Industry segments are distinctive subsets of the larger industry along lines such as geography, product or service type, channel, customer or consumer needs, and so on. Mapping industry segments is rarely straightforward; it takes work, reflection, and, often, the willingness to explore beyond the current or obvious segments to segments that do not currently exist. In many cases, the accepted, traditional industry maps are imperfect. Like the old maps of a flat world that showed edges you could sail off, industry maps have limitations; only by exploring the edges of those maps can you see things differently.

As we mentioned before, if your customers don't all fall in the same operating segment — if they don't all want the same things, in the same order of priority — you have two basic choices. Your first choice is to focus without apology on one operating segment, to build a single service model around one segment's needs and keep your finger on its pulse. That's what Walmart and Southwest do. If a customer outside these companies' core constituency wants to do business with them, Walmart and Southwest will certainly take the money. But the companies won't contort their service models to also meet the needs of these secondary customers. Your second choice is to build different service models for the operating segments you uncover. Think emergency rooms and outpatient clinics within a single hospital.

We have come to describe the way most companies segment their customers as marketing segments — the buckets used to identify and communicate with different kinds of customers. We call the classifying of customers by service priorities operating segments. The vertical axis created in an attribute map can be thought of as an operating segment. The goal of the diagnostic stage is to reveal all the different operating segments — that is, all the different orderings of attributes that you are currently serving.

As we mentioned before, if your customers don't all fall in the same operating segment — if they don't all want the same things, in the same order of priority — you have two basic choices. Your first choice is to focus without apology on one operating segment, to build a single service model around one segment's needs and keep your finger on its pulse. That's what Walmart and Southwest do. If a customer outside these companies' core constituency wants to do business with them, Walmart and Southwest will certainly take the money. But the companies won't contort their service models to also meet the needs of these secondary customers. Your second choice is to build different service models for the operating segments you uncover. Think emergency rooms and outpatient clinics within a single hospital.

UNCOMMON TAKEAWAYS

• To achieve service excellence, you must underperform in strategic ways. This means delivering on the service dimensions your customers value most, and then making it possible — profitable and sustainable —by performing poorly on the dimensions they value least. In other words, you must be bad in the service of good.

• The primary obstacle to service excellence is not the ambition to be great, but the stomach to be bad. This is an emotional obstacle.

• It's difficult to compete without understanding your customers' needs and how well your competitors are meeting those needs. Fortunately, customers are typically very willing to give you that information. And it's cheap and easy to ask them for it.

• There is an important distinction between marketing and operating segments. Marketing segments tell us how to identify and communicate with different kinds of customers. Operating segments tell us how to serve customers differently.There is rarely a one-to-one mapping between these segments.

• There are two key ways to improve service: (1) meet your customers' existing needs more effectively, or (2) convince your customers that they need something you already do well.

• There is a difference between financial models and service models. Service companies need to be "bi-lingual" to excel.