5. Sunsetting a Feature - When and How to Retire a Functionality

Sunsetting a feature—essentially taking it offline—is a topic everyone in product management knows is necessary, but it’s rarely discussed in detail. As a result, we often do it less frequently than we should, leading to overly complex products full of unused features. Let’s face it, retiring a feature is much harder than launching a new one.

We need to talk more about how to kill off features and how to sunset them in a way that doesn’t hurt the business or damage customer relationships. Here’s a guide to understanding when and how to sunset a feature, along with best practices for making the process smoother.

Evaluating Whether to Sunset a Feature

The main reasons for discontinuing a feature typically include:

  1. Misalignment with the current company or product strategy

    • Example: Uber shut down Uber Eats in certain markets.
  2. Too many bugs and high maintenance costs

    • Example: Twitter rethinking its approach to blue verification checkmarks.
  3. Engages only a niche audience that doesn’t align with company strategy

    • Example: A scheduling feature in Easy Taxi mentioned in a previous article.
  4. It solved a big problem in the past, but engagement is now low

    • Example: Old rewards program at Nubank, which became redundant with newer features like Ultravioleta.

A good way to identify if a feature is ready for sunsetting is by asking a few critical questions:

General Questions

Adoption

Satisfaction

Revenue

B2B-Specific Considerations

In B2B environments, the decision-making process can be even more complex. If only a small number of high-ticket customers use the feature, taking it offline might not be the best approach. Here are additional questions to consider:

Communication and Timeline

Even if you determine that a feature needs to be removed, it’s crucial to consider how this will impact key clients. In such cases, an extended communication period is essential. Some companies give six months' notice (or even 12 months for larger features) before sunsetting a feature, freezing any new development but keeping it available to existing users.

A clear timeline for customers is vital. For example, Mixpanel did an excellent job with the sunset of their "Messages & Experiments" product, providing ample time for customers to transition and suggesting alternatives.

Additionally, consider whether you can charge an extra fee to continue supporting the feature for specific high-value customers. This is common in enterprise SaaS products, where features that are no longer part of the standard offering can still be maintained for clients on custom plans.

However, keeping the feature active could create support issues, so make sure any additional revenue from maintaining the feature can justify the costs of expanding your support team.

Communicating Internally and Aligning Stakeholders

Internally, you’ll need a well-structured rationale to show why the feature should be discontinued, ideally linked to company strategy. Expect resistance from stakeholders in different departments who may rely on the feature or see its value.

Be sure to align closely with your leadership and peers, and map out which stakeholders might object to the change. A stakeholder matrix can be a helpful tool for structuring your communication plan.

Practical Steps for Sunsetting a Feature

If you've made the decision to sunset a feature, here’s a step-by-step approach to ensure smooth execution:

  1. Involve Product Marketing to craft clear messaging, if possible.
  2. Offer alternatives so users know what to switch to once your product no longer supports the feature.
  3. Thank your customers for their support and use of the feature.
  4. Provide clear contact information for any questions or concerns customers may have.

Conclusion: Best Practices for Sunsetting a Feature

By following these best practices, you can make the sunsetting process smoother for both your business and your customers, ensuring that the transition is as seamless and minimally disruptive as possible.