Building a data-informed company – interview with CEO of Feather Insurance. Part 2
In the first part of the interview we roughly sketched out a way to a data-informed company:
Step 0. Focus on getting real customers in. It might take a whole year and data won’t help here – you must talk to your customers so they can literally tell you their problems. Data is just a way to scale this approach.
Step 1. Once you get customers using your product without your assistance, start looking at basic metrics – rates (signup, purchase) and funnels (drop out rates in the onboarding, etc).
Interview
Rob: Let’s maybe decipher the last step a bit more.
Anatoli: Indeed, it might seem like conversion rates and funnels are too simple to bring any significant impact. It’s just a percentage, right?
I must point out that real insights come from asking great questions. It’s like having a magnifying glass and then trying to zoom in many places to understand the details. 
Your example about drop out rates in an onboarding is great: looking at the overall signup or activation rate gives us some idea of what’s going on (say, it’s 5%). A real insight comes from knowing that 99% of people churn at a certain onboarding step – maybe your copy is misleading or there’s a critical bug that doesn’t allow people to go further.
Rob: Exactly! It’s all about getting the basic stuff right.
Even now as a company we try to be good at basics everywhere. If you nail simple stuff it already makes you better than everyone else.
Anatoli: Rob, it’s so awesome that we talk about it.
It might feel as it’s 2021 and the history is over – everything in the world is done, everything is perfect and all markets are full, no niches left.
It’s so not true, I believe it’s quite the opposite.
Rob: I used to play ultimate frisbee, I’m several times European champion and vice world champion, I also coached a lot of national teams. 
There’re a lot of parallels between coaching frisbee and running a business. One of them is getting the basics right.
People could do tons of complex strategies, but if you can’t do basics – throw-run-catch you’ll lose 100%.
Anatoli: Wow, it’s so freaking cool! This sports metaphor really hits me, it’s genius.
I know companies with millions of users and they’re still working on getting their basics right, no need for complex machine learning or anything.
Even if you feel like you’re on top of the world, try to challenge the status quo. Try asking what are the simple things that will move the needle – I bet you’ll find a lot of them.
Rob: Exactly, and I love how SQL Habit teaches precisely that. 
I remember after every chapter I’d spend a month building it at Feather, getting the basics right.
Here comes the thing that frustrated me the most, wait for it – I shared a screenshot with our CTO of a table from the SQL Habit dataset, it was the marketing spends table. I told him I need this table in our database to dig into our marketing.
I needed this table so I can continue with the course – I literally build our data warehouse as I go through the course.
Anatoli: I’m freaking out right now haha, I just realized that the actual database used in SQL Habit inspires your next steps! It makes me feel very proud. 
Rob: Exactly, I need our data warehouse to have the same structure as in SQL Habit.
Anatoli: That’s really awesome, I guess we can put this as a Step 2 – building a data warehouse, bringing all the data in.
Now let’s go broader and talk about how data is used at Feather. Who uses data internally and how? Do all employees have access to it? Do you encourage them to use research data?
Rob: I believe as a company you can’t build an awesome product without data transparency.
If you’re trying to hide stuff – you’ll never build a great product.
All important numbers are accessible to everyone in the company, it also helps with showing where we want to go.
For example, an important metric for us is the number of protected customers (yep, it’s not a profit we’re making on them). I want everyone to know this number, to see where we’re at.
People need to understand these metrics because it also shows them how the company works.
Anatoli: I absolutely agree, I also arrived at the same thought early in my career.
About 10y ago I started working at this big company (400+ people) in Moscow and everything was hidden from employees – everyone was kept in the dark about how much money we were doing, no one knew our metrics.
I remember I came in on the first day, they showed me my desk in our open space, handed me a laptop. The first thing I did was installing an SQL client and building a report of the month-to-month revenue.
I was like “Yo folks, I think we’re growing pretty awesome, look at this!”. All engineers were standing around my desk and then the CTO ran towards us screaming “Anatoli, WTF, where did you get this data?! You keep it quiet!” 
I believe hiding data is like lying – building a house of cards that is meant to fall apart. It also drains so much energy from you.
Rob: Exactly.
If you want to build a great company you should never hide your metrics.
Afterthought
So far we’ve identified these 4 steps on a way to a data-informed company:
Step 0. Focus on getting real customers in. Measure their problems by talking to them.
Step 1. Start calculating simple metrics, querying your production database is already enough.
Step 2. Start building a data warehouse and collecting multiple data sources in one place. That way we can analyze the whole user journey.
Step 3. Make sure everyone in the company has access to data. #datatransparency
In the next part, we’ll talk about misconceptions about learning Data Analysis and SQL.
In the meantime, check out open positions at Feather Insurance. I personally recommend it because of its amazing product, company culture and data transparency.
If you want to start building a data-informed company – check out SQL Habit for teams. You’ll learn how to use data from a story of a growing startup company.