What we cover in Episode 27
CreditKarma is a hyper-growth company. When Matt joined 3.5 years ago, he had eight engineers, today that number has grown to over 100.
Matt helped to create a culture of growth at the company.
During his time the growth function went through many different structures. Today it's decentralized across teams.
In this episode of the GrowthTLDR podcast, we talk.
- Why decentralizing growth can get you a better scale, and how to do that
- How you can create a culture of growth at your company
- How to build empathy between teams (e.g., marketing and engineering)
- What common mistakes do companies make when building a growth team
1. Why decentralizing growth can get you better scale
Credit Karma's growth team has gone through a couple of different structures, which is typical for such fast-growing companies.
"I think we've had four different structures for our growth team; it's varied a lot over time."
One thing we've noticed from speaking with product-led companies is a growing trend of decentralizing growth.
Instead of having a central team who owns all things growth, you have a centralized team who creates best practices and tools so other teams can adopt a growth-led approach for their areas.
"The primary core of our growth team today is what we call growth services or product services. They mostly build tools and platforms for other teams to execute growthy tasks and then each of our orgs or business units at this point has their learning and optimization team.
And those teams leverage that platform, or those platforms and tools, to do growth in those specific areas, and we don't have a centralized growth team anymore."
For example, the centralized growth team could build an internal email platform that allowed marketers to email their userbase, and run A/B tests and pull reports.
Why build not buy? Casey Winters has a great article on why this might happen more.
In this case, the growth marketers are leveraging the tools/platform to run their experiments.
2. How to create a culture of growth
When setting up a growth team, you're asking teams to both think and work in ways that are different from what they would typically be asked to do.
For example, as an engineer on a growth team, you have to get comfortable with throwing out a bunch of code you write, and it not being seen as wasted effort.
"You have to tell engineers to write a lot of versions of the same code, and then throw 90% of it out, and if you wanna enrage a bunch of engineers, that's how you do it.
Saying, "Hey we're gonna throw out 90% of the code you write", well that doesn't engender a lot of trust, so when you start you need to hit those cultural things head on."
Here are three ingredients that have worked for Matt & Credit Karma in creating a culture of growth:
- It all starts with having the right leaders in place who can work together to instill that culture across teams. You need leaders who are comfortable with collaboration and shared ownership of metrics.
"You need to have a group of people that are willing to work together and collaborate on metrics. You need people who can drive change across the organization."
- You need to invest in the onboarding and coaching of new people onto the growth team.
"You have to be willing to teach people, and you invest a lot of time in saying, "Here's what we wanna do and why we wanna do it," and you have to reinforce that teaching over and over and over again."
- Investing in talent
Matt is another example of a growth engineering leader who's had a lot of success hiring in junior talent and coaching them to be impactful members of the team. We heard the same thing from Pinterest.
"We've had extraordinary success bringing on more junior folks and teaching them how we want them to do it, so they aren't already disdainful of the way that we're proposing. They come in, and they get excited and then, as you teach them and you reinforce that teaching, and you reward that teaching."
3. Building empathy between teams
When teams begin to work together, you have to build awareness of the differences in how they approach problems.
One of the best examples of this is marketing often lead with a solution, whereas, engineering like to lead with a problem.
Marketers want to think offensive; they want to go after metrics and do things that will move the needle as quickly as possible.
Engineers also love moving metrics, but they're more aware of the potential repercussions of implementing something, they're thinking more about defense.
"You can get a conflict when one group is thinking about getting the work done as quickly as possible so they can start to see an impact on their numbers. And the other group is thinking, well, if it doesn't wake me up at midnight, and I make sure that I test it correctly, it can have net value to my life."
For example, there are hidden costs of doing growth that marketers may not take into account:
All great growth teams want to run a high velocity of experiments, but they come with costs.
"Testing is a big cost, you have to do it, and the reason you have to do it is that there's a variety of interaction effects that can happen when you change something. So maybe you add a new feature that allows you to rank different campaigns and score their projection. Well, if you're running 500 campaigns a day in email, or marketing, if each of those things slows the system down just a little bit, what happens if it takes an extra six hours to run?"
Another common cost of doing growth is maintenance.
When prioritizing experiments ongoing maintenance is something you need to factor in when determining its cost.
"Another big cost is long term maintenance. I think one of the moments when you start to become a senior engineer happens after you're debugging some really hard problem, and you look at some piece of code and you say, "What idiot wrote this code", and then you go back and look at Git, or Subversion, or CPS, whatever you grew into your engineering skillset, and you said, "Uh-oh, it was me. I was that idiot.
I wrote this code two years ago, what was I thinking?", and a lot of what happens, especially in the growth world, is you write these experiments, and then you figure out which version won, and then you don't go back and clean it up afterward."
Growth can be a lot of experiments, a lot of change, and it can become difficult to maintain at some point, creating a considerable amount of complexity for how work gets done.
When asking engineering to do something, marketers should keep the above in mind to build empathy on why things never "just take five minutes."
"So if you're going, "Hey, I just need you to adjust this button and change the color to red", and then when it takes you four days to do it because the test environment was down or something and you have this challenge, or the brand team didn't like that color and it wasn't on pallette, and it takes you four days, someone's like, "Well I just asked you to do this thing, why'd it take you so long?", god that sucks."
The podcast provides a more in-depth look at these topics, so if you enjoyed reading the above, please do give it a listen.
And until next time,
Subscribe Rate and Review
>> Google Play
Follow us on Social Media