5 mins

By modeling the behavior of your engineers, managers can boost execution and raise the bar of their whole team.

Execution is one of the main factors that drive consistent engineering success, but raising execution standards can prove tricky for managers.

Behavior modeling is the practice of instilling certain behaviors through effective use of incentives in order to achieve a certain goal. Done right, it’s a powerful set of tools that can help to ensure consistently excellent execution. 

What kinds of behavior can be modeled?

As engineering leaders, we are constantly thinking of ways to maximize our teams’ impact: we may opt to adopt iterative and incremental development in order to improve prioritization and velocity, we incentivize risk-taking in order to enable more innovation, we promote inclusive behaviors in order to improve psychological safety and collaboration. These are some common examples of behavior modeling.

People’s behaviors are an externalization of their beliefs coupled with an instinctive reaction in response to a situation. Effective behavior modeling requires, first and foremost, the belief that changing behavior has benefits. As a result, it is fundamental that people understand and agree upon these benefits.

Humans are inherently averse to change, so top-down enforcement is generally met with resistance. This means that engineering leaders need to hold conversations with their team members and gain buy-in. When doing so, it’s important to clarify value propositions. 1:1 meetings are great spaces for these conversations, as they allow more two-way empathetic communication, where both parties can understand each other’s thoughts, reactions, and concerns.

It is also important to be transparent with intentions in these conversations, talking about the problem we are trying to solve with sincerity. Here’s a rule of thumb: if we as leaders show resistance to being transparent, we should reconsider if what we are proposing is the right thing to do.

One pitfall to watch out for is being over-prescriptive in behavior modeling. This type of micromanagement is often harmful to creative roles, aside from being detrimental to morale.

Humans are incentive-driven 

While working to alter or modify certain beliefs is fundamental to behavioral changes, these typically don’t happen without a driving force. This is where incentives come into play.

Good incentives are ones that yield long-lasting effects. For example, recognition tends to be very effective for modeling behavior, whereas monetary incentives like compensation are better suited for retention.

Recognition can come in many forms. Centralized ways of recognition, such as career ladders, performance reviews, and promotions, are useful for normalizing expected behaviors throughout the organization. For example, at Duolingo, we have “ability to receive feedback productively” written as an expectation in the career ladder for all roles at all levels. For us, this behavior is fundamental for long-term success at the company.

Decentralized recognition is also important because it allows for more frequent reinforcement than more formal methods. At Duolingo, we have things like peer bonuses and internal communication channels where the purpose is for people to give praise or express gratitude to others. In meetings, we reserve time at the end for shoutouts.

One personal example of how positive reinforcement may look in practice was a project where upstream dependencies were blocking the work of engineers. Naturally, this affected the deadline and so the engineers rallied to figure out ways to meet this timeline in the face of this bottle-neck situation.

Expectation-setting is often the go-to approach by leadership in cases like this, but it normally ends up causing some level of frustration. Instead, we opted to praise the behavior of moving forward productively in the face of ambiguity. This not only led to more durable, resilient behaviors but also sparked conversations on the value of navigating ambiguity.

The case against reprimanding

Negative feedback, which is different from constructive feedback, causes what is called “amygdala hijack”. This is a physiological response to a perceived external threat, leading the primitive parts of our brain to overtake the rational parts. This means that our ability to think critically decreases, together with our ability to internalize feedback, which is counterproductive when the goal is to encourage certain types of behavior.

In fact, negative instinctive reactions are an enemy to producing productive behaviors. Strategies to combat this enemy include nurturing an environment that minimizes situations where amygdala hijacks occur, which can be achieved via an inclusive, collaborative culture.

Examples provide visibility and specificity

Incentives alone are often not enough to reach desired outcomes. This is because they do not show the specifics of the behavior. For example, “move faster” can mean very different things for different people depending on their backgrounds. For some, it can mean building a barebone minimum viable product (MVP), whereas for others it can mean reaching decisions quickly.

It is therefore important that we set concrete examples of expected behaviors. Effective examples are ones that people can identify with, learn lessons from, and apply the actionable items.

Needless to say, leading by example is crucial. A leader cannot be effective at instilling behaviors if they themselves are not doing them. Advocacy requires inspiring a deep belief in the reasons and the benefits of actions, so leaders need to position themselves not only as champions but as embodiment of the principle they’re trying to push.

Getting started

If this seems like a daunting task, we have good news: if you lead a group of people, you are likely already engaging in behavior modeling. Most modeling happens organically; we identify a problem or carve out an opportunity, develop solutions, and then we act on instilling behaviors that enable our organizations to further increase their success.

When generating solutions, it’s helpful to leverage people in our support networks. Having sounding boards helps validate the framing of the message and what sentiments it might spark. Personally, I like to involve my own direct reports early on in these validation exercises. I hold candid and kind conversations with them to share where I am coming from and what I am thinking and to hear their thoughts, reactions, and concerns. This way, I have a collective approach to instilling behaviors, which helps build a high sense of ownership and trust.

With all of this said, I should note that effectively shaping behavior is not easy and involves intentionality. We need to gather data, ask ourselves questions, and constantly challenge our assumptions on what we believe are productive behaviors. There is a lot of effort involved and a lot of trial and error, but the reward of having a high-performing team awaits at the end.