Photo by Michał Grosicki on Unsplash

Over the past year I’ve become increasingly aware of a fundamental divide in the prevailing wisdom on how good products are built. I’m not talking about waterfall vs. post-waterfall methods. I’m also not talking about the differences between specific methods like Agile and Lean. I’m talking about different philosophies on the best way to build products in an already post-waterfall world.

These differences have been apparent for a while, but they came into stark focus for me over the past week, as I finished reading two books in quick succession: Getting Real by Basecamp and Inspired (2nd ed.) by Marty Cagan. Both books are interesting and worth reading by themselves, but even more so when you read them right after another.

Basecamp and Marty agree on the biggest challenges in building good products, but diverge quite often on how they believe teams should deal with those challenges. Let me provide a couple of examples.

The same, but different

Both books are adamant that value (defined as whether someone will use/buy your product) needs to be validated as early and as cheaply as possible, and that the old ways of doing things are expensive and wasteful. Basecamp says you do this by “racing to running software” and making sure that it’s cheap to make changes:

It’s ok to do less, skip details, and take shortcuts in your process if it’ll lead to running software faster. Once you’re there, you’ll be rewarded with a significantly more accurate perspective on how to proceed. Stories, wireframes, even HTML mockups, are just approximations. Running software is real. […]

With real, running software everyone gets closer to true understanding and agreement. You avoid heated arguments over sketches and paragraphs that wind up turning out not to matter anyway. You realize that parts you thought were trivial are actually quite crucial.

And earlier:

Change is your best friend. The more expensive it is to make a change, the less likely you’ll make it. And if your competitors can change faster than you, you’re at a huge disadvantage. If change gets too expensive, you’re dead.

On the other hand, Marty believes in building prototypes really fast, and testing those with real customers before you commit to code:

One of the most common traps in product is to believe that we can anticipate our customer’s actual response to our products. We might be basing that on actual customer research or on our own experiences, but in any case, we know today that we must validate our actual ideas on real users and customers. We need to do this before we spend the time and expense to build an actual product, not after.

And later, on prototypes:

Product discovery [coming up with a validated product backlog] involves running a series of quick experiments, and to do these experiments quickly and inexpensively, we use prototypes rather than products.

On the topic of “functional specs”, both books agree that writing long specs filled with “requirements” is a terrible way to build software. I don’t think any of us disagrees with that. But again, they diverge on the best alternative. From Basecamp:

So what should you do in place of a spec? Go with a briefer alternative that moves you toward something real.

Write a one page story about what the app needs to do. Use plain language and make it quick. If it takes more than a page to explain it, it’s too complex, The process shouldn’t take more than a day.

Then begin building the interface — the interface will be the alternative to the functional spec. Draw some quick and simple paper sketches. Then start coding it into HTML. Unlike paragraphs of text that are open to alternate interpretations, interface designs are common ground that everyone can agree on.

Marty favors a technique called the Opportunity Assessment for the vast majority of projects:

The idea is to answer four key questions about the discovery work you are about to undertake:

1. What business objective is this work intended to address? (Objective)
2. How will we know if we’ve succeeded? (Key results)
3. What problem will this solve for our customers? (Customer problem)
4. What type of customer are we focused on? (Target market)

[…] You need to ensure that every member of your product team knows and understands the answers to these four questions before you jump into your product discovery work.

Many ways to skin a cat

To summarize this another way: most people in this post-waterfall world agree that the biggest reason why software projects fail is that various risks are assessed too late, which ends up being too costly for the business to survive. Most even agree on the core principles to follow to fix this: tackle risk as early and as cheaply as possible. Where we are seeing the divide is in how this should be done.

One perspective is what we can call the prototyping movement (I’m deliberately staying away from naming specific methodologies). The goal is to align around business objectives and build functional prototypes to meet those objectives as quickly as possible, and test those with real users before engineering gets involved.

The other perspective, the real software movement, says that even that takes too long, and that nothing can replace the feedback you get from working software in production — as long as you’re able to make changes very quickly.

So which is it?

In our search for easy answers and silver bullets, the obvious next question here is, “ok, so who’s right?” But I think good product managers eschew such easy answers. Good product managers are always learning about different perspectives, but they have to learn through an added dimension—the lens of their own product and culture.

So for me, the real takeaway from these books hasn’t been the prescribed solutions — although those are certainly helpful as idea starters. The real takeaway has been all the roadblocks to good product development that I noted down as I was reading. The different kinds of risk we need to validate. The constraints we tend to miss when we brainstorm and plan. The heavy processes that do nothing more than slow teams down and make them unhappy. How to address those challenges in a particular culture is where the true art of modern product management lies, and what makes the job itself so difficult to pin down, define, and get good at.

I’ve learned a great deal about product, culture, and teams this year. But no lesson has been more valuable than this: the challenges to building good products are universal, but the solutions are not, and the biggest value I can add to the team is to work with them to figure out the best way for us, in our context to address these challenges in a way that ensures the team is happy and productive, and our customers love our products.

I’m guess if I have to make a New Year’s Resolution, getting better at this would be it.

There’s a lot you need to get right when launching a new product. There’s figuring out the user needs and the business goals; there’s testing hypothesis and finding product/market fit; there’s prioritizing the roadmap (or more likely these days, deciding not to use a roadmap at all)… the list goes on and on. So naturally there’s quite a bit of writing on this topic. Most product management advice (including, ahem, a certain book) covers this area of fresh or updated products very well. What we see less of is articles and books about how to manage a well-established product.

What do I define as a well-established product? It’s fuzzy, but I see it roughly as a product that has been in the market for more than a couple of years, has a steady and growing revenue stream, as well as a few healthy competitors. The reason I’m interested in well-established products is because, well, Postmark (where I work) is one. We have a product that most of our customers love, and it’s growing in both users and revenue. Of course we have ideas on how to make it better, but we’re in a different stage of our journey. We’re not frantically playing catchup with competitors or trying to complete a bunch of features that aren’t adding value to any customers yet.

This is, I’ll admit, a wonderful position to be in as a product manager. But it doesn’t mean there’s no work to do. I’ll stop short of pulling out the “software is never done” trope (in fact, I’d argue that software can be done). I will say that we’re definitely not done with ours yet. So as a product manager, I learned this year that well-established products face a different set of challenges than newer products. Here’s an incomplete list of my hypotheses about managing such products:

• Prioritization becomes a much finer art (with a little less science).
• Roadmaps become more important.
• Process becomes more necessary.

I’d like to spend a little time discussing each of these issues separately.

Prioritization

When I started at Postmark we went through a full product discovery and prioritization exercise. I wrote a long post about it on our blog: How we built a product vision and roadmap. I also followed that up a few months later with From roadmap to shipping: effective product management for remote teams — a post on how we moved from strategy to execution. Something I didn’t get into in those posts is how my understanding of prioritization evolved as we moved further into the details.

Postmark is a tool for developers. That means that feedback from customers comes in often, and it comes in detailed. Feature requests are usually incredibly specific (and useful), ranging from a field that’s needed in an API, to missing functionality and how that affects their business. Due to the sheer volume of high-quality feedback we get, I learned quickly that I needed an efficient way to categorize and prioritize things as they come in, otherwise everything would just go die on a backlog somewhere.

The biggest shift here was to move to a system where we prioritize themes, not individual problems or features. A theme is loosely defined as a particular customer problem or opportunity we need to solve — similar to the “hypothesis” nomenclature in lean development. Now, when new ideas come in to JIRA (Yes, yes, I know; more on JIRA later) — either internally or from customers — I follow a set of steps to (1) process and (2) prioritize them:

1. First I do a gut check: is this something we’d ever do? Does it point to a customer problem or business opportunity that would be beneficial for all of us to solve? If so, great. If not, the request gets rejected with a note so we can get back to the customer or internal stakeholder.
2. If a theme already exists for a particular idea (e.g., Improvements to the Bounce webhook, or Improvements to the Statistics page), I add it there, and considered it processed. It would be too overwhelming to think through the details of every single idea, but I know that when the time comes to work on a particular theme, it’ll be there, and we’ll go through it then.
3. If a theme doesn’t already exist for that idea, it gets a little trickier. Is this important enough to group together with other things to create a new theme? Do we expand a theme to “make room” for the idea? This is where things get decidedly less certain, and I have to be comfortable with a certain amount of fluidity. I might create a new theme, only to merge it into something else later. That’s ok. As long as it’s somewhere manageable where we won’t forget about it, I’m happy.

During our quarterly planning I don’t make the team go through every single JIRA ticket (to their enormous relief). Instead, we discuss our OKRs (Objectives and Key Results) for the quarter, and then decide which themes to work on that line up with those OKRs. Once our themes are prioritized, I work with the teams to figure out how best to address it — it’s usually a combination of some of the ideas already inside that theme, and a few new ones we come up with.

And this leads me to my next point…

Roadmaps

I’m still quite amazed at how out-of-fashion roadmaps have become. It seems to me that a lot of people got burnt by 2-year corporate plans that weren’t allowed to change, and then decided that the only way to deal with that (admittedly atrocious) situation was to get rid of roadmaps altogether.

I respectfully disagree.

I’ve written about this before, so I’ll just reiterate one bit here:

The purpose of a product roadmap is to set forth a long-term vision for the business, and break that up into smaller, meaningful pieces of work, based on what you know now. It’s fallacy to believe that this is an unchangeable list of dates about where the business is headed. A product roadmap that doesn’t react to day-to-day changes in the market and within the company is a pretty dumb document.

This is, in my experience, especially true for a well-established product. There is a lot more leeway to mess around and see where things go on a new product. With an established product, it’s way too easy to get off track (Evernote, anyone…?). So what I argue for is a flexible quarterly roadmap of prioritized themes. You don’t write down dates next to a long list of features. You don’t make hard commitments for releases that might never happen.

Instead, you put together a list of themes you’d like to work on, in order of priority, based on how closely they align with your objectives and key results for the quarter. And then you meet every week as a core team to discuss a couple of things:

• Did any new theme arise over the week that is important for us to discuss?
• If so, and if we should work on it soon, what does it replace?
• If not, then let’s get back to work.

This keeps the roadmap flexible, and it keeps the team aligned on what the most important themes are for the product and our customers.

Process

Which brings me to my last point… process. I’ll quote Michael Lopp on this until the day I die:

Engineers don’t hate process. They hate process that can’t defend itself.

Considering the sheer number of detailed input we get on the product as well as the amazing ideas coming from inside the team, there’s no way I could just wing it. Maybe there are some PMs who can hold everything together in some kind of virtual kanban board in their heads, but I just can’t do that.

So yes, we use JIRA, and you know what? It works for us. The Customer Success and Marketing teams submit ideas and hypotheses with links to the customers who request it. If and when those ideas make it into a theme, they continue to add more customers to it as we go. They can always see where something is in the development process, and when it gets released they get to go tell a bunch of customers that the thing they asked for is now live. And that is really cool.

I understand why people hate JIRA, I really do. But I think I’m such a big proponent of it because we were able to come up with a way for it to benefit the whole team, and actually make their lives easier, not harder. We have a process that can defend itself. And I don’t think you can work on a well-established product without something like that.

---

This year has been an enormous learning experience for me. I’ve never worked on a product like Postmark before. It’s a product that is basically universally loved, where feedback comes in from people who are passionate about it and can’t imagine their development lives with out it.

As much as it is a dream to work on a product like this, I feel like I had to learn a whole new set of skills to be effective as a product manager in such an environment. I know I still have a long way to go, but I can tell you one thing — I’m excited to learn more in 2017.