During his time at Coursera, their product and engineering team grew from 50 to 160 people. As CPO, Tom had to ensure that the team stayed motivated and continued to ship products at a rapid clip. In this AMA, he shares how he did that.
Tom Willerer joined Venrock a year ago, to focus on consumer investments, as well as some scalable enterprise SaaS investments.
Prior to that, he was Chief Product Officer at Coursera, and helped it scale from 50 people to 300, from one business line to three, from $1,000,000 in revenue to… $100,000,000. During his time there, his team grew from 10 people to 160, with all the associated challenges when you transition from being an individual manager to a manager of managers, to a leader of large teams.
Before joining Coursera, Tom has spent seven years working in advertising and joined Netflix at the time they were still shipping DVDs around the US. When he left the company as VP of Product, Netflix was global streaming and doing originals.
How can we learn from his 17 years of experience in companies that benefitted from spectacular growth?
Our mission, at Plato has always been to try to learn from the best engineering and product leaders. To do so, we match product and tech managers with top-of-class mentors, so that they can become great leaders.
Fortunately, Tom sat down with a few mentees from the Plato community for an AMA. You can watch the full video here, or scroll through a few of our favorite questions and answers below, sorted by topic.
A few mentees asked Tom what he thought about vision and how he helped his teams understand it and act through it.
A: My journey on vision was somewhat painful. I came from Netflix, and Netflix is famous for being extremely focused. I never set out the 10 year vision for the company. It was always clear that we weren’t a DVD service, that we were going to work toward streaming, so in that sense the vision was there but it wasn’t explicit. We were much more of an experimental company- test and try things, see what works, and follow the data. I took that approach when I moved over to Coursera.
We built a large AB testing engine and were were doing 500 AB tests a year. There were some people on my team who would say that we needed a vision. For a while, I disagreed thinking “no one can tell the future. We have a plan for the next 6 months, let’s see what happens and create a direction out of the results.”
However, the team kept patiently prodding me, asking for a vision. Soon, I realized they were right. The vision isn’t about predicting the future, the vision is connecting the metrics and data to emotion and heart. What I’ve learned is that people are much more invested when their work is tied to a larger purpose and this comes through with a vision.
I set a vision for Coursera. This is a new skill. Before, I was providing a rational framework, but not as much of an emotional framework that people could connect with. Most of what we do in life is much more emotional than rational. My process for creating a vision was, 1) pick a timeframe. Ex: 3 years, 5 years, 10 years. Each will look a little different. I decided to craft a 1 year time frame and a 3 year timeframe. The 3 year timeframe had much less fidelity than the 1 year timeframe. For 1 years you can get more specific. For 3 year, it was more general. The shortest time frame for setting a vision is probably around 6 months and the longest time frame should be around 5 years.
There is a strong temptation as a leader to set the vision on your own. I think that is the wrong way to go about it. I involved my direct staff and we iterated on a weekly basis. I also collaborated with my peers on an executive level to get feedback. Collaborating with other people was a way to get better ideas out of it. Additionally, making it collaborative gives people a way to be heard and gives them a voice. This is helpful because the vision has to be executed by the company. If the company doesn’t have buy in, they won’t be as invested in moving toward the end goal. By making it collaborative you get buy in even before you roll out the vision.
I tried not to be super specific through this vision setting process. For Coursera, we had different users, a learner and an instructor. The framework was, “for those two users, how do we want the experience to be different in a year?” I worked with one of the designers who made sketches for how the experience would look different. We told a visual story instead of a roadmap. It wasn’t based on features and metrics.
Instead of delivering a presentation on my own to roll out the vision to the company, I involved the whole team. There was even an element of surprise. We had different characters in the story and showed the journey and how it changed over time. This helped because months later people referenced characters from this presentation in their own presentation. It helped the vision to get adopted.
A: I talked to the CEO and told him that my team is asking for this. Since the company is manifested through the product, it makes sense for me as the product leader to run with this, but are you okay with that. Luckily for me, my CEO at the time liked the idea. We determined how and when he wanted to be updated and involved. Once we nailed that down, we told the rest of the executive team and involved them as well.
It doesn’t always work like that. Sometimes it may be the CEO who wants to set the vision. The main point is that setting expectations ahead of time with your boss and your peers is crucial. Especially when it comes to getting buy in and involvement from peers. In my case, a lot of this was lead by my team but this was because the rest of the executive team delegated to us and felt like they still had the opportunity for their opinions to be heard.
A: I think they are two different things. The mission of Coursera was to make the world’s best education available to everyone over the world. The vision was more specific. At the time, we were working on personalization in content and higher stakes learning. The vision was to make Coursera an amazing experience for our learners. However, the mission was stable and did not change year to year.
Netflix’s mission for a long time was: movie enjoyment made easy. This really didn’t change except for maybe to broaden into “entertainment made easy” since now there is a focus on TV shows as well. The vision, however, changed. At the beginning of streaming, we were against originals. It started as a tiny experiment that worked; then, originals became a cornerstone of the vision and strategy. However, the mission never changed, it was still about entertaining people.
A: We did not do the one year planning every month. That was a once a year update. We had our best guess on vision in December and January. We took the whole vision in broke it into 3 month increments and adapted as we went. We did not feel beholden to every single piece of the vision. There were parts that worked and parts that didn’t. We didn’t chase those just because they were originally part of the vision. Instead, we updated it. We met at the halfway point of the year to talk about what we had learned in the past 6 months and how this changes our vision and roadmap from what we originally anticipated.
Tom shared his experience on how to set objectives and key results to ensure optimal delivery from engineers or product people.
A: We tracked a few things. We tracked an aspirational metric that tracked if we were moving toward our mission. This was essentially: number of courses completed and number of people who had completed a course. These were longer term metrics that couldn’t really drive the business, but they were motivating.
We also tracked revenue as a direct translation into if we had the type of impact we wanted. We also tracked employee pulse surveys as an operational metric. These drove the business week to week.
Next we tracked strategic initiatives. For instance one of our goals was to get full online degree programs onto Coursera. We measured this by counting these.
In Netflix’s case, this was simple. We wanted to drive retention and increase acquisition. When they got into originals, goals shifted and we had to take more bets and spend more money.
A: No, we didn’t do too much of that. My hesitation with that is if people can see right through it. For example, it is not good to set artificial deadlines. That will work one time. Then everyone will see through it and be angry with the arbitrary rush. I think there is use in having a target date but not a hard deadline, that could go wrong.
For bug fixing, you do have to have to find a fun way to get that done. One thing we tried was to set aside a period of time where knocking out bugs is your number one priority. It’s amazing how much satisfaction engineers got from that. PMs never wanted to do it because they were focused on the future. However, engineers wanted to maintain a good product. I would have to force my PMs to do that. I would also celebrate bug fixing and call them out in an unusual way because very few PMs will go out of their way to do it.
A: One of the most interesting experiments we did was related to quarterly planning. We found that even though teams had their own swim lanes, there was tons of overlap between teams so they needed to align. We experimented with blocking off a two hour chunk of time during the planning sessions. This was at a time where planning was far enough along that each PM knew their OKRs.
We brought all the product leaders and put them in a room with no agenda. I kicked it off and said this is the planning meeting- if you have an item that has another team as a dependency, and that other team doesn’t have that item as one of their OKRs then you have to delete it from yours.
This was a very interesting, non-obvious way to approach it and it was very bottom up approach. This worked well for us and we kept it around.
A few of our mentees asked Tom to explain how he managed to keep people motivated. Though there’s no miracle solution, some of his tips and insights are gold!
A: Hiring great people and giving them a lot of responsibility is key to keeping people motivated. For instance, let’s talk about product managers. I was reluctant to hire a large volume of product managers and instead I wanted to hire senior PMs who I could give tons of responsibility. This was for the exact reason you described. I didn’t want to break down the product so small that people felt like they were constantly bumping up against each other.
We organized ourselves into a growth team: 10 engineers, 1 PM, 2 designers, 1 analyst. A learning experience team with sub teams: 20 engineers, 2 PMs. A partner experience team (building tools for content providers). Netflix had a similar structure. In each of these sectors, you have dedicate PMs, engineers, and designers. There are squads with a purpose. They should each have a set of metrics, that do not overlap with other teams, that they are trying to optimize. That gave each of those teams enough autonomy to see a clear indicator of if what they are doing is successful. As a leader of these teams, I did not need to make decisions on a daily basis.
A: You can do an employee pulse survey, or just check in with the team. You will hear problems pretty quickly. When people are not motivated they will be dissatisfied with their job and they will tell you.
Secondly, i’d say the productivity of the team is telling. If the activity level plummets then you know the team is not motivated or they don’t know where they are going. If the team just doesn’t hit their metrics, that is a little bit trickier. It could be that the metrics were too high.
A: At Netflix, we had a continuous feedback cycle. If someone was doing something you had feedback on, positive or constructive, we just went and told them directly. We tried to be very continuous about it.
The other thing we did at Netflix was we separated out performance and compensation. Feedback should be about you improving, not about getting a raise. Compensation raises were all market based. That worked at Netflix.
There was a culture that was developed where people gave each other feedback from a standpoint of: how do I make all of us better. It was key that it didn’t seem like it was coming from a place to undermine one and other. Even though feedback was generally constant at Netflix, we did have a yearly 360. This was a way for employees to know exactly what they could approve on, rather than information just for their manager.
At Coursera we were more traditional. We used a rubric with different levels. We had 360s, and these were tied to compensation. I think either can work.
What you want to strive for is a culture where people trust each other and feedback isn’t about trying to undermine the other person. The feedback is about trying to improve and make the company better. If that’s the case you don’t need a yearly 360 review because that’s happening on a regular consistent basis. Now, we don’t live in the utopia. So it’s helpful to have that moment where you’re like, okay, I should probably think about the last six months and how I want to structure my feedback to these different people.
Generally, I am terrible at giving and receiving positive feedback. When someone gives me formal positive feedback, my instinct is to think they aren’t telling me all of the truth. I think to counteract this feeling, it’s better to give positive feedback in real time so it comes off as more authentic.
Some other interesting questions have been asked, see below!
A: At Netflix, we were exceptionally data driven. We did thousands of AB tests. I wanted to bring this to Coursera so our decisions were data backed and not just what I felt was right. This was important because it helped individuals and teams to be on board with new features and to get a black and white answer about why we are or are not moving forward with something. This is culture building. You are building a culture that is going after truth. AB testing is a mechanism that can reinforce that.
At Netflix, we also had a strategy meeting that happened once a week. PMs would bring forward the types of things that they were trying to do and just get feedback. We pulled that into Coursera.
Last, we implemented the OKR process and experimented with that.
A: One at Netflix- in 2010 we were very convinced that if we didn’t make Netflix a more social, viral product in regards to sharing on Facebook, someone else could come in and take our space by piggybacking off Facebook. We did a lot of experimentation to test this and ultimately found that this was not something our members cared for at all. They did not want to share all of their watch history via Facebook. This is something that seems intuitive now. We eventually rolled all this back. That was something that felt like we should go after for strategic reasons but it ended up being a disaster.
At Coursera, I mapped my Netflix experience to Coursera. At the time we had 500 asynchronous classes. These were pre-recorded so hypothetically could be watched at anytime. However, instructors did not want this. They wanted to be available to students and virtually run the course. But there was no reason they had to do it this way, why wasn’t it on demand? It turns out learning and entertainment are exceptionally different from each other. TV is much more addictive than learning.
When we made Coursera classes on demand so people could follow at their own pace, it removed all the urgency. It went from important and urgent to important but not urgent and thus would move down the priority list. Less people would finish courses.
We then moved to a model where each course had a start date and end date because this mattered for education where it does not for entertainment.
This started as a disaster but we ended up able to save it.
Plato helps engineering managers and PMs get mentorship from top tech leaders at companies like Google, Lyft, Netflix, and more.
Get a taste of this mentorship by joining a group mentoring session for free.