At the first Starbucks I pass on my way to work, the line often stretches out the door. Yet if I hold my caffeine addiction at bay and drive a few minutes further, I can order my desired beverage with little or no wait at all. The cafes are almost identical in size and ambience; both menus are the same. Yet some customers routinely choose to wait in line — often for longer than it would take to drive to the other outlet.

In all likelihood, those two Starbucks aren't identical after all. While the company makes an effort to reduce variation and deliver the same customer experience, its branches perform differently — with different financial results.

This is true across all service industries, including healthcare: When organizations grow and become geographically dispersed, they begin to see greater variability in the performance of their customer-facing outlets. How much variability to tolerate, and how to improve the performance of all outlets, becomes a central leadership challenge.

And it's a particular challenge today for healthcare executives, many of whom have been on recent acquisition sprees and are struggling with post-merger integration issues. When athenahealth examined the financial data of its clients, it discovered clear diseconomies of scale — and a wide range of results among the individual practices in large healthcare systems.

The roots of variation

Successfully managing variation begins with recognizing why it happens — and when it can be avoided. When I was an executive at L. Brands in the 1990s, I met with many underperforming retail store managers and their superiors. They would all insist that their branch (or district or region) was struggling because of external factors: The store was tucked away in a bad location of a struggling shopping mall, or parking was an issue for customers.

Healthcare executives hear similar excuses for variation from practice leaders — the patient panel is different, the payer mix is tough, and so on. Yet I have found that, even in complex business environments, there are ways to reduce variation.

In one extreme, top-down, highly-engineered systems attempt to standardize as much as possible and put in place a management process that defines and demands compliance: Think of the white-gloved McDonald's inspector who arrives at each franchise with a long checklist.

At the other extreme, the “affiliate" model takes a laissez faire approach. Crossfit is an example. Its CEO, Greg Glassman, requires affiliates pay a yearly fee to use the brand, and that's about it; there are very few requirements or specifications for how they should run the business. If affiliates are doing something wrong, Glassman believes the market will let them know, and put them out of business if they don't respond.

Across the service industry, this lighter-handed approach seems to be increasingly adopted — especially if you include sharing-economy businesses such as Uber and Airbnb. (After all, what else is Airbnb if not a hospitality company that allows all its properties to operate as they please?)

I'm a believer in empowering front-line workers to tailor their businesses to best serve their customers, so I have a natural affinity to the affiliate model, and believe it will eventually become the dominant model for franchise businesses. But a management structure that loose doesn't transfer directly to healthcare. The consequences of business failure are too high when we are talking about health as well as economic outcomes.

So what are healthcare executives to do?

The middle ground — and the bottom up

My advice is to follow a middle path to managing variation — such as the one we used successfully at L. Brands. Called peer-based improvement, or PBI, it's a bottom-up performance management methodology in which corporate managers group together locations within systems that face similar external environments, then compare the stores within each peer group along several key performance metrics.

Because the approach only compares apples to apples, it compels managers to focus their attention on what they could be doing better internally — as opposed to blaming their woes on external factors. If done well, PBI can improve underperforming locations within each peer group, promote relevant group learning, and identify additional opportunities for high performers.

Successful implementation of PBI involves several steps.

• First, managers develop credible peer groups by determining which drivers most affect performance — location type, competitive intensity, size of the outlet, and so on.
• Next, managers identify the key performance metrics they will use to compare peers by prioritizing them based on the size of financial impact. That way, best practices from high performers in peer groups can be shared widely, along with standard business-improvement coaching.
• The final step is the most important: building the organizational capability to sustain improvements over a long period of time. By giving frontline workers the tools, resources, and latitude to make meaningful change, leaders can create the structure for a continuously learning organization.

Technology can aid in this effort by collecting, and making accessible, both performance metrics and remediation tactics. At L. Brands, we developed an integrated point-of-sale system that provided local and comparative peer group data at the store level. That way, all Victoria Secret locations, for instance, could benchmark their performance and share data and tactics on a timely basis.

But such technology isn't essential. What matters is that information sharing and learning occurs. Food company Danone Group gave up on its attempts to install an IT system to support PBI after several false starts. Instead, they created events — termed “marketplaces" — where operating managers share what works with their peers. The goal is to identify what they call “nice stories" about successful local innovation that can be shared across geographies.

Making it work in healthcare

In healthcare, some organizations are already putting versions of this system into practice. Privia Health, a fast-growing, venture-backed physician network, develops precise road maps to help recently acquired practices transform their organizations to be able to succeed under risk-based contracts.

At Privia, performance consultants, many of whom have experience running practices themselves, review detailed performance reports with doctors — including unblinded data about how those physicians are performing versus their peers.

In addition, peer groups meet in person periodically to discuss what's working and what's not. They develop monthly action items tied to concrete metrics, such as increasing portal adoption or scheduling visits for high-risk patients.

Because healthcare is administered by and to human beings, we can never stamp out variation entirely. Peer-based performance improvement is a methodology that accounts for unavoidable variation while removing excuses for underperformance.

If done right, this method improvement drives results. But that's not the only advantage. A data-driven approach to managing growth can lay the groundwork for collaboration, peer-to-peer sharing, and continuous improvement.

Len Schlesinger is Baker Foundation Professor at Harvard Business School and the former vice chairman and chief operating officer of L. Brands.