To Kill or Not to Kill

After my rant last week about product managers, the excellent Joshua Porter (@bokardo) made a great point about it. He said, “In my own experience the hard part is knowing when to kill something vs. when to give it more breathing room, as sometimes a really new idea can’t really be tested in low fidelity.”

As much as I’d love to send my pageviews soaring by starting a flame war with somebody popular on the internet, I have to admit that he’s 100% right. Killing a feature or product is exceptionally difficult. It’s tough to know when to do it. It’s tough to figure out if you made the right decision. And it’s tough emotionally to let go of something you really thought was going to be great.

First, let’s talk a bit about why you kill products or features. You kill them because they’re not succeeding or because you don't expect them to succeed. That could mean that they’re not getting enough traction or because you’ve determined that they’re never going to turn into an important part of your business. You kill them because the ROI isn’t high enough to justify investing more resources in them. You kill them because they are using resources that would be better spent in other places.

They’re hard to kill precisely because you never know whether they’re just a few days away from taking off and turning into everything you thought they’d be. After all, every successful product went through some period of time before everybody found about it.

So, let’s look at a few questions to ask yourself before killing a product or feature. For the purposes of this post, I'm just going to talk about killing existing features or products. I'll probably address how to decide to kill things before you build them in a future post.

These questions won’t make killing easy, but hopefully, they’ll make it possible.

Why Isn’t Everyone Using It?

There are four reasons people don’t use your product or feature. Yep. That’s right. There may be thousands of reasons that people do use a product, but there are really only four basic reasons that they don’t.

People don’t use your product because:

• They don’t know about it 
• It doesn’t solve a problem for them 
• They don’t understand that it will solve a problem for them 
• The problem it solves isn’t worth the investment of time, money, or effort

Before you kill your existing product or feature, figure out why it’s not popular. For example, if you’re simply not getting any traffic to your page, it means not very many people know it exists. On the other hand, if you’re getting tons of traffic, but none of it is converting or engaging, then your problem is one of the last three. People are finding your product, but they don't want it or understand it enough to convert.

You can find out if the product solves a serious problem for people by talking to the types of people you expect to have that problem. Develop a persona that represents the sort of person who might suffer from this problem, and interview them about that portion of their lives.

Don’t just ask, “do you suffer from x problem?” Have them tell you stories about their real life experiences in situations where they might have experienced the problem. For example, if you’re testing to see if people need turn by turn navigation on their phones, you might ask them to tell you about the last time they were trying to get somewhere and they got lost. Then you might ask how often that happens. If it’s extremely rare, they probably don’t have the problem you’re trying to solve.

If you’re trying to figure out if they understand the problem your product or feature solves, you can do that by showing them your product or feature (or a mockup or prototype) and asking them to tell you about it. Don’t prompt or prep them. Just show them the product and say, “Tell me what this does. Who do you think it’s for?” You will be shocked by how often your perfectly crafted prose and imagery cause nothing but blank stares.

When determining whether or not your product is worth getting, don’t forget that money isn’t everything to your potential users. Sometimes there are switching costs that they’ll have to deal with or just the cognitive load of learning how to use a new product. I can’t tell you how often I’ve seen people stick with a completely suboptimal solution to a problem, just because that’s what they’re used to.

Regardless of which it is, determining the reason people aren't responding positively to your product will go a long way toward telling you whether to kill it or keep it. 

Who Is Using It?

So, once you’ve determined that there are people who are using your product or who you expect will use it because it solves a serious problem for them at a price they’re willing to pay, it’s time to look at who those people there are and how many of them exist.

A great company with a very engaged group of users recently killed a feature. Unsurprisingly, there was a huge outcry. They got many, many complaints telling them how sad users were that the feature was going away. If they had gone entirely by the comments on the blog post about removing the feature, they would have been justified in thinking that they were making a huge mistake.

Luckily, they didn’t do that.

It turned out that the number of people using the feature was an incredibly small percentage of their user base. More importantly, the people using the feature were not, by and large, paying customers. In other words, a couple percent of very vocal users who didn’t earn a cent for the company were upset by the removal.

While it’s always best to avoid making your users angry, there are certainly users that it’s safer to anger than others. Keeping a feature or product that is disproportionately useful to people who aren’t benefitting your business in some real way means that you have fewer resources to devote to things that might make you some money. 

The other thing to consider here is how many people you might reasonably expect to have use this product if everybody knew about it. Unless there is a huge potential market for your feature or the small market that exists is willing to pay quite a lot to use it, you may want to consider killing it.

Note: for those few people who inevitably write to me and complain that “it’s not all about money,” I would like to point out that it very frequently does have to be about money or you will go out of business. If you want to keep your 10 free users super happy, you go right ahead. I’m going to cater to the large number of folks who pay me. 

And yes, I do understand the difference between long term and short term gains, and I expect my readers do, as well. Assume I'm optimizing for lifetime value here and not simply what makes the most money right this second.

What Is the Actual Cost of Keeping It?

Now that we’ve determined that people are using your product or feature, we should figure out how much it costs to keep your product or feature alive.

Of course, if you’re talking about your whole product, this math is relatively easy. The only hidden cost to keeping your product alive is the opportunity cost of building something else. If you’re working on your current product, you can’t be working on something more promising.

However, if you’re talking about a piece of your overall product, sometimes it can be harder to figure out how much it costs to keep a feature alive. Obviously there is the cost of the engineers or customer support people or sales people, but often they’re working on other things as well, so it’s not clear that cutting any particular feature will really save you any money. If even a few people are using a feature and it’s already built, why not just let it hang around indefinitely?

Well, consider some of these hidden costs to keeping a feature:

• Bug fixes 
• Customer support 
• A more complicated code base 
• A more complicated user interface (more features means more cognitive load on your new users) 
• Server and infrastructure costs 
• Additional work if you decide to do a site redesign or visual refresh

These may seem like small things, and in some cases they are, but don’t ever think that a feature is free just because you no longer are actively building it.

Of course, there's another alternative, which is to continue to iterate on the feature or product. This obviously adds hugely to the expected costs. Let's say that you create a new search feature for your product, and very few people end up using it. The actual cost of that feature needs to include all the iterations and changes that you're willing to try before people start using it or you give up on it. 

What Is the Actual Cost of Killing It?

In a similar vein, sometimes things can cost more to kill than you think. Unhappy users can cause trouble in forums or for support staff.

Of course, I did mention above that it’s sometimes acceptable (and inevitable) to annoy some of your users, but don’t underestimate the work that it will take to keep that unhappiness from spreading throughout your entire community.

There are engineering costs to turning off a feature, as well. Either you need to pull it out of your code base or leave it there to rot. Neither of those options is free, although both can be ultimately cheaper than maintaining the feature.

If you’re killing your whole product, you are often throwing away a huge percentage of your customer base. Just because you’re pivoting doesn’t mean that all of your users will pivot with you.

And the same goes for your employees, if you’re lucky enough to have any. Killing a product or significant feature can be absolutely terrible for morale. Obviously you’re not going to keep a failing product just to make your employees happy, but make sure that you are prepared for the fallout – and possibly resignations - when you do decide to make a major change.

So, Should You Kill It?

In the end, it really comes down to the expected ROI, and the future is notoriously difficult to predict. Good customer development techniques can help you get a clearer idea of the eventual potential of a product or feature that seems to be failing. An honest assessment of real costs can help you determine the investment that you're really making into the product.

But it's an art, not a science. In the end, you're still going to have to make the decision. And it may be the toughest decision you ever make as an entrepreneur. Good luck!

Like the post? Here's what you should do next:

• Follow me on Twitter
• Pre-order the book
• Attend a workshop

Expect More From Product Managers

I talk to a lot of startups, and I've noticed a really troubling trend at some of them. A huge percentage of Product Managers at startups suck.

Not you, obviously. I mean, you're a product manager at a startup, and you're the exception. Good for you. I mean all the other ones.

I mean the Product Managers who are failing to do the following obvious things:

Understand their Users

I was visiting a company, and I asked about what sort of ongoing user research they were doing. It turned out that they had an intern for a few weeks who was doing a giant research study to try to understand how people viewed the product. The Product Managers would occasionally attend a session, but basically they seemed uninvolved in the research. They knew nothing about the planning. They didn't monitor to see how things were going.

The research involved talking to a lot of people who had heard about the product but who had never used it before to see why they weren't using it. The Product Managers' main job with regard to research seemed to be to read a report produced by the intern at the end of the few weeks.

Meanwhile, nobody was doing targeted usability research on whether people were confused by the product. Nobody was doing any sort of inquiry into how big customers were using the product. Nobody was calling people who had stopped using the product to find out why. Nobody was following up with new users to understand their initial experiences.

Unsurprisingly, none of the PMs in charge of deciding what to build next had any real idea about usability problems, what separated their biggest customers from everybody else, or why people tried the service once and never returned. And that meant that they weren't very good at figuring out how to build their product, so when they did add a feature, it very rarely improved any of their key metrics.

The single most important thing you can do as a Product Manager is understand your users. This is the foundation for making every single decision about your product that you will ever have to make. If you don't know who your customers are and what motivates them, you can't consistently deliver features that will make them happy. It really is that simple.

Know Exactly How their Product Works

I was talking to a different startup about their onboarding flow, and I asked the product manager to explain how it currently worked for certain types of users under specific circumstances. The product manager said he'd have to check with the designer to get all the details.

The designer, of course, knew exactly how the product worked, since she was the one who had designed it. She also knew all about the user needs and the issues with the current flow and what she wanted to try next. All of this led me to wonder why on earth she wasn't the one in charge of the product.

When I attended meetings with both of them, the product manager constantly had to refer to the designer when talking aout what things they had tried, what had worked, and what users were currently seeing. Whenever an engineer asked a question about how something was supposed to be built, the PM would simply pass the question along to the designer and then relay the answer back to the engineers.

As far as I could tell, the PM was nothing more than a buffer between the engineers and the designer. Once the designer started working directly with the engineers, the PM had almost nothing to do except track the project schedule.

If you don't know everything about how your product works and why it works that way, how are you supposed to make good decisions about it? The role of a product manager isn't to be a passive conduit for information. A great PM needs to understand the product well enough to make important decisions about how to change it.

Validate Assumptions Before Building

I'm not going to tell a specific story about this one, because I've seen it happen absolutely everywhere.

Essentially, Product Managers seem to fall in love with ideas. Maybe they get the ideas from meetings or from their bosses or the ideas appear to them in a flash of light one morning in the shower. But regardless of how it happens, today's idea is always the Big Idea that is going to change everything.

Once they've got the idea, of course, they jump straight to the question of how to get it built. Typically, they get the feature on the schedule, break it down, get a designer in to mock it up, and get the engineers working on it. At no time before all this happens do they stop and ask the question, "How can I tell if this idea is any good before I spend a lot of time and money on it?"

No matter how good your instincts are, you don't really know that adding this new feature is going to be an enormous hit with your users. Sometimes you're right, and the feature is a game changer, but just as often, an unvalidated feature has a negative ROI.

Good product managers do as much work as possible ahead of time to figure out if they're spending their resources on the right stuff. Maybe they devise a small experiment to test whether people will use the feature. Maybe they do a very small version of the feature first. Maybe they do a concierge version of the feature. Hell, maybe they even sell the feature before they build it.

Whatever their strategy, good product managers validate their features before they build them, and that's why their ideas are so much more likely to improve the bottom line of the company. They don't necessarily have better ideas. They just kill the bad ones before spending too much time on them.

Prioritize Changes Based on Business and Customer Needs

Everything can't be a top priority. That's not how priority works. But you'd never know that to talk to some product managers. I can't tell you how often I've seen product managers try to build everything at once, because they simply couldn't make the decision about what was most important to either the business or to users.

Unfortunately, this tends to cause problems as engineering gets spread too thin and the team gets confused about what they should be working on.

Without clear prioritization, nothing really gets done well or quickly. Everybody ends up working on different projects, and all the projects move much more slowly than they would if everybody worked together.

A good product manager can make priorities clear without micromanaging. She makes the decision about what to work on based on things like expected ROI and the outcome of early validation tests. She balances features that are great for the business with features that are great for the users, and she always tries to find features that are good for everybody. She looks at things like long term vs short term pay off and makes sure that features are delivering value to all the different types of customers - new users, power users, business users, etc.

She clearly says that the entire team's goal is to improve a particular key metric and then makes sure that the team understands what that means. She then prioritizes which features should be delivered first while monitoring the metrics, and she keeps the team motivated to follow up by making progress toward the goal very clear.

Keep Engineers from Getting Jerked Around

How many of you have been working on something when a product manager has suddenly told you to stop working on that and work on something entirely different? How often does that have to happen before you become completely unmotivated? Not very often, right?

Of course, sometimes things change. Bugs are found that need to be dealt with. Business people make new deals. Company priorities get shifted around. Organizations get shuffled.

A good product manager protects her team from as much of this as she possibly can. She does this by pushing back on pressure from above to change priorities midstream. She does this by making sure that what her team is working on is important enough that she has a good reason to keep them on it, even if her boss suddenly wants something new and shiny. Sometimes she does it by making sure that there is somebody who can triage high priority bugs in order to keep the whole team from getting pulled off their work in the event of an emergency.

A good product manager manages up as much as she manages down.

Do Jobs that Aren't Technically Product Management

Just because the word manager is in the title doesn't mean you're off the hook for actually building things yourself. This is especially true at startups, where there are very few people trying to do a whole lot of work.

When I was working as a product manager, I frequently wrote copy, answered customer support questions, touched up images, built prototypes, consulted with the CEO on strategy, ran scrums, responded to people on Twitter, and made user research calls. Oh, and about hundred other jobs that had nothing to do with management.

I did those things because they had to be done, and there was nobody else to do them, but I also did them because the act of doing them made me a better product manager. By answering customer support questions, I learned how users felt about the product and where they got confused. By touching up images, I learned how much time it took to do the job so that I could gauge how efficient other people were when they were doing it. By monitoring Twitter, I learned who was talking about my product and what they were saying. These things were all critical to doing my job well, plus it meant that we didn't have to hire a bunch of specialists to do these things.

If you're not doing things outside of your normal job description, take a good hard look at whether you're really doing the most important parts of the job. I'll give you a hint, if what you're doing involves hours of meetings every week, there are more important things you could be doing.

This Sounds Hard

This sounds hard because it IS hard. Product Management should be hard. You're in charge of creating something. You have to make important decision affecting a huge number of people all the time. If you think it's all just going to meetings and making schedules for other people to stick to, you're doing it wrong.

Like the post? Follow me on Twitter!
Want more info on Lean Startup? Check out the Official Lean Startup Workshops Series I'm working on.