This article originally appeared on Forbes.com.

In many consumer markets, companies that once relied on developing new product features and improving customer service increasingly see competitive advantage rooted in the entire experience that’s wrapped around the product.

For a mortgage, many customers care not just about the interest rate and amortization, but also the ease and speed of the approval process, the quality of the online payment system, and the bank’s service resolving problems that arise over the life of the mortgage. For a laptop computer, some customers might consider ongoing technical support as important as the initial hardware and software.

This shift in the sources of value pushes managers into largely uncharted territory. While product development has well-worn practices, many firms have not even started to identify their most important experiences, much less manage them. We are seeing more companies adopt a new key unit of management: the episode, which can encompass a variety of shopping, usage or service activities.

When customers have a task to complete or a need to fulfill through the company, that’s an episode. It has a clear start and end, marked by the customer completing what he or she set out to do. Episodes range from a single interaction (such as paying a bill online) or an intricate series of interactions spanning weeks (getting fixed broadband service moved to a new home).

While short episodes individually have little impact unless something goes wrong, they tend to be more frequent, so they should be easy for customers to transact, with no pebble in the shoe to cause irritation. Longer, more complex episodes, though less frequent, usually contain at least one high-stakes “moment of truth” and thus have a large effect on a customer’s attitude about the company.

Consider how an episode perspective changes cost-reduction initiatives at a telecommunications carrier. If it targets cost reduction from its technical support call-center operation, rather than from the entire tech support episode, it could end up with shorter calls or cheaper or fewer agents. That likely would lead to bad tickets of work, pushing more work into the back office and the field, with a higher number of expensive “truck rolls.” A better response might be to add call-center resources, in order to reduce the number of truck rolls.

We are not proposing that companies completely replace their management of individual functions with episode management. Rather, the two should work side by side. To avoid duplicative work or a new layer of bureaucracy, creating a successful episode management scheme hinges on making thoughtful choices in four areas:

1. Define episodes at the right level and focus on those that matter most. The customer’s high-level needs can be broken down into variants of episodes. For example, banks can usefully separate the overall need of “make a payment” into “pay another person,” “pay a bill” or “make an international transfer”. Each is defined by what the customer wants to do, with a clear start and end, and each consists of a number of underlying processes to redesign and improve. Emotional impact, frequency and economics should be considered when selecting key episodes.

2. Charge Agile teams with continuous improvement of episodes. Many companies have experimented with Agile methods for software and digital projects, but Agile also works well to manage an entire episode as long as it exists, with responsibility for the quality and efficiency of the episode. Up to 10 members drawn from each relevant function gather in co-located, cross-functional, self-organizing units to improve the episode in two-week sprints. Besides their Agile work, members remain part of their function as well.

3. Install clear metrics. While individual metrics will vary depending on the episode, most fall into one of three categories. Episode frequency, or the rate per customer, is particularly important for negative events, such as a credit card being declined at the point of sale. It might also be relevant for some positive episodes, such as redeeming reward points.

   Episode quality could include such metrics as speed to completion, resolution of a problem at the first contact, and Net Promoter Score (a key metric of customer loyalty) for that episode relative to competitors.

   Episode economics could include the all-in cost to complete the episode and the change in customer lifetime value. The benefits of a sharply improved fraud-management system, for instance, include avoiding not only the fraud, but also the lower spending from customers due to repeated disruptions. Measuring cycle time also comes into play for many episodes, because a fast cycle time shows that no errors occurred that would require costly rework.

4. Assign strict accountability for each episode. The shift in perspective does require governance that transcends current boundaries of department or function. Governance might rest with a current executive, or it might require a new role. One telecom carrier created a new role to oversee the episode of “installation of a new landline,” because of the technical complexity and numerous departments involved. Each of the departments, though, still is held accountable for the success of this episode.

When the experience matters as much as the core product, companies will need to work harder to earn their customers’ loyalty and advocacy. They can do that by delivering great episodes, reducing costs in the bargain.

Gerard du Toit, Rob Markey, Jeff Melton and Frédéric Debruyne are partners with Bain & Company’s Customer Strategy & Marketing practice.