The 10 SaaS Metrics You Didn't Know You Needed
Download Now
The Role Forward

Steve Groccia on Keys to Effective Cohort Analysis

Steve Groccia explains why cohort analysis is so valuable, why it's challenging for any finance team, and how to do it effectively for strategic insights.

Subscribe to the podcast

The Role Forward
Steve Groccia on Keys to Effective Cohort Analysis
Never miss new content

Subscribe to the newsletter to keep up with the latest strategic finance content.


Episode Summary

The contemporary business world depends on in-depth and high-quality data analysis. But it seems like many departments don’t have adequate time or tools to focus on data.

In a study conducted by Mosaic, only 14% of surveyed finance leaders said they used cohort analysis. Therefore, it is critical to determine the reasons behind this small percentage and offer solutions.

In this episode of The Role Forward, host Joe Michalowski welcomes Steve Groccia, the Head of Customer Operation at Mosaic. Steve and Joe discuss the reasons finance leaders don’t use cohort analysis. Steve also explains the difference between segment-based and time-based cohort analysis and the steps in the process.

Watch the Full Video

Featured Guest

steve groccia headshot

Steve Groccia

Head of Customer Operations

Linkedin link

Steve helps finance teams successfully implement and adopt the platform so they can maximize all the benefits that Mosaic has to offer. Before joining us, Steve managed and then directed finance at Fundbox.

Key Themes from the Episode
  • Cohort analysis is critical to getting to the "why" behind your data.
  • Cohort analysis has to be digestible to everyone in the business, otherwise it won't deliver real value.
  • Customer success isn't just a department in the business — it's a core responsibility for everyone in an org.

Episode Highlights from Steve Groccia

4:10 — Reasons Not Many Finance Leaders Use Cohort Analysis to Look at Revenue Retention

”If I think about it, what are the challenges? Sometimes, just the fundamentals of having time and resources to do these deeper analyses.

I remember my first couple of years at my last company just doing everything from closing the books to payroll. […] And so, there are always these big burning fires, and you never have the time to do this deeper level analysis.

[…] Time becomes a challenge. Getting data in a structured format, even if you had the time — it takes a lot of time to set up the data to make it usable and make it work.”

8:22 — Segment-based Cohorting versus Time-based Cohorting

”I think about two different flavors of cohorting. Let’s call it segment-based cohorting and time-based cohorting.

And so segment-based — we had one investor when we were going through debt financing at Funbox that wanted what they called a layer cake chart. […] You wanna look at your ARR over time. And so, if the business is doing well, that ARR over time is going up fast. But then, segment that ARR by your different cohorts, and each layer of that ARR stack is different […]; it looks like a colorful layer cake. It’s very insightful because it tells a lot about how much of your revenue is from existing customers.

You learn a lot by segmenting your different data by cohorts. And then, you’re trying to find that balance; you want a lot of layers because that means you have a ton of existing customers using your product, but you still want that top layer to be high because you want to keep growing and keep growing fast.

[…] I think the other way — where we spend a lot more time cohorting at a deeper level — is time-based. And that’s to understand your behavior, the behavior of something over time. […] It’s meant to analyze behavior, to find insights, and then see what behavior works and what does not work.”

15:40 — End-to-End Cohort Analysis

“Cohorting was the underlying methodology for us [Funbox] to plan the business. […] So the complexity there was, we’d have to cohort everything. We’d have 6, 7, 8 types of cohorts. So that hopefully paints a picture of how wide this was.

And then the actual fundamentals of doing it; I’d do the super manual way first. And then, we upgraded to a BI tool and hired a team of four people to structure our data. But before that, it was taking raw data. I have it in a massive spreadsheet and I look at the key fields I care about.

If you’re comparing cohorts over time, what is the starting and ending point? Usually, the starting point is when a customer first becomes a customer. I’m adding additional columns and rows to my Excel sheet. I’m doing all these formulas and making sure everything works appropriately. And that’s step one — to take some data and restructure it, so it’s usable.

And then step two is to have another sheet and set up what the cohort looks like and how it works.

And then step three depends on if you want to take an average, a percentage of the initial balance, or a running sum.

That was to get your starting point. And then the second point was what’s going on with these cohorts? […] And our assumptions are driven off these master cohorts. And so we’re bucketing half of our potential new users into what we think is going to be their behavior over time.

It gets a little scary when you’re forecasting, but it’s also very insightful like, ‘Hey, marketing team, retention team. It is what we’re seeing.’ If we could move the needle a little bit on behavior, it’s gonna have a massive impact on our customer base. And so, that’s the fun part of finance. You find some of that and present it to the rest of the organization to find actionable insights.”

Full Transcript