Amazon’s acquisition of Whole Foods last August was the corporate equivalent of mixing tap water with organic extra virgin olive oil. You’d be hard-pressed to find two companies with more different value propositions.
Even so, it was surprising to hear reports shortly after the marriage about Whole Foods customers, really angry customers, regularly encountering empty shelves at their favorite retailer. Then stories surfaced about Whole Foods employees crying over their new performance-driven working conditions imposed by Amazon.
“This is not a story where there is a good guy and a bad guy”
Harvard Business School professors Dennis Campbell and Tatiana Sandino took notice, suspecting a clash of corporate cultures was at work. Their forthcoming case study discusses the limits of trying to force one culture or management style on another organization. The case was co-written with James Barnett and Christine Snively.
A tale of two cultures
From the very start, Amazon made its name on being fast, cheap, and efficient—using data to drive its product mix and enforcing strict employee discipline to squeeze out cost savings to pass on to its customers.
Whole Foods, on the other hand, always prided itself on its personal touch, empowering individual stores—even individual employees—to make decisions about products that emphasize high quality, healthy, and local foods. That decentralization, however, caused enormous inefficiencies that drove up prices to the point where critics referred to the store as “Whole Paycheck.”
The acquisition was initially met positively by Wall Street, amid hopes that Amazon’s data-driven mindset might be just the thing to enable Whole Foods to scale up and add more stores while maintaining its employee-empowered culture.
That’s not quite what happened.
“A lot of it from our perspective was centered on a culture clash,” says Campbell. “Whole Foods has a very high-empowerment kind of culture, so these ‘draconian’ standards, telling people where to put things on the shelves and the loss of autonomy, employees were feeling angry from that.”
The new inventory system was actually something Whole Foods had started to implement before the Amazon deal, pressured by activist shareholders who had seen the grocer’s stock and sales margins slipping for two years.
“This is not a story where there is a good guy and a bad guy,” says Campbell. “It’s a story about what the limits are to scaling this high-empowerment model, and what are the limits to a model where it’s all about standardization and data.”
A model of empowerment
For decades after its founding in 1980, Whole Foods’ decentralized model had earned a cult following, driving rapid growth from its Austin, Texas, home across the country. Managers operated stores like autonomous fiefdoms, able to tailor offerings based on customer preferences for fresh, local produce. Employees—also known as team members—built relationships with customers to cater to their needs and came up with ideas, like a bike messenger service or a new bread recipe, that sometimes found their way to other stores.
“That model of empowerment brings with it a lot of wonderful, creative solutions and a great relationship with local suppliers—and customers are very into that,” says Sandino. “They were appealing to a wealthy customer with a curated selection of healthy products that people could trust.”
The company also earned loyalty from team members—it was named one of America’s best companies to work for by Fortune magazine a staggering 20 years in a row.
“It’s one of those cultures that works for some people, but not others. There is a lot more performance pressure and accountability”
In some ways, however, the company became a victim of its own success as a pioneer of organic sustainable foods. Peaking in 2013, it started losing market share to less pricey alternatives such as Walmart, which had entered the organic space. By 2017, Whole Foods had started closing stores.
Turning data into customer value
Amazon, by contrast, had always been about low costs and efficiency, pursuing a frugal focus and rigorous performance measurement in its warehouses and distributions centers. This sometimes resulted in reports of workers stressed out and exhausted by the demanding environment.
“It’s one of those cultures that works for some people, but not others,” says Campbell, “there is a lot more performance pressure and accountability.”
The model paid off for customers in low prices and fast shipping, and they became intensely loyal: According to one survey, 85 percent of Amazon Prime members visit the site, and 46 percent buy something at least once a week.
After Amazon’s acquisition, Whole Foods pushed forward with the inventory system first introduced by the activist investors, started to centralize decisions about product selection, and slashed prices by as much as 40 percent on some items.
Employees struggled, however. They were frustrated about having to do paperwork instead of helping customers, and stressed over new performance metrics with demerits if they failed to meet them. Last year, the company dropped from Fortune’s best companies to work for list for the first time in two decades.
What went wrong?
The question that Campbell and Sandino ask in their case is: Given the pressures Amazon was facing to turn around Whole Foods’ slide, should they have approached the acquisition differently?
While there are no easy answers, Campbell says that part of the issue is realizing the limits of standardization, even for a company that has perfected data-driven management.
“It’s not totally clear that data will be a perfect substitute for human judgment,” he says. “That might work in a digital platform, where you have tons of data on customer history you can use to drive a recommendation engine, but in a store environment, there is a lot of learning that takes place from employees interacting with customers that can be very localized and specific.”
That kind of tacit knowledge is not easily captured in data and performance metrics, adds Sandino. “Amazon has been an expert on delivering non-perishable foods, but that is different than learning how to prepare certain foods, or knowing how customers want their fish or a cut of meat, which may vary in the moment.”
She suggests Amazon may have been better off pursuing a management concept known as structured empowerment, where a company standardizes operations but allows flexibility for employees to make their own choices in key areas where having high-touch contact with customers matters.
In addition, she says, Amazon might have changed its performance measures to focus more on results rather than processes, holding employees accountable for goals, but giving them more leeway on how they achieve them.
“Instead of this assumption that data should take over everything, there is a huge opportunity here for data to inform and complement human judgment”
“They can make some tradeoffs and incorporate their own knowledge, rather than having to follow a recipe,” she says. That, in turn, could give them more incentive to use the data Amazon is serving up to further drive results.
Is the best yet to come?
Campbell stresses that the ideas in their case are speculative, based on second-hand reports in the press. They can’t be sure how much Amazon tried to integrate its culture with that of Whole Foods, or what its ultimate goal is for the acquisition. There is always the possibility the company is just acquiring the stores for their locations, and plans to phase out the Whole Foods brand.
Assuming that Amazon wants Whole Foods to succeed, however, it might do well to consider the benefits of the grocery chain’s empowerment model, and what elements of it to keep, before throwing out the proverbial olive oil with the tap water.
“Instead of this assumption that data should take over everything,” Campbell says, “there is a huge opportunity here for data to inform and complement human judgment.”
This article was written by Michael Blanding from Harvard Business School Working Knowledge and was legally licensed through the NewsCred publisher network. Please direct all licensing questions to firstname.lastname@example.org.