How to Sunset Features without Disrupting Users

Dorra Mlouhi at UXDX Community: Accessibility, AI UX, and Feature Sunsetting. Video: https://youtu.be/IDlpS5gdz7s

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Introduction and agenda

[00:00:00] Dorra: Today I will start by talking about something unusual to talk about publicly, especially for product managers, because we're usually known for building something with the engineering team. But in real life we build, and we also sunset things, when we think there is a need to take this decision, for several reasons that we will go through together today. I will share with you the best practices, but I will also share my personal experience. And mainly I will tackle it for products in general, because in sunsetting a product or a feature you are still impacting customers.

[00:00:41] So this is Dorra. I'm a product lead at AWS. I previously worked for trivago, Tesco and Al group [?], and I'm based in Berlin, Germany. The agenda today is: why we sunset a product, when to sunset a product, and how to sunset, but most important, how to measure to make sure that this is a success, and the risks that I mitigated through this process. I will share a few use cases from the market in the last five years, and also my story.

[00:01:19] I wanted to open with a quick question. If you are here working in the product field or the engineering field: did you ever retire any product in the past in your career? Ask yourself that question. Or, if you did not, did your company do that? And also try to remember any product or company name that retired a product in the market in the last five years. By the end of this presentation we will see together if this is really something common, or if it's just because it's unspoken that we didn't realize it's coming.

Why sunset a product

[00:02:01] So, why sunset a product? I tried to summarize the five reasons that I've heard about, or that other colleagues experienced, or that were my personal experience. The first reason, by far the most usual, is low adoption or usage. Companies today realize that if you have low adoption or usage for a specific feature or product, it's time to move to something else and try to see where you can shine more, rather than investing more time and money in optimizing that. We will see more in the use cases.

[00:02:48] Obsolete technology: today IT is moving so fast, and some companies decide to build new products from scratch with new technologies. In my previous company, we had to rebuild everything from scratch on microservices, so we literally had the new e-commerce built instead of the old one. They wanted to really move toward new technologies.

[00:03:13] We also have strategic shifts, especially for big companies that have several products and are managing a huge portfolio of products. They shift based on the priorities and based on their financial goals, and they decide, okay, this is not fitting our vision anymore, and we think it's not needed anymore to invest in this product or feature. Performance or quality issues: this is really the least common one I've heard about, I will say, because usually we always have a plan for how to make things better. But for small features, sometimes it's even optimal to do this sunsetting.

[00:03:51] And finally, mergers or acquisitions. If you've had any startup experience, for sure you will know about this. Mergers and acquisitions are hectic, and they can end with you having to make a decision on whether you want to keep this product or merge it into the existing product, and in the end sunset that product. Those are more or less the five reasons that are the most common in the market.

When to take the decision

[00:04:21] So when to take that decision? Each time I hear this question. I even had someone asking me last week, because they are going through this journey: how to know that it's the moment, how to know really that I need to sunset that product or feature. This question is really hard to answer, and you need to convince yourself before convincing your stakeholders or your entire team and company. So it's a lot of thinking from the product side, because product will be the first person who starts having that doubt, whether they really want to sunset this, and starts thinking.

[00:05:00] How I process this is to always gather your thoughts in one place and use data and data and data. Nothing is more important than data. The first thing is to collect your performance metrics. If you have any performance metrics that you are tracking, make sure that you have enough view of the history, and really ask yourself whether the performance of this feature or product is really the problem. Once you gather that data, document it.

[00:05:41] Then also do a market analysis. This is really, really important, because you need to know, if as a product or as a company I lose this feature or product, how I will be positioned against my competitors. Am I really losing a huge market opportunity, or is that a trend, and my competitors are also not going in that direction? It's good, when you are talking with stakeholders and any leadership board that you have, that you have the right arguments to use, because you will be asked about that.

[00:06:17] Strategic alignment: it's always good to see in which direction your company is going and in which direction your product is going, and whether this feature or product is aligned with that strategic direction. That might be your answer: you don't feel that you are aligned anymore with that vision, and you need to make this hard decision.

[00:06:48] Identify customer impact, because in this situation customers will be impacted, and it's not a hundred customers, it's millions of customers. So check how customers will be impacted. And I always recommend having your trusted circle of customers, the friends who are customers that you test new features with often. Try to have a meeting with them and hear what they think about this product or feature, and see if they find it useful or not. Are they using it massively as before or not? How has the usage decreased over time? How do they see this feature, how do they evaluate it? Is it really something they're interested in?

[00:07:33] Of course, when we talk about sunsetting a product or a feature, if you have any legal commitments, you need to check them, especially with customers with subscriptions, with contracts. The legal team is your best friend in this situation. You need to go and discuss together and see how this will impact them legally, and what your duty is. For example, we will see in the use cases later that there is always a very long period of time between when companies announce any sunsetting and when they literally start executing that sunsetting, because of this legal commitment.

[00:08:20] And feedback and strategic and stakeholder input. This part is the last part. As soon as you go through all of those, you do all your homework as a product manager and your research, you will start collecting feedback and sharing this data and your thinking with a small circle, and enlarge that circle slowly, and collect feedback and incorporate feedback. Because it will not be taken positively in the beginning, but data is your key, and that's why I was insisting on this.

Executing the plan: timelines and plan B

[00:08:56] Now the hardest part. The hardest part is not the decision but executing that plan. Because it's not like implementing a product or a feature, where you know approximately how much it will take you in terms of time and effort. With sunsetting products more precisely, because it's really huge, it will take you time, depending on how customers are feeling, depending on how the first steps of your plan are going. So you will be really under construction. You will be adjusting your sunsetting timeline almost every week, to make sure that you are doing the right thing and that you are giving customers enough time, but also that you are sure about every step that you have.

[00:09:52] Of course, as with any other project, you need a timeline, and you need a timeline that is really tailored to the sunsetting. In the beginning it was hard for me to digest the differences. I was trying to make milestones and show this positively, but it's actually different. It's another nature of project that you are working on. The project timeline is more about your execution in general: when you are announcing, when you will start the sunsetting, how long customers have to retrieve their data, what your legal obligations are, from when to when.

[00:10:34] And there is another timeline, which is the communication timeline, where you will start communicating to customers, really depending on how big the product is, or whether it's a feature. For example, for big products you need to send them monthly updates, reminding them about the sunsetting timeline and date, and communicating any updates or information that they need to know.

[00:11:01] And of course, when we talk about sunsetting a product more precisely, you cannot just announce it to the customers without providing them a plan B. This plan B can be that you provide them an alternative internally in your company. For example, in the case of an acquisition or a strategic shift, for sure you have something that customers can use internally in the company. But in this situation you can even go deeper and recommend partners and external third parties. Recommend them, and make this information really visible, because you will help them to speed up their offboarding and make it a really smooth and less painful process. And that's your duty. It's not like there's a red line here and there is competition. No. Here there is the interest of the customers, and you are just trying to ease the customers' pain.

Customer support, data and closure

[00:11:58] Of course, in this situation you will also work a lot on customer support and data management. That means: who will be the contact person for customers in case their offboarding is not going well? How can they share their feelings and feedback with you? Because you will hear some frustration from customers. And data management, which is: for the data that they have in your tool or in your product, how long do they have in front of them to really retrieve the data?

[00:12:31] And all this feedback that will be shared with you from customers, you really need to monitor it and build your own framework for how you structure the feedback you are receiving, from sharing the news the first time with the small circle to making it public information. Monitor it in a way that it's really less painful for customers, with clearer communication. It's a learning process. Your first communication is not as clear and intuitive as your last communication. And also share this information with the stakeholders, the leadership team, anyone who is involved in this.

[00:13:08] Of course, the last step is closure and evaluation. It's not that because we are sunsetting a product we are not celebrating it. If you are doing it properly, without impacting the customer really badly and without having negative publicity, you are, as a team, able to celebrate this. And you need to share your learnings and really do a retro, because this is a unique experience. I'm sure that in my network not many product managers have gone through this. So sharing this makes other people conscious that, yeah, this is a possible direction, and I should not be afraid of doing it, because we have public references available on the internet now.

Measuring success and mitigating risks

[00:13:54] As with any other project that you are working on, you will be measuring. But I will say it's less sophisticated with sunsetting. That's at least one positive point. You will mainly just measure two things: the user feedback that you are receiving, and the migration metrics. How many customers are offboarding? Who is asking for extensions for certain circumstances? It's really less sophisticated. You don't need those sharp dashboards. It's really a few metrics, but they have a huge value as they progress.

[00:14:36] Of course, anything has risks, and the risks to mitigate in sunsetting products are really huge risks, because they even put the company's reputation in the game. So you really need to make sure that customers are always, I will not say happy, but you need to listen to the customer and try to understand how to ease their pain. Maybe their pain will be eased by just referring them to a third party that can help them with the offboarding, or maybe by just listening to them, explaining the timeline to them more, and accommodating any possible extensions, if you can afford that.

[00:15:21] This leads to the second item: this will help you to avoid negative publicity, because especially for big names, the key players in the market, they will do everything to avoid any negative publicity, especially in those situations.

[00:15:38] Employee morale. When you are working with a team, they are not building. They are removing something that maybe they even contributed to building since day one. It's not easy. You need to make sure that the team is motivated. You always need, as a team, to have your dailies and make sure that you are talking openly about this, sharing honest feedback and motivating each other. Because this is not a few-months process. As I said, in the majority of cases it will be a year or more. So the team needs to be aware that this is just a step that we are going through. Even if we are not building something, this is adding value to our company and helping our company grow in the right direction.

Examples from the market: build fast, kill fast

[00:16:27] Now, if you took some time in the beginning to think about a few names: I added four names that I hope people here have heard about at some point, if we don't have too big a gap in age. Google Inbox, for people who used it. I didn't use it, to be honest, but I heard about it. It was announced in 2018 and deprecated after, I think, two years. Also Adobe Flash Player, for the Adobe lovers. They announced the deprecation in 2015, and it took them three years to finish that and end the support.

[00:17:19] As I said, the bigger the product or the company, or the more important the feature that you are sunsetting, the less easy it is to do it overnight. You cannot turn off the light one day and just pretend that you didn't offer that service, because you have commitments before that. It's a legal commitment, it's a moral commitment between the company and the customers.

[00:17:45] Of course, Google Plus. I think this is not really unknown. Google Plus did not really have the success expected by Google, and the deprecation was announced in 2018, and I think it was officially turned off only last year or the year before, if I'm not wrong. Windows Live Messenger service also. I tried to bring examples from different companies to show you that this does not reflect that the company is not successful or that the company is not doing the right thing. It's under the code of "build fast, kill fast."

[00:18:36] So always try not to be emotionally attached to what you are building, and criticize yourself first. What you are delivering: is it really adding the right value that customers are expecting? Do I have another opportunity in my product or in my company where I can spend the same time and energy and money to deliver something that will add more value than what I'm doing today? So always think of it not as a failure, because it's not a failure. It's not a mistake. None of this. It's just happening. Everything has its own time. The cameras of 10 years ago are not the ones of today. Even the iPhone when it was released is not like today. Everything progresses, and it's normal that we build better versions, and this is what companies are doing.

[00:19:26] And I'm happy that in the market today people are more open about talking about this. It's really more common now. If you Google this, or if you are a ChatGPT person, as we were talking about AI in the previous session, you can ask ChatGPT about the sunsetting that happened in the last five years and how it impacted the market and the customers, and you will see how this is a tailored process, but one that's been well defined through time.

My experience of sunsetting a service

[00:19:59] Now I will share my experience. My experience of sunsetting a service or a product was really unexpected. When I started, I was really thinking deeply in the beginning about whether it was really the right thing to do in my career. But I realized, when I was talking with stakeholders and talking with my team, that this is something that adds value. It means that it's something you do as product: I built products, and I'm now building a product, but I also sunsetted a product. So you have the 360 overview of the product life cycle. It was a journey that had ups and downs, but overall it was a very, very interesting journey that I learned a lot from.

[00:20:52] The first thing is that you never, ever talk about the decision of sunsetting something before you have the right data and you know who the right people to talk to about it are. Because, as I said, people can take this discussion in different directions, and it can be discussed negatively and impact the team. So it's really good to be sure about what you are using in your arguments and whom you are talking with.

[00:21:21] I also learned a lot about customers. I never talked to customers as much as I did when sunsetting a product, because you will have 10 or 15 customer calls per week in your calendar, and you will start hearing different opinions. You will even discover customers who were expecting that this service or feature would be sunsetted. And customers are really smart. They inspired us a lot in building the alternative solution that they are transitioning to. I was really happy that I had several rounds of testing with customers and interviewing them to really build the next replacement for them.

[00:22:05] But I also prepared myself emotionally that some customers would reject the alternative that I was offering. I used that as a learning, because I was asking them where they were moving, and I was using that data to see how to optimize the alternative that I offered and how to make it a better version to compete with the other solutions that customers preferred at certain times during this experience. So this is more or less it. Please feel free to ask your questions. I finished a little bit early, maybe I was speaking fast, but feel free to ask your questions, and if you have any other questions after the session, feel free to message me on LinkedIn.

Q&A

[00:22:56] Host: Excellent, thank you very much, Dorra. That was a great overview. It's an incredibly important topic, because you do have to recognize that sometimes products don't work as you expected, and holding on to them and dragging on for too long is doing damage. So hopefully people out there recognize that.

[00:23:15] Dorra: It's just losing time, money and energy.

[00:23:18] Host: Exactly. So if anybody has any questions... We've got some comments coming in: "RIP Flash," and "Magento 1.9 took around five years to sunset, but Magento 2 is a totally new product." So people are seeing the pain that they've experienced already. But if you have any questions, please do post them in and we'll ask them through to Dorra.

[00:23:43] One thing, and I guess it's becoming almost a meme on the internet, is how Google likes to kill products, and two of your four examples were killed by Google. I've been burnt with Google Reader, with Google+, with Google Domains, with Google Photos, their brand accounts, recently. Just different things that keep getting killed. I've heard a lot of people say they don't trust Google Cloud now, because they're worried that something they build on top of might get sunset, because Google's quite quick to kill things. So is there a risk that you could be damaging your brand, and people start losing trust in your products because features they rely on start getting cut?

[00:24:28] Dorra: I would answer this question differently, after experiencing this. We always analyze, unfortunately, the titles, but we do not get into the details. I was a customer of Google for a few of those products, and I always read the email carefully. If you read the email, you will find the reasons, but you will also find what they are proposing to you. I do not agree with this reputation, first because, according to me, smart companies build fast and kill fast. You see, the competition in the IT world today is really huge, and everything is moving fast, so all the companies are making sure that they are providing the right product that they can offer and guarantee continuity for.

[00:25:35] So for any product that you are using today, you might ask yourself if you are doing the right thing just relying on that thing, or whether you have a plan B, for anything that you have today. Imagine if we wake up tomorrow and iCloud is not working, is not there anymore. What would we do with our iPhones? So you always need to make sure, especially for people using cloud, that you are using the right thing that fits your needs and requirements, but also that you trust the brand. Because no one will kill something as massive as Google Cloud without providing the right steps and alternatives. That's for sure.

[00:26:29] But I will say that Google Cloud is not the same case as other products. Google Plus was a B2C experience, but Google Cloud is mostly B2B, so it has other dimensions and other complexity, and I'm sure Google will not really put any risk there.

[00:26:57] Host: Yeah, no, it's definitely a cash cow for them, so hopefully that's enough to keep their services alive. One of the other things: you mentioned using user feedback as one of your metrics. I'm just thinking your user feedback is not going to be positive at the start. So how do you use that as a metric when it's veering negative, and how do you track that? Because you might be able to pull it back a little bit, but I'm sure people are still going to be upset.

[00:27:28] Dorra: Yes, communication is the key there. If you remember, in one of the slides I was insisting on splitting your project timeline, or deprecation timeline, and your communication timeline, because communication will be your key in easing the pain the customer is going through. The first thing is the way that you are communicating. You will make sure that the wording in the monthly or weekly emails that you are sending to customers becomes more adjusted to this feedback. The first weeks they were super angry. Later they started asking constructive questions. Later they started sharing their plans and asking for our feedback. So you will see this progress.

[00:28:19] The second thing: it's better that they say those aggressive things and give that feedback to you than say them publicly. I always say it's like having a teenager, and you're trying to have all the teenager's ups and downs happening in your house, not outside in the street, not in the school or high school. So that's really important. And by the way, you will get very extreme feedback, but you will also get surprising feedback. I got feedback like, "Oh, really, finally you are doing it. We were expecting this for a good period of time." It was like, okay, now we can speak the same language.

[00:29:05] Customer communication is the key: arranging round tables, private calls, especially for the ones who, in the first days after the email, will see just a black screen. They will not even accept to discuss things, but slowly they will get to discussion. And help them and offer them things. Offer them vouchers, offer them a budget to help them migrate to other alternatives. Because after all it was your decision, and you will not really let the customer pay the bill for this decision fully.

[00:29:47] Host: Makes sense. Well, that brings us to time. So thank you once again, Dorra. That was a really great angle, because it is something that we don't talk about enough. We always talk about creating, but you also have to see the end of life of products and features as well.

[00:30:02] Dorra: Thank you. Thank you so much for this session.