DevOps process: Customer feedback

In software projects, developers often work on products for months or years, sometimes for multiple releases without validating whether the features they're building are actually helping users solve their problems, or whether the features are being used at all.

Customer feedback is part of a wider group of capabilities, including visibility of work in the value stream, working in small batches, and team experimentation, that together represent a lean approach to product management.

When these capabilities are applied together, they help predict the following:

  • Software delivery performance, which is measured in terms of delivery speed, stability, and availability.
  • Organizational performance, which is measured in terms of profitability, market share, and productivity.

DevOps Research and Assessment (DORA) research shows (PDF) that teams achieve higher performance when they work in organizations that utilize those capabilities and also do the following:

  • Collect customer satisfaction metrics regularly.
  • Seek out and attend to customer feedback on product and feature quality.
  • Use this feedback to help design products and features.

How to implement customer feedback

When you are developing a product, it's important to establish key metrics to gauge its success. These metrics help you understand whether you're solving a real problem, whether your solution is being adopted sufficiently quickly, and whether users continue to use it and recommend it to others. These metrics must be explicitly derived from how customers interact with the product. Using these metrics requires you to carefully gather and analyze customer feedback.

One set of metrics that's popular in consumer-facing SaaS products is AARRR: acquisition, activation, retention, referral, and revenue. (The order is sometimes different depending on who is presenting them.) These are sometimes known as pirate metrics because the acronym spells out a word that's stereotypically associated with how pirates talk.

The idea is to look at five key metrics and iterate your customer experience to improve on them:

  • Acquisition: The percentage of users that come to your site who create an account.
  • Activation: The percentage of acquired users that activate their account and use the service.
  • Retention: The percentage of activated users that return to the service.
  • Referral: The percentage of retained users who refer other users to the service.
  • Revenue: The percentage of referring users who actually pay money for the service.

If these metrics aren't suitable for your business or product, it's essential to discover some that are suitable, monitor those metrics regularly, and use them as a key driver of your product strategy.

The approach described in this article is not just for external-facing products. It applies equally to internal products and services that are built for other users in your organization. Building internal tools requires exactly the same approach: early and frequent engagement with real users to ensure what you're building actually solves their problem. Otherwise, you might end up building and using tools that nobody uses. This is discouraging for those who work on the tools and frustrating for those who have to use tools they don't like. It also represents significant costs for the company in building a poor technology match for its users.

A team should use the following pattern in order to maximize their chances of successfully solving customer problems:

  1. Gather customer feedback first, before defining any potential features.
  2. Validate that you're solving a real problem.
  3. Iterate on a solution that actually solves that problem (and nothing more).
  4. Ensure the solution results in a viable business (for example, the cost is less than the anticipated revenue).
  5. Track key metrics to gauge success (for example, AARRR).
  6. Iterate through the above to improve those metrics.

Success requires you to not only deploy and release software quickly, but to address customer needs better, smarter, and faster. This can be achieved by experimenting more thoroughly than your competition. Retail is one industry that has seen particular success with this (PDF). Techniques such as hypothesis-driven development, defining and measuring your customer acquisition funnel, and A/B testing allow you to perform user experiments. This can foster creativity, speed up innovation, and create organizational learning.

Increased engagement with customers and participation in product management processes contributes to stronger identification with your organization's goals and values (PDF). This in turn helps contribute to organizational performance.

Common pitfalls

Some common pitfalls to effectively using customer feedback include the following:

  • Failing to gather feedback. It's common for software development companies to not gather customer feedback at all.
  • Gathering feedback too late. Sometimes companies gather customer feedback so late in the software delivery lifecycle that they cannot act on it.
  • Not understanding the problem (or misinterpreting customer feedback). Companies can have unrealistic expectations of their customers, or might not understand what customers want from the product. The team might even explicitly ignore inconvenient but accurate customer feedback. This situation can arise when a development team hasn't adequately evaluated or managed the risk of delivering the solution at hand, or simply doesn't understand the problem to solve. A development team wants to keep the scope of their work to the minimum that's required to solve a problem. In some cases, if the problem is larger than what the team has designed for, the team might incorrectly dismiss the additional work as "scope creep." This results in an incomplete solution from the customer's perspective, and ultimately results in a failed product.
  • Failing to allow teams to act on feedback. Delivery teams must be empowered to make changes to the design or specification of the system in response to feedback, as discussed in the document on team experimentation.
  • Measuring success based on the wrong metric. In some cases, solutions without sufficient customer validation are provided to development and delivery teams as finalized requirements. In those cases, the success of the delivery team is then measured based on whether they delivered the feature as specified, not on whether the team actually solved the problem or achieved the outcome for customers.
  • Failing to address the customer's problem. Industry data (PDF) from A/B testing shows that in successful products, only about one-third of proposed features improve business outcomes when actually delivered. The remaining two-thirds of features deliver zero or negative outcomes for businesses. So, if you're not performing user research, odds are that at least two-thirds of the work that teams are doing is either not achieving the outcome they want, or it's actively making things worse. If a team is working on early-lifecycle products, particularly innovative ones, the team's odds at achieving the outcome they want are considerably worse.
  • Underestimating the time and effort required to respond to customer feedback. Teams should expect that their proposed solutions are wrong and prepare to discard at least two-thirds of them. As for the remaining one-third of solutions, teams should be prepared to evolve them significantly in response to customer feedback.

Ways to improve customer feedback

The field of user experience design (UX) provides a framework for understanding improvement. Many organizations treat UX as just making a product UI look good. However, UX is about whether you're solving a real problem for users; more broadly, UX is about every experience a user has when they interact with your organization. It cannot be overemphasized how critical good UX is to building successful products and services.

It's essential to build customer feedback gathering into the delivery process. Every significant feature you build should start by considering the problem to be solved. This should include performing user research to determine possible solutions and select a candidate. User research should be analyzed before a single line of code is written.

For early-lifecycle products, teams should adopt the ideas put forward in the lean startup movement to validate the underlying business model of the product before any code is written. You should validate that you're solving a real problem, and then iterate on a solution that solves the problem while providing a viable business model.

You should follow a similar pattern for existing products that are implementing a new solution to a known business problem. When the solution design has been validated, you should create a prototype so that you can perform further research and testing. Only when testing validates that the feature achieves your goal should the full production feature be completed.

Many organizations skip all of this work and fail. However, even strictly following these steps does not guarantee success. The point of this effort is to do as much as possible to minimize the risk of failure.

Ways to measure customer feedback

Sophisticated data gathering isn't required to establish whether customer feedback is gathered, visible, and acted upon. The following questions can help you determine how well you're taking advantage of customer feedback for your product design:

  • Do you have metrics that measure customer satisfaction? Are these updated and broadcast to teams regularly? Do you act on them?
  • Do you validate all features before building them and perform user research with prototypes as part of delivery?
  • Do you make changes to features based on this user research?
  • Do you actively and regularly gather feedback from users and act on it?

What's next

Was this page helpful? Let us know how we did:

Send feedback about...