It’s difficult not to love a bit of fancy terminology that makes us sound clever. What’s even better? Getting to the bottom of what it means to unlock the value really on offer.
Today, let’s do that with cohort analysis.
Lots of people and blogs drop it into technical analysis articles including, David Skok, KISSMetrics and MixPanel. So it must be important.
It’s a method of looking at datasets from particular time-frames to find valuable, actionable insights on how groups of users behave and interact with your product.
Looking at smaller chunks of data this way tells you the real story, breaking apart vanity metrics and unrealistic “growth” that looks so sexy when you plot it into a line graph.
Analyzing your data in these blocks (or “cohorts”) helps you get to the bottom of your actual growth in terms of customer lifetime value (CLTV), monthly recurring revenue (MRR), churn and other important metrics.
Let’s get into it.
What is cohort analysis and how does it work?
A decent definition of cohort analysis might look something like the following.
“A selection of people from a predetermined timeframe and who share a common characteristic”
In the case of a SaaS owner, that “selection of people” can be visitors, users, customers, leads, or anything else that’s relevant to your product.
One benefit of using groups this way is that only a relatively small quantity of data is needed before the numbers are calculated for you automatically. You extrapolate or project the numbers based on a common characteristic shared with users in a sample set to show you what it might look like on a larger scale.
The main benefit though is that cohort analysis lets you analyze data more accurately to realistically build a picture of how each month’s metrics develop over time.
Your SaaS is constantly evolving, so these time-frames are super important. It’s a method to see how well your business is actually doing as while you make tweaks to try and improve it.
You can enter the numbers highlighted in the blue text at the start and the spreadsheet will calculate the data for you.
As Janz states on his blog:
“It’s almost impossible to get a really good understanding of a service’s usage without looking at activity and retention numbers on a cohort-by-cohort basis.”
So yes, it’s worth getting your head into all those numbers.
We’ll use the spreadsheet in this article as a demonstration of how to use it for your introduction to cohort analysis.
An example of cohorts in action
Yup, that’s a whole lot of numbers when you open it up.
Let’s go through them together and start with the customer cohort example. Here’s the lowdown.
As mentioned, the blue is where you enter in the raw data. The first table is how many new signups you have each month, showing how many stayed with you for each subsequent month.
If your data goes back further, simply extend the table accordingly. All of the other charts (after this blue-text chart, A1) are calculated automatically. For the sake of our examples, we’ll use the default numbers already entered when you downloaded the template.
As you can see in chart B1, the metric of “customer lifetime months” is used. It simply means the number of months elapsed since a customer originally signed up. Here’s what the annotations mean.
- In this cohort analysis example, we’re analyzing monthly customer cohorts. So in the January cohorts, our SaaS generated 80 new customers
- The numbers in the column of blue digits extending below represent the number of new users each subsequent month. That means each number is a different cohort
- The chart shows us how many customers were retained each passing month, meaning that of the 105 customers that signed up in March, 94 were still using the product by June
- The complete number of users each month is tallied up at the bottom of the chart. So the total number of users active in June was 580
- Below that, we can see the percentage of customer retention. So in the 5th lifetime month of the March customer cohort, 85.71% of users were still active
- Finally, at the bottom of the green-filled chart we can see that the average percentage of users who were still active seven months after they signed up
What else can we learn from the cohorts?
When users sign up to use your product, they won’t stay with you indefinitely. It would be nice if they did, but the best we can expect is for the number of retained customers to go down as slowly as possible.
Ideally, you’d see an improvement in retention with newer cohorts compared with older cohorts. Because you’re likely improving your onboarding flow and other aspects of your SaaS. And adding new features should improve your product which should improve the retention of each cohort, even if you see users drop off slowly over the long term as above.
As you can see in the below chart, that’s not the case. January and February enjoyed a little over 93% retention by customer lifetime month 2. But August is at 89.4% by the same stage.
Clearly, something you did in October didn’t work. If you know that information, you can investigate it rather than sit around for a few months thinking everything is fine because your number of active users appears to be growing steadily overall.
If we look at B2 table below, it shows us a little more clearly what our percentage of churn is from month to month without having to do the mental arithmetic by looking at the previous month. It just makes a little easier.
It’s also useful to look at churn over time to find out where it peaks and whether or not it’s stabilizing in the medium to long-term.
For example, a peak in month 2-3 may indicate that some people don’t quite get what they expected, so more of them are dropping off. But if the number stabilizes shortly thereafter, you’re doing okay.
Remember that according to David Skok, the average acceptable churn rate for a SaaS business is between 2% and 3%. So, if you’re seeing significantly more than this, you know you’ve got work to do.
This spreadsheet also includes the churn rate relative to the previous month, in addition to the baseline number as illustrated above.
This can be especially useful when you’re adding new features, or doing an interface overhaul, for example. Because you can see the percentage of customers that churn as a result of any changes that you made.
Let’s look at an MRR cohort example
As mentioned, when you look at customer cohorts, it’s natural to see a slow decline over time.
That’s ok, provided the churn cohorts indicate that the rate of decline isn’t too fast.
But when it comes to MRR, it’s important you see increases over time. For example if you offer a basic service package for $24.99 and one for $99 per month (maybe some kind of Pro version) cohorts can generate more cash even if a few more users drop off than the previous month.
Breaking down the MRR into cohorts this way also gives you insights as to how well you’re doing with your up-selling.
Similarly, the chart above might indicate that a new strategy implemented during July or August had proved especially successful in increasing the monthly MRR by around 7%.
Armed with that information, you can see what it was and then make sure you implement the offer (or another strategy) again in the future.
Another win for you cohort analysis.
Customer acquisition cost (CAC) cohorts
The final section of the spreadsheet is set up to help you find the point your monthly cohort actually becomes profitable, based on the churn rates and the monthly recurring revenue.
It’s important to note here that you have to realistically enter all the data for a genuine idea of when your cohorts start making money. And that includes staff salaries to marketing costs, and everything in between.
And this brings us down to our final point which is about keeping your data clean
Keeping your data clean
If your product goes through several changes over the course of a few weeks, it’s important to break down your cohorts accordingly.
Because, for example, if you start comparing the August cohort with the March cohort, but the two are actually using completely different UI’s or different onboarding flows, you have to keep that in mind when you’re looking at the data.
This will apply differently to different product owners, but it’s worth mentioning. Remember, there are seriously valuable insights on the table, so stay on top of this organizational aspect for the best results.
We hope this article gave you a decent understanding of how cohort analysis works and why it’s so important.
It’s a pretty intimidating subject at first, but after you enter a few of those blue numbers in and see how your data looks, it can actually be pretty fun.
Want to stay up to date with the latest in SaaS KPI boosting strategies?
Make sure you sign up to myTips blog to get the good stuff before anyone else.