality Manager, sums up in ArsTechnica the vision problem that blinded the company at the time: “We have always operated with a long-term strategy. The problem was that at the time this strategy changed every three to six weeks. ” In other words, it had become a tactic like any other.
One of the developers, Slade Villena, ended up describing on Reddit the atmosphere that reigned behind the walls of the company: “The bigger we got, the more we had to listen to analysts rather than our players. No one realized that we were losing more than we were gaining. (…) One day, we pushed some code that permanently trapped 10,000 players in a level. 10,000! I was told, ‘Don’t worry about them.’ Management prefers to attract new players, keep them for 3 months by taking $ 5 to $ 10, then let them go. “
The data that reflected the exodus of users and the free fall of loyalty were nevertheless there, lost in the midst of other data. But rather than looking in the thousands of weekly reports which questioned Zynga’s overall strategy, management preferred to focus on KPIs that reinforced its tactics of immediate profit.
The result: a massive user leak and a share that has lost 71% of its value in 2 years.
Like Zynga, and like many others before it, Tesco’s real problem lies in these few words: the pioneer had become the number 1. During the 90s, Tesco had made its three promising trends its own. : small local units, house brands and digital sales. It was this same acuteness, this same ability to sense the market, that later prompted her to strengthen her analytical intelligence. Thanks to her loyalty card, she began to collect an insane amount of data that she patiently segmented, analyzed, in order to know and even anticipate the purchasing habits of her customers.
A decision crowned with success, since Tesco ended up swallowing more than 30% of the large-scale distribution market across the Channel. Except that by becoming a leader, she lost her values. Seeing its business model confirmed by the sector and analysts, it visibly sank into a form of self-satisfaction. No mock trial, however.
As this article from 2011 demonstrates, Tesco has indeed reacted to the war unleashed by competition: by halving the points of purchase obtained thanks to the card, it has undertaken to carry over the savings made on prices in stores. The company’s spokesperson made no secret of the reason for the decision: “Our customers have told us that what really matters to them right now is the price.” As the graph opposite illustrates, Tesco then deliberately chose to ignore that in 2009 it had stemmed the fall in its market share by rewarding card holders more. Unfortunately, this new pricing policy to the detriment of loyalty programs has not had the desired effect. Was Tesco wrong to listen to its customers? Of course not. She just didn’t complete her diagnosis.
Data driven, but in which direction?
HBR’s analysis hits just one point: data is not the alpha and omega of the economy. Too many data driven companies forget it: data is not there to set the course, it is just there to tell where the wind is blowing. Like Zynga before it, Tesco was undoubtedly overwhelmed by reports of all kinds, drowned in figures and segmentations, blinded itself with performance dashboards. It is quite possible that this data-bulimia, coupled with the critical mass reached by the company, prevented it from seeing in time the change in behavior among English consumers, the growing mistrust of loyalty programs and the impact of the crisis on portfolios. By dint of looking at everything, measuring everything, monitoring everything, she did not see the essential: the wind had changed direction. But once again, this information was there, available, analyzable, exploitable like the others. Tesco could therefore have integrated it into the other KPIs it monitored and used it to support its strategic orientations. More than data, it is therefore the selection of data and the attitude towards it that is at stake: are we looking to be comforted, reassured or worried by the data that we have chosen to analyze?
In turn, a data driven company is not a company that monitors all of its data, or that only looks at the data that corresponds to its initial strategy, or even that hides behind activity reports to make strategic decisions that are just tactical adjustments.
A data driven company is a company that only worries about the data that matters, that seeks in the data what will feed a strategic orientation, which wants to make decisions for tomorrow and not be reassured about the decisions taken yesterday.
As such, this Telegraph article highlights what appears to us to be the crux of the problem: Tesco had become a caricature of a data driven company. His loyalty program was costing him £ 500 million a year, arguably more than he earned. For analysts, the company therefore had an economic reason and a unique opportunity to radically change its image: by removing its loyalty program which made its fortune to better lower prices, it would have struck a big blow and demonstrated to its consumers that she had lost none of her audacity and sharpness. Because the data did not tell Tesco that it had to adjust its strategy, it told it that it had to change its business model. In fact, they were actually asking him this absolutely dizzying question: what should a data-driven company do that reads its data that it must collect less data? Hence this paradoxical conclusion: in reality Tesco did not fail because it was a data driven company; she failed because she was not enough.