All I have done wrong running a product team


Two years ago I had an opportunity to shift my career towards product management. It took me a while to accept the job, it was no easy decision. I needed to be ready to lead people doing a job I had never done before. I needed to take product decisions. While I felt ready on the leadership part, I was afraid of not being able to get up to speed with the product part fast enough.

Two years passed and I learned a lot. Along the way, I made a few mistakes and in this article I want to focus my attention on those four or five things I consider crucial. Now these learning seem just common sense to me. I often find myself summarising this experience in a joke:

Common sense is not so common

This is the reason why I feel compelled to share the learnings I consider basic product management now. I will present my list in no particular order.

Data is nothing. Information is the key #

We have introduced blazer at work recently. I’ve wanted to add this little tool to our portfolio of internal applications for a long time. The reaction coming from other departments was great. Everyone was happy about the tool and it was a stimulating process for me. I started asking myself what the real difference between these two processes is:

  • A product development team providing data for the rest of the company, and
  • Everyone being able to ask questions on their own

My assumption was that people were happy about shortening the feedback loop. It’s like writing tests instead of deploying code, right? It takes less time to figure out you broke something if there are tests around.

I was wrong. My technical background made me focus on the wrong aspect of this problem. Everyone wants answers faster, no discussion there. However, people’s enthusiasm was not driven by the shorter lead time, they were happy because they could finally participate in the process of transforming data points into useful information. One day I came to realise data is nothing. Blazer was useful because it could help with information, not because it increased data availability for the rest of the company. No one really cares about the conversion rate comparison in two different countries. Everyone wants to know the why of the differences and similarities. The information that explains the data is the key. People can do a better job if they can extract knowledge out of data. Knowing why a company performs better in a market is the key information. Finding the answer will improve your marketing strategy, and it could change the story of your company drastically.

Working with product managers and senior management meant being in conversations where all you would hear is data. It took me a while to realise I was confusing data for information. I had no process in place to make that transformation happen. What we needed was a systematic approach to this process. To cut a long story short,

When presented with data, ask yourself: why that data?

Ask questions all the time. You hear data points and you automatically ask yourself why. What is the explanation behind it? You need to transform all the data points you hear into useful information. I believe this process should back up your roadmap. It’s a three-step process:

  1. Look at the data
  2. Ask yourself why
  3. Extract information

And I think it should be a big chunk of the product discovery phase. Most of the time the answer to the question “what’s next?” is already there, well hidden in your data points. You don’t know the answer because you’re not systematically asking yourself why.

Say no to CEO features #

I define “CEO features” as features we build because someone says so or just because it sounds cool to build them. According to my definition, those features do not always come from the CEO. The key mistake is fast tracking features without the necessary scrutiny. It was very tricky for me to figure out how to handle the features coming specifically from the CEO. Here is the problem:

  • My CEO set up the vision of the company
  • My job was execute that vision translating it into a product vision
  • My team and I translated that vision into a product roadmap
  • We built things from our roadmap

The problem lies in the tiny border between vision and roadmap. I think they need to overlap, they must complement each other. And that’s exactly why this has been so tricky for me. I struggled a lot with the concept as I could correlate with my CEO saying “we should do this because it’s the right thing”. It happened to me often while leading developers: to struggle between giving people enough room for decision and needing to do what I believed was the right thing. So how do I say no to a “CEO feature” and keep the overlap between his vision and the product vision?

It was hard, however, it had a very simple solution. The word feature is not present in the sentence “translating the vision into a product roadmap”. It’s about the vision and its implementation: a roadmap is not a set of features. A crucial mistake I believe only a newcomer like me could make. The roadmap is a visualisation of your strategy, a high level concept. Nothing about the CEO vision and the product roadmap is about the features you build and it took me a while to truly understand it.

Once again, it is just common sense. Of course you should not let your “CEO features” be fast tracked. Of course it makes no sense to fast track any feature. Now I think I have a good understanding of the relationship between the CEO and the product team and I can back up my “No, we won’t be doing that” with good reasoning. It’s not about a list of features, it’s about clear strategy and its execution.

Say no with reason #

Saying no seems to be a required feature for good product management. I can agree to that sentiment but I think it’s an incomplete statement. When I started running my first product team, we did not have enough people and experience to say no. I struggled with it as I felt we were not saying no often enough. It had bad consequences:

  • Irrelevant features were creeping into our building cycle
  • We would often build the wrong thing and only realise after the roll-out
  • We would build more than we needed

It was not the nicest situation and it was that rare case in which hiring was the answer (hint: use hiring as the last answer to your problems, it brings the best in you and the people around you). We hired more people and brought in the necessary skills. Finally my team was saying no to feature requests. Oh, it felt good. We were saying no to things that made no sense to us. But it had consequences for the rest of the company. People were frustrated by this new attitude. I was expecting it but I still got something wrong: I thought it was a temporary phenomenon. I was telling myself that we just needed some time to adapt and people would understand why saying no to most requests was the right thing to do. Here we go, it’s just common sense again: why would people understand a no without any context?

We were saying no but we were not explaining its benefit. It turns out this is easy to fix: you say no to people and always provide them with an explanation on what you are going to do instead. It’s as easy as that. This approach comes with mutual benefits for both your product team and the rest of your company:

  • The rest of the company understands better how you prioritise product work. It’s vital they get it.
  • You have to convince people your prioritisation is correct. Great side-effects:
    • You learn how to express things clearly and in plain English. No jargon, no acronyms. It makes your communication more inclusive.
    • Together with other teams you can catch errors in the prioritisation and that is a great outcome too.

Do not ship what you can’t measure #

I must confess I felt guilty while reading about feature factories. We were showing clear signs when it came to impact analysis. What I did wrong here was realising too late how important it is to design a process around impact. It felt especially bad to realise I was doing it wrong; it’s obvious to me there is value in analysing the impact of features.

Nowadays, I would be able to design a process around impact analysis but I have never tested it yet. I think it would be as easy as:

  • Require all the feature specs to provide an expected impact.
  • Set up a recurrent meeting with the product team to evaluate expected impact versus reality.
  • Share the outcome of the meeting twice: with the development teams and with the rest of the company.

I consider this process a good start and I would love to evaluate it during retrospectives. I am curious to hear other folks’ opinion on the subject as my approach is still just a hypothesis that needs testing and feedback. I like to say:

In theory, there’s no difference between theory and practise, in practise there is always a difference.

Get rid of features no one uses #

Deprecating features simplifies your product, simpler is better. I think feature deprecation is strongly connected to impact analysis. We did not do this enough because we did not know enough about some of our features. This way our product kept growing in terms of complexity and we missed the opportunity to make our life easier and, above all, the experience of our customers better.

Given my background as a developer, I found myself thinking about this problem the same way I think about refactoring code. It helped me realise I can approach decommissioning of features as I approach a code base. My approach is

you should not be refactoring your code

It’s more of a catchy title for a short article but I think it conveys my message. You should be improving your code base with every single change. You should be doing it up to a point when you don’t need any set of changes (developers tend to call it “pull requests” these days) consisting only of refactored code. Applying this approach to product management can translate loosely into:

Simplify your product with every new feature. Give more to your customers by having only the features they need and use.

Conclusion #

Running a product team was great experience. I learned a lot and had fun building features for such an exciting product. Product management is tricky: the job is not well defined and being on the border between technology and business is challenging. Over time, I enriched a metaphor I had read somewhere in the past (maybe it was Dr. Alan Cooper’s writing) and I use it often to explain what’s tricky and exciting about product management. I compare product development to the construction of a building:

  • Product people are the architects. They decide how the windows work, where the stairs are, where and how the lifts work and so on.
  • Developers are the construction workers. They build all the blocks and care about all the tiny details that bring the building to life.

The kind of decisions product managers take are hard: stakeholders think they know which windows you should put in your building because they use windows. As a developer, I’ve never experienced this problem; there are few people who actually know how windows work. Stakeholders just want to use windows or give them to their customers. This difference between product management and development was an exciting new challenge for me.

I’m glad I accepted the job back then and I’m happy I will be dedicating my time to other things now. I have a lot of writing ahead of me and I have never been so excited about it in my life.