Flywheel effect, part 1
In the previous post I presented 3 sets of actions that you can do as a new director in order to stabilize your org and set it on a new trajectory. Those actions are mostly focused on the first few months of the role when you need to show some positive progress, but can also be taken following that. However, they are not enough in the long term and will have diminishing returns: you’ll have eliminated some very bad bottlenecks, have a solid roadmap in place and automated some of your org’s processes. So, what comes next?
The answer is in the title 🙂
A flywheel is a heavy wheel that stores energy as it spins. Its inertia makes it hard to get moving from a standstill, but every push adds a bit more speed. That same inertia then keeps it going: once it’s spinning, it takes far less effort to keep it spinning than it did to start it.
An org can work the same way: when you make an improvement that frees up some of your people’s time, they can put that time into the next improvement, which frees up even more time, and so on.
As a director, you’re not the one pushing the wheel every day, your managers and their teams are. You need to get it moving in the first place, choose how and when to push it even more, and keep friction away from it.
The actions from the previous post are designed to remove some big barriers and free up time to think and work, and that’s your first push. What does the flywheel look like after that? Well, the true-but-you-hate-to-hear-it answer is: it depends.
Personally, I’ve found it very powerful to adopt a flywheel state of mind: when presented with ideas, proofs-of-concept or products, either from people inside my org or from the outside, I always judged them through the lens of the flywheel: is this something I can use to improve what a day looks like for an employee? If the answer had a hint of “yes” then I spent more time looking into it. Why was this so important to me? Because time is finite, and I wanted to increase my employees’ value to the company by enabling them to focus on work that matters. At the director level this adds up quickly: if you save each of the 50 engineers across your teams an hour a week, you’ve gained more than a full-time person’s worth of time without hiring anyone.
Here are some avenues you can use to jumpstart and accelerate the flywheel, and also prevent it from being slowed down. In this post I’ll focus on the ones that deal with your people and the information they work with: culture, metrics and visualizing business outcomes.
Culture as the bedrock #
I’m starting with culture because everything else in this post and the next one rests on it. Building and cultivating a workplace culture is one of the things that changes when you become a director. I think of culture as what people in your org do when nobody is telling them what to do. When you keep shaping it, people push the flywheel on their own without waiting for you, and you’ll see it in a few places:
- A succession plan is based mostly on existing talent in the org. Your managers won’t hold their roles forever, and neither will your senior engineers. In a culture that develops people, when someone moves on, someone else is already there to step into a managerial track or take on more IC responsibility. Because they already know the people, the systems and how things get done, they can hit the ground running in their new role, and you skip the months an outside hire would need to get there. Hiring their replacement means hiring for a lower position, allowing you to add new people to the org with a softer risk-reward balance.
- The cost of on-boarding new people is reduced. Here cost is taken to mean the effort by the new hires and that of current staff to facilitate the on-boarding. So, new hires are able to contribute sooner, and your current staff can put the time they would have spent on on-boarding into improving how the org works.
- People behave in an autonomous and expected manner, allowing more free time to experiment, think, debate, learn and improve. Less time is needed to repeatedly instruct people on what and how to do things. As a director this matters even more: you’re a layer or two away from most of your people, so you can’t instruct them directly even if you wanted to. When people know what’s expected of them, your intent reaches the teams you see less often.
- People can have a real stake in the success of the org they’re a part of, which has a positive correlation with talent retention. In turn, talent retention means less time spent hiring and training.
As a director, you don’t shape culture directly. You aren’t in most of the places where it actually happens: the stand-ups, the code reviews, the one-on-ones - your managers are there. So your org’s culture ends up being what you reward and what you tolerate in your managers, repeated across every team they lead. Your managers watch this closely: when you praise a manager whose team ships fast but burns out, or call out one who takes credit for their team’s work, the rest of them take note and adjust.
Metrics #
Every metric that you gather is something that you can use to understand how the system performs or how the business behaves. This understanding is fundamental in driving decision making, otherwise everyone is just playing roulette.
As a director this matters even more, because you’re further from the work than you’ve ever been. You no longer sit in your teams’ stand-ups or read their pull requests. Metrics let you see into teams you don’t manage directly, without going around your managers.
You might have stepped into an org that collects few metrics, if at all. There’s an infinite set of metrics you can gather, but you don’t have infinite resources to analyze them. You’ll have to pick out some key metrics that make sense to the org, and grow from there.
Granted, today you get a ton of performance metrics from any cloud vendor of IaaS/PaaS/SaaS. There are also quite a few observability vendors in the market, that with very little effort can orchestrate the collection of even more metrics on top of the vendor. That will go a long way in answering the question how does the system perform? Answering how does the business behave requires more effort on your end because every business is different and cares about different things. In practice, system metrics are mostly your managers’ and their teams’ ground: they’re closest to the systems, and they should own them. Business metrics are where you come in: you’re the one closest to the business, so defining “how the business behaves” for your org is largely your job.
Some points to take into consideration:
- Be wary of stepping into an analysis paralysis, in which there’s so much information that you’re unable to make any decision. No decision is a kind of decision.
- Although you can collect a lot of metrics, it might be wise to exercise prudence in which metrics you wish to monitor, as monitoring too many metrics tends to create an over-fitting problem in the org which draws it away from what matters.
- Numbers win many arguments. Not everyone can be right with 100% of their assumptions, so showing the numbers creates a factual baseline for a discussion. This matters most in the conversations you have as a director with peers and executives, for example headcount, budget and priorities. Say you have several cost reduction opportunities and capacity to take on only one or two, and everyone has a favorite. When you put the expected savings next to the effort each one takes, everyone starts the discussion from the same place.
Visualizing business outcomes #
Humans are visual creatures, and leveraging that is extremely powerful for you.
As a director you have something few others in the org have, which is access: you see the finance and revenue numbers, sit with business stakeholders, and hear how the company is actually doing, when most of your engineers don’t. Turning that into something they can see is part of being the link between the business and your teams.
What business outcomes do you show? Again, depends on your business, but the trivial answer is expenses and revenue. Expect this to evolve over time as the business context evolves.
When do you show them business outcomes? Whenever you want to, but I found recurring events to be a good rule of thumb. There is no better way to connect people to the company other than repeatedly showing them how their actual work makes a difference.
How do you show the business outcomes? There are endless options, and today’s AI workplace tools could help you achieve very good artifacts very fast. You can start from a small table of data, and the tool will draw up all kinds of drafts, and you’ll evolve the product until you get a clear result you can show. You can send recurring messages or emails, or display the results in your workplace area. Whatever makes sense to you.
Once you’ve done this a few times, encourage your managers to do the same at the team level, connecting their team’s work to the numbers you share, and from that point on, this part of the flywheel turns without you.
In the next post I’ll go over four more avenues, this time around how your org builds and makes decisions: dogfooding, automated tests, written documents and new technologies.