The writing in this post is my continued developed OKR narrative. Personal viewpoints, thoughts, and experiences.
Retention is not something we fix in 3 months
We set goals on a 3-month basis. It's enough time to get meaningful stuff done and it fits nicely with 4 four sets of goals during a calendar year.
But don't expect to significantly improve our big metrics during one OKR cycle.
Big, foundational, lagging metrics like "percentage of new customers that come back within 90 days" do not qualify for key result metrics.
Overall, there are two things wrong with having retention as a key result metric:
It's a lagging metric with long lead times
It focuses on OUR problems, not our customers'
And even though we're running a meaningful company with hundreds of thousands of customers, detecting a 5% improvement in something like retention is... difficult.
We'll likely make a lot of different efforts to improve it. To isolate exactly what change will make a difference will be... difficult.
Each change would need to happen under clinical conditions with perfectly isolated customer segments that weren't exposed to all our changes.
Things get complicated...
But here's another way.
What if we ... interact with customers, learn about their problems, and connect them to the company outcomes we're trying to achieve?
From there, we can hypothesize solutions and use leading metrics to measure progress.
Sounds easy, right?
A few examples:
Percentage of customers that receive their goods on time --> Retention
Percentage of customers without any customer service tickets --> Retention
Page load time --> Conversion rate
All of these leading metrics are much easier to measure. They're also focused on customer problems. This makes them a lot more motivating to work on.
We can approximate their importance based on how often customers complain about them. Delivery time doesn't matter? Then customers probably don't complain about them (or we've nailed it already).
Browsing the "problem space"
Our 3-year goal is to significantly improve our retention rates.
Great, then let's investigate what's causing people to churn. There are tons of ways to do so, but to name a few:
Go sit in Customer Support for a day or two and see what customers are contacting you about
Collect NPS or survey responses from customers and reply to them with follow-up questions
Reach out to churned customers to try and understand what made them churn
Based on these activities, we'll enter the "problem space" and learn so much about our customers. Now, go discuss all the customer problems you've observed with your team.
You might end up with problems like:
"Customers can't see everything clearly on their phone"
"Customers are unsure what they'll pay for shipping"
"Customers don't know when to expect their order to be delivered"
This was the hard part. Phew...
The easy part is now figuring out how to solve their problems, and make sure you're both qualitatively and quantitatively validating that you are, in fact, solving them.
What if our dashboard was leaked to our customers?
I heard this great phrase from Jeff Weinstein at Stripe:
... if you accidentally leaked your dashboard to your customers, they'd be extatic to know that that was what you were working on the whole time...
I now come back to this phrase whenever I think about metrics.
Is the metric focused on what the customer wants, or what the company wants? Customer problem or company problem?
The company outcome we want is improved retention, but the customer doesn't care about that.
They want a clean mobile site, a clearly stated shipping price, and a precise delivery estimate. So let's fix these.
Then, in the next braindump, we can discuss how to make sure that solving them also moves the business forward 👋
/Lasse