A Human-Scale Service Vignette
Each spring, my family plans and invariably abandons several home projects, usually because we don’t have the required specialized tools. This year, however, we decided would be different. There were only two options available to us: either buy (and store) each tool, or borrow (and return) as needed from someone we know. Buying each tool wouldn’t be cost-effective with our plans, so I reached out to my neighbors for their input. We put our heads together and came up with an elegant but simple solution: a spreadsheet of tools our households could share with each other (also known as a Library of Things).
We coordinated informally through text, divvied up tasks, and iterated a shared spreadsheet. In less than a week, we compiled a list of tools that virtually tripled each of our own personal sheds, without spending a single cent. And this thing we built is serving us well. But why does it work?
The Anatomy of Failure
Rewind to the early 1980s. The world was transitioning from a manufacturing to a service economy (based on GDP), and a team of three researchers set out to measure how customers thought about service quality. You could measure whether a widget met a spec, but not the quality of a visit to a bank. Their goal was to help service-based businesses improve waning service quality. They conducted focus groups across five industries and found that customers kept describing gaps between expectations and experiences, not outcomes. In their earlier work, these researchers identified five gaps:
- The Knowledge Gap: the difference between customer expectations and what managers think they expect;
- The Policy Gap: the difference between management’s understanding of customer needs and how they translate that into actual policies and standards;
- The Delivery Gap: the difference between documented specifications of an experience and the actual results of the experience;
- The Communication Gap: the difference between what was communicated to the customer and the delivery of the experience;
- The (Cumulative) Customer Gap: the difference between customer expectations and their perceptions.
These gaps became the basis of the SERVQUAL model of service quality. The five industries studied (retail banking, credit cards, securities brokerage, product repair, and long-distance telephone) were all dominated by large national firms. By focusing on these organizations, the researchers inadvertently studied the pathology of scale. While seeking to measure the factors contributing to service quality, they created a diagnostic tool that could roughly locate the threshold of failure. In other words, these gaps can actually help us answer the question raised in the previous article, “how big is too big?” Let’s dive into the gaps, one by one.
At human-scale, the person managing a service knows what their customers expect through direct and intimate contact. In large-scale organizations, layers of management between the customer and decision-maker attenuate the customer’s signals of need and expectation. The mechanism is that each layer necessarily summarizes, and summary is lossy. As the hierarchy grows, so does the gap.
In 2017, United Airlines gate agents at O’Hare selected a passenger for involuntary removal from a fully boarded flight. Dr. David Dao, a physician who needed to see patients the next morning, refused. Security officers dragged him from his seat, breaking his nose and knocking out two teeth, while other passengers recorded on their phones. When the footage went viral, CEO Oscar Munoz initially called Dao “disruptive and belligerent” in an email to employees, because that was the distorted version of events that survived the chain of command. The person with authority to override the policy was insulated from the moment by layers of organizational distance. The gate agents who could see what was happening had no authority to stop it. That is the knowledge gap.
At human-scale, this gap likely wouldn’t exist. The person serving you set the terms, and if those terms fail, they feel it directly. In large-scale organizations, standardized policies are written for the average case and fail for the outlier, to institute order and predictability. The person who wrote the policy will never meet the person for whom the policy fails. Reducing the policy gap would require granularity that scales with customer diversity, and that complexity would quickly gridlock the organization.
For years, Bank of America charged overdraft fees that brought in roughly $4 billion a year. The fees landed hardest on the customers with the least money in their accounts. From a revenue standpoint, the structure worked. But the people who designed it never had to sit across from someone paying a $35 fee on a $6 coffee. In 2022, after sustained public pressure, the bank overhauled its fee structure.
At human-scale, the person who delivers the service is the same person whose reputation depends on the result. Here, quality is not a management objective; it is a condition of survival. In large-scale organizations, service delivery is compartmentalized so that it happens in chunks across teams and departments. The mechanism is that compartmentalization severs the connection between the person doing the work and the outcome of the work.
Consider a patient being discharged from a hospital. The physician writes the discharge order. The nurse explains the medication schedule. The social worker arranges follow-up care. The insurance advisor handles the authorization. Four people, four departments, four sets of performance metrics. If the patient is readmitted within thirty days, that outcome is tracked by a fifth department. Nobody owns the full experience, so nobody owns the failure.
By now, I think you see the pattern: as feedback loops lose their integrity due to increasing scale, service providers sacrifice quality. Let’s keep going.
At human-scale, the one making the promise is either the same person or closely connected to the one delivering the service experience. In large-scale organizations, the people making promises are separate from the people keeping them. The mechanism is specialization: growth requires it, specialization creates departments, departments develop their own measures of success, and their incentives diverge.
Fatima’s Grill in Downey, California was a small Lebanese-Mexican fusion restaurant. In 2020, a TikTok video of one of its dishes went viral. Lines grew to three hours. The owner considered closing. The viral video was, in effect, a promise the restaurant never made and couldn’t keep. The communication gap opened not because marketing and operations were in separate departments, but because the audience for the promise (millions of TikTok viewers) exceeded the scale at which delivery was possible. The gap formed instantly, from outside. Even a small business can cross the communication gap threshold when the promise outpaces the capacity to deliver.
At human-scale, the previous gaps are small enough that perceived service quality is likely to exceed expectations. In large-scale organizations, problems grow faster than solutions. The four preceding gaps generate failures continuously; they are persistent structural conditions, not isolated incidents. Remediation, by contrast, is episodic, resourced from the same constrained hierarchy, and addressed sequentially. The result is a rate problem: failures compound while corrections accumulate. Large-scale organizations cannot fix themselves faster than they generate new problems.
During a recent family vacation to Mexico, we went snorkeling on the Mesoamerican reef. Afterwards, we overheard a customer suggesting the owner advertise the snorkeling service on TripAdvisor, then walking his own suggestion back after realizing (out loud) that the popularity would probably diminish the quality of the service. He understood intuitively what the gaps describe structurally: the conditions for good service have an upper bound, and crossing it does not mean you need better management. It means you have outgrown the conditions that made quality possible without management at all.
The Toolkit and Its Blind Spot
To sum it up, large-scale organizations created the very gaps the SERVQUAL model discovered. Each gap has a scale threshold beyond which it becomes unclosable, not because of management failure, but because of organizational geometry. Large-scale organizations can only ever manage these gaps. They cannot close them.
How do they manage them? Through VoC programs, UX research, journey mapping, design thinking workshops, service blueprinting sessions, and other tools in the service design toolkit. These are sophisticated, well-intentioned tools. They are also prosthetics for damaged feedback loops. Consider what journey mapping actually does to the knowledge gap: it imports signals of customer expectations, but those signals arrive abstracted, averaged, and attenuated. The mapping enters the organization as a project with a timeline, producing a deliverable that re-enters the same hierarchy that attenuated the signal in the first place. The tool reimports what scale expelled, but through the same structure that expelled it. This is not to say the tools are useless. It is to say that the context in which they operate, a large-scale organization built on standardization, compartmentalization, and subordination, structurally constrains them.
Where This Leads
Ultimately, the right question is not how to design better service delivery systems within organizations of any size. It is: what size of organization enables good service to emerge without requiring design at all? In our neighborhood shared tool library, there is no knowledge gap because we are neighbors creating this out of a shared need. No policy gap because we are operating on norms with shared stakes. No delivery gap because the person lending a tool owns the tool and cares whether it comes back without damage. No communication gap because everyone promising is also delivering. And therefore no cumulative gap to close.