Trust

By | March 13, 2014

Those following along at home will know that I recently wrote about the place of responsibility and ownership in software development. While valid throughout software engineering, ownership is especially important in the context of a rapidly scaling software business – we need to be able to rely on developers to get things done without a massive layer of management whom we neither have the time to recruit, nor the culture to support.

The seminal Netflix culture  slide deck (yeah it’s a doozy – I’ll wait while you read it), illustrates this wonderfully as the point where the growth rate of the business out-paces the density of high-performing individuals. This is traditionally the point where layers of management and process are introduced to deal with the ensuing chaos.

First ever 747The clue to solving this is imbued throughout that Netflix deck too: trust. Without trust we can’t expect responsibility to flourish. Without trust we sit on the shoulders of developers, imposing restrictions and demanding feedback. Proponents of Agile would call this an impediment.

So what does trust look like in a rapidly growing engineering team?

Solving the right problems

Newly minted development managers coming from a technical background (and there’s an argument they should only be from a technical background), often feel to manage developers one needs to prove one’s technical prowess. I’ll admit to falling into this trap myself, especially when working with a new team. If they don’t respect me as a coder, how can I have any authority?

Here’s the thing: the problems you need to solve as a technical manager are by definition non-technical. You need to create an environment in which the engineering team can execute to their potential, and otherwise just get out of their way. The last thing you need is technical authority. I suggest you get your technical jollies forking MEAN at home, and use your fading technical knowledge when liaising with the business, but otherwise close Sublime Text or Visual Studio and walk away.

Do you trust your team to choose the right libraries, architect solutions, solve loosely-defined technical problems, and code to their best abilities? How about do you trust the team to deliver a solution that is the best mix of JFDI and polish for a given timeframe? If you don’t, then you have a problem infinitely larger than your technical abilities could ever solve.

So, what problems should you be solving in order to build responsibility and ownership in your engineering team, and build a self-propelling engine of international awesome?

Focus

Work with the business to make sure the engineering team are working on the right thing, right now. Help them understand how to build the right way. The right thing of course means the number one (or two or three) thing that must be built to bring delight to your customers (new or old).

Here’s one I prepared earlier.

Clarity

You are undoubtedly right to abhor a 50-page technical specification, but you do need to demand enough information so that engineers can understand the scope of the task at hand, and ensure they can connect to the right people to get the detailed answers as they build. This might be Product Management, or perhaps direct customers. Facilitate and engage, then leave them to it; don’t dictate.

You need to be a navigation buoy in the flow between engineering and their stakeholders, rather than an hourglass through which all information must travel.

Foundations

Provide space and time for the team to lay their foundations. Documentation, tooling, code reviews, training, recruiting, onboarding, communication, experimentation. All of those things that are “invisible” to customers but oh so incredibly important to the smooth operation of a dev team and the production of world-class software.

You need to be the advocate for this behaviour, explaining to the business why it is essential. But trust, trust that the team can execute this foundation work themselves.

Growth

If you’re getting a lot of the above right, then your team will be a fecund swamp of talent. Engineers will be popping up and seeking out challenges. It can be super tricky to find opportunities for growth in a small team and company, but one simple way to promote growth is to delegate ownership as much as you can, because this supports the growth of both the team and the individual – things you need to be happening as the team size grows.

You can also facilitate internal presentation sessions to foster talent within the team, then support your engineers if they want to present externally.

Use your gut feel and previous technical experience to pinpoint areas for improvement and work with each engineer individually to help them improve those areas. We’ve found some benefit in plotting a modified Urban Airship Tech Ladder across four axes, which can show where an engineer perhaps excels in some areas but can improve in others. Great food for growth!

Output will flow

I imagine some readers are having heart palpitations because I haven’t mentioned delivery, or deadlines, or output. My assertion is that if you get the above right – especially focus and clarity – output will flow. Sure you can tweak the dials on time spent on foundation vs delivery work, but only for short periods of time.

Trust the team. They will deliver.

TL;DR version: Are your engineering managers solving the right problems, or are they an impediment?

 

 

 

 

One thought on “Trust

  1. Greg

    Killer clarity, thanks Ben. I especially like the idea of not being the neck in the hourglass; a visual metaphor I can carry around with me like [insert other visual metaphor portmanteau here].

    It’s also telling how well this approach translates; I am working on the exact same framework with my team(s) of weirdo fruity creatives and their production crew, only a very few of whom are developers, and it’s paying off handsomely. Once we help them know how hard they want to work, and how smart, they just do.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *