George Kapernaros | YOCTO Agency

WEBVTT
Kind: captions
Language: en

00:00:00.240 --> 00:00:05.520
What is cohort analysis? If you've ever opened 
Shopify, if you've ever opened Recharts,  

00:00:05.520 --> 00:00:11.360
Key or Loop, any of the subscription platforms and 
seen this colorful table that has like percentages  

00:00:11.360 --> 00:00:16.800
and numbers and you have no idea what it means, 
then this video is for you. Because I'm not just  

00:00:16.800 --> 00:00:21.680
going to tell you the three ways. There's three 
ways to read a cohort chart. I'm also going to  

00:00:21.680 --> 00:00:28.560
show you step by step what they mean and what you 
can do. And on top of this, I'm going to walk you  

00:00:28.560 --> 00:00:34.080
through two common scenarios that are extremely 
relevant for an e-commerce brand. What to do if  

00:00:34.080 --> 00:00:40.480
your retention is poor at the beginning of the 
journey, meaning people cancel soon after they  

00:00:40.480 --> 00:00:47.440
subscribe to your products. Take the discount and 
run type of behavior. And what to do if your case  

00:00:47.440 --> 00:00:53.840
is slightly more complex. In other words, how 
can you identify when the true biggest leak in  

00:00:53.840 --> 00:00:59.200
your attention system is? As you may know, my 
name is George Scalanaro San Yokto. We are one  

00:00:59.200 --> 00:01:03.600
of the world's best life cycle marketing agencies 
and it would really mean the world to me if you  

00:01:03.600 --> 00:01:08.720
could subscribe to this YouTube channel. I've just 
started posting videos and your support is really  

00:01:08.720 --> 00:01:14.400
going to mean a lot to me. Let's get started. 
Okay, let's start with the very basics. What is  

00:01:14.400 --> 00:01:21.040
a cohort? A cohort is simply a group of people who 
started in the same month. And in e-commerce when  

00:01:21.040 --> 00:01:26.640
we say started in the same month we usually mean 
who placed their first order in the same month.  

00:01:26.640 --> 00:01:32.400
So it's new customers in a specific time window 
which is typically monthly. And the reason why  

00:01:32.400 --> 00:01:37.840
I say typically monthly and in incommerce we 
refer to customers is that the cohort analysis  

00:01:37.840 --> 00:01:42.000
can apply to other context as well and have 
different meanings. It can be different time  

00:01:42.000 --> 00:01:48.320
windows and can be different metrics. As you can 
see in this example that I have open right here.  

00:01:48.320 --> 00:01:56.640
What we use as our metric is of course net sales 
but it could be something else and just to show  

00:01:56.640 --> 00:02:04.560
here it could be cumulative sales and then it 
could be I don't know AOV it could be retention  

00:02:04.560 --> 00:02:09.120
this which is what people typically refer to 
when they look at customer cohorts but all of  

00:02:09.120 --> 00:02:16.880
those metrics are simply different ways to look 
at people grouped by a specific time interval yeah  

00:02:16.880 --> 00:02:22.640
month in this case who first entered our brand's 
universe at the beginning of that interval. So,  

00:02:22.640 --> 00:02:29.200
we're always looking at all people that started in 
November when we're looking at this. And there's  

00:02:29.200 --> 00:02:33.360
actually three ways to look at a cohort. 
And I'm going to show you how to do this in  

00:02:33.360 --> 00:02:38.800
recharge so that you see that it's entirely the 
same setting as you see on Shopify. By the way,  

00:02:38.800 --> 00:02:44.240
this is a native Shopify report. So, if you go to 
analytics and if you head over to reports and type  

00:02:44.240 --> 00:02:49.600
in customer cohort, you will get this report. it 
will by default use customer retention rate. But  

00:02:49.600 --> 00:02:55.680
as I showed you, you can view different metrics 
as well. Okay, so I've opened a totally different  

00:02:55.680 --> 00:02:59.840
brand with totally different technology. In 
this case, it's recharge just to show you that  

00:02:59.840 --> 00:03:05.280
it really is the same as what you have on Shopify 
native. The only difference is that in this case,  

00:03:05.280 --> 00:03:12.000
we are looking at subscription customer cohorts, 
meaning people that pay us in a recurring way. And  

00:03:12.000 --> 00:03:17.040
if that's your use case, you're going to find this 
very helpful because what we see here is a cohort  

00:03:17.040 --> 00:03:22.640
of a very well-developed brand. And when you look 
at this cohort, there's three ways you can gather  

00:03:22.640 --> 00:03:29.120
insights from it. One is of course to look at 
a cohort horizontally. This means uh in this  

00:03:29.120 --> 00:03:37.440
direction. And when you do this, what you really 
see is how well you retain customers over time.  

00:03:37.440 --> 00:03:43.280
So, you have initially 8,000 new customers, 
new subscription customers specifically,  

00:03:43.280 --> 00:03:51.840
entering your brand's universe in November. And 
then by month zero, you've only 84% of them left,  

00:03:51.840 --> 00:03:57.920
which means that they cancelled. 16% of them 
cancelled within the first few days, within the  

00:03:57.920 --> 00:04:05.600
first month really. And then by month one you only 
have 52% of them left and then 36 and then 31 and  

00:04:05.600 --> 00:04:12.320
then 128 etc. So what this shows you really the 
horizontal view of a cohort is very simple. How  

00:04:12.320 --> 00:04:18.240
well am I retaining customers right? The second 
view that you can have which is the vertical. It  

00:04:18.240 --> 00:04:26.320
goes the other direction from up to down. It shows 
you how well your efforts affect the retention of  

00:04:26.320 --> 00:04:33.680
your brand. To be more specific, it shows you how 
different cohorts perform in the same life cycle  

00:04:33.680 --> 00:04:38.560
month. In this case, if you look at this like 
this, it always refers to how well does month  

00:04:38.560 --> 00:04:44.640
zero do or how well does month one do or how well 
does month two do. And the reason why I say this  

00:04:44.640 --> 00:04:48.800
shows you how well your efforts affect retention 
is because presumably you're making improvements  

00:04:48.800 --> 00:04:56.000
in the life cycle and those improvements should 
over time show better retention results. So, let's  

00:04:56.000 --> 00:05:01.840
take a look at this example. This is actually one 
of our clients. They used to have 84% retention  

00:05:01.840 --> 00:05:08.480
by month zero. They're currently at 88% which 
is not too bad. It is a pretty big improvement.  

00:05:08.480 --> 00:05:15.680
And then they used to be at 52% by month one and 
now they're at 65%. Which is not too bad either,  

00:05:15.680 --> 00:05:21.120
right? Like it seems like there is a pretty big 
lift in terms of retention results for the same  

00:05:21.120 --> 00:05:25.840
life cycle month. The last thing you can do 
which is actually something that not a lot  

00:05:25.840 --> 00:05:30.160
of people like realize is that you can read the 
cohort diagonically. And when you read the cohort  

00:05:30.160 --> 00:05:36.480
diagonically what you see is what's happening in 
your customer database in this case subscription  

00:05:36.480 --> 00:05:44.560
database in the same calendar month. So let's 
take as an example this month two of the November  

00:05:44.560 --> 00:05:51.600
cohort. If I I'm going to arrive here which is 
actually January right? So it's month zero for the  

00:05:51.600 --> 00:05:58.000
January cohort and it's month two for the November 
cohort and it's month one for the December cohort.  

00:05:58.000 --> 00:06:07.440
It always refers to January. So I can see what 
if any seasonal impact exists and how it affects  

00:06:07.440 --> 00:06:12.640
my cohort separately. And the reason why this is 
important is because you may have for example a  

00:06:12.640 --> 00:06:18.160
big spike in sales come November or come any other 
promotional peak that you may be having as a brand  

00:06:18.160 --> 00:06:23.200
but then you realize that hey all of those people 
that came they were bargain hunters they just  

00:06:23.200 --> 00:06:28.560
came for the discount and they actually have very 
poor LTV for us. So next time you run a big promo,  

00:06:28.560 --> 00:06:32.800
you have a completely different approach because 
ultimately you saw that even though from an  

00:06:32.800 --> 00:06:36.800
acquisition standpoint that first purchase 
cohort looked fine when you had that sale,  

00:06:36.800 --> 00:06:43.920
it actually performed horribly over time. If you 
start thinking about it in this way, you will be  

00:06:43.920 --> 00:06:48.480
able to make much more intelligent decisions as 
a retention market. Okay, I'm going to walk you  

00:06:48.480 --> 00:06:53.360
through a real case. This is a real company that 
went through our sales process recently. And I'm  

00:06:53.360 --> 00:06:59.280
going to walk you through how to reason through 
what we call the month zero turn problem. And what  

00:06:59.280 --> 00:07:05.600
this is is very simple. It essentially describes 
customer behavior that looks like this. They find  

00:07:05.600 --> 00:07:09.680
your brand. They see that you have a bigger 
discount when somebody subscribes. So they're  

00:07:09.680 --> 00:07:13.840
like, "Okay, I'm going to get a discount and then 
I'm going to cancel right away." And you can see  

00:07:13.840 --> 00:07:22.160
this very clearly here. They have 79% retention 
by month zero. This means that like 15th,  

00:07:22.160 --> 00:07:28.160
that's crazy. 1/5ifth of their subscribers are 
just grabbing the discount and running away. And  

00:07:28.160 --> 00:07:32.960
this is a very big problem because if they cancel 
right away, they won't experience the product's  

00:07:32.960 --> 00:07:38.240
value proposition over time, which means that they 
are unlikely to stick to the brand over time. And  

00:07:38.240 --> 00:07:43.520
it also means that the company loses unnecessary 
margin. So how would you reason through this very  

00:07:43.520 --> 00:07:48.960
common problem? You would reason through it in the 
following way. Number one, you would think through  

00:07:48.960 --> 00:07:54.960
when does this decision take place. And obviously 
it takes place before they buy. So it doesn't  

00:07:54.960 --> 00:08:00.080
have to do with how the retention system is set up 
because by the time they buy, they've already made  

00:08:00.080 --> 00:08:04.080
the decision to cancel, right? They literally 
cancel before they even receive the product  

00:08:04.080 --> 00:08:08.800
in many cases. So, what you would do is you would 
take your acquisition team and you would sit down  

00:08:08.800 --> 00:08:15.040
and create an offer that is meaningfully different 
than the onetime purchase for your subscription  

00:08:15.040 --> 00:08:19.680
customers. And it would be meaningfully different 
not just in terms of savings, but also in terms  

00:08:19.680 --> 00:08:26.320
of added value, particularly added value unlocked 
over time. So, instead of saying subscribers get  

00:08:26.320 --> 00:08:34.160
20% off and that's it, you would maybe have an 
offer that unlocks certain benefits over time.  

00:08:34.160 --> 00:08:39.280
free gift on your first renewal or something along 
those lines. We've worked with companies that have  

00:08:39.280 --> 00:08:44.400
done this successfully when it comes to physical 
products and we've done this with companies that  

00:08:44.400 --> 00:08:51.040
have even included digital products, things 
that other can't get unless they subscribe. So  

00:08:51.040 --> 00:08:54.320
that's how you would reason through this. And 
then from a life cycle standpoint, of course,  

00:08:54.320 --> 00:09:00.000
you could address the behavior itself. That would 
be a supplement to fixing the subscription offer  

00:09:00.000 --> 00:09:04.400
because that of course that would be the the main 
lever here. But from a life cycle standpoint, what  

00:09:04.400 --> 00:09:09.120
you could do is address the behavior right away. 
Something like a flow that triggers telling them,  

00:09:09.120 --> 00:09:14.240
"Okay, you grab the first time offer. That's a 
smart to move. Now, here's how you can actually  

00:09:14.240 --> 00:09:19.120
benefit from the products." And then make a case 
for why consistency is the key. And then using it  

00:09:19.120 --> 00:09:24.320
once or twice or for a month is not going to do 
anything. And that actually your best customers  

00:09:24.320 --> 00:09:29.760
see results after three or four or five months 
or whatever the case may be. And that's how you  

00:09:29.760 --> 00:09:34.880
address the month zero churn problem. Okay, let's 
take a look at a totally different case that has  

00:09:34.880 --> 00:09:40.640
totally different dynamics. In this case, 
the brand has 90% retention by month zero,  

00:09:40.640 --> 00:09:44.320
which is I would say very good based on what 
we see. Typically, it's lower than that. And  

00:09:44.320 --> 00:09:49.280
then what would usually be the case, which is 
a drop from month one to month two, doesn't  

00:09:49.280 --> 00:09:54.240
seem to be there. Like they are at 76%. And I'm 
looking at the weighted average because that's  

00:09:54.240 --> 00:09:59.120
like the best way to understand what's happening 
and like exclude potential seasonal dynamics. uh  

00:09:59.120 --> 00:10:04.960
they go from 76 to 71 which is pretty good right 
like it's just 5% drop so they have an amazing  

00:10:04.960 --> 00:10:11.120
retention performance so how would you approach 
this what should you do the default thought that  

00:10:11.120 --> 00:10:16.000
most e-commerce marketers have there's reasons for 
this of course I I want to be fair to everybody  

00:10:16.000 --> 00:10:22.080
but the default thought is that let's fix our 
post purchase flow let's improve what our like  

00:10:22.080 --> 00:10:27.600
transactional emails look like and then maybe add 
more value and all that and of course that's good  

00:10:27.600 --> 00:10:32.560
because usually Usually the issue is very early 
into the life cycle and by improving on boarding  

00:10:32.560 --> 00:10:38.240
you can affect in a positive wave customer 
retention. But the thing is some cases that's  

00:10:38.240 --> 00:10:43.440
not the case like in this case they have amazing 
retention at the very beginning of the life cycle.  

00:10:43.440 --> 00:10:48.400
So for me as a strategist to look at this and then 
propose we need a new post purchase flow or like  

00:10:48.400 --> 00:10:54.720
we need to add more value as we on board new. It's 
not a good use of time and I'm being kind here. So  

00:10:54.720 --> 00:11:01.680
what I should rather do is look at my cohort and 
identify where's the biggest drop. And of course  

00:11:01.680 --> 00:11:07.280
you want to do this in percentages. You also 
want to do this like by actual absolute figures,  

00:11:07.280 --> 00:11:14.240
right? But suppose that then you would see okay 
the issue seems to be between month two to month  

00:11:14.240 --> 00:11:20.160
three. So we have a lot of people caning in uh 
as they transition to month three which in this  

00:11:20.160 --> 00:11:25.280
case does seem to be the case at a clients. Now 
what do you do? Do you create month three flow  

00:11:25.280 --> 00:11:31.440
and just a discount or like what would you do? 
No, what you would do is create a report through  

00:11:31.440 --> 00:11:38.480
recharts and you would need two data points. You 
would need data points on when people start their  

00:11:38.480 --> 00:11:44.640
subscription and when does their subscription 
end. And you can create this through recharges  

00:11:44.640 --> 00:11:49.520
port building functionality. or if you don't have 
access, if you are probably like an agency or a  

00:11:49.520 --> 00:11:54.320
fre freelancer, maybe you don't have full complete 
access, you can just ask recharge support and  

00:11:54.320 --> 00:11:58.400
they're going to do this for you. And then I'm 
going to show you what to do. Okay, you've now  

00:11:58.400 --> 00:12:02.960
created your report and it probably looks a little 
bit like this. I'm going to make it big so that  

00:12:02.960 --> 00:12:09.360
we can see the entire thing. So we have when did 
the subscription start. Yeah. and then when did  

00:12:09.360 --> 00:12:16.880
the subscription stop and then you can of course 
do this for specific time windows, right? And of  

00:12:16.880 --> 00:12:22.640
course this is 2025 so no video would be complete 
without an AI prompt. So I'm going to show you an  

00:12:22.640 --> 00:12:27.440
AI prompt that you can copy and paste and do the 
analysis that I'll show you. So you basically  

00:12:27.440 --> 00:12:33.040
take this and then you say my goal is to create 
a report that analyzes on which days within the  

00:12:33.040 --> 00:12:39.920
first 60 days or any other window that you have. 
users most often cancel and this will help me know  

00:12:39.920 --> 00:12:46.240
when to time my initiatives that would impact 
turn and then you have essentially the rest of  

00:12:46.240 --> 00:12:50.800
the information and what you get is something that 
looks a little bit like this right so you have the  

00:12:50.800 --> 00:12:57.120
days listed out and then how many cancellations 
are happening and then what percentage of the  

00:12:57.120 --> 00:13:04.800
total group do those cancellations represent and 
in this case what we see is that like half or  

00:13:04.800 --> 00:13:10.480
like 40% % approximately of them are happening in 
this time window. Yeah. And coincidentally, this  

00:13:10.480 --> 00:13:15.360
is when the billing reminder gets sent. Probably 
we need to rework that. But in another case, you  

00:13:15.360 --> 00:13:21.680
might see that it happens on a totally different 
time window. And the whole point that I'm trying  

00:13:21.680 --> 00:13:28.160
to make here is that the timing of the initiatives 
that you take really should be informed by data.  

00:13:28.160 --> 00:13:33.920
A customer cohort shows you where to focus 
and then you need to zoom even deeper to  

00:13:33.920 --> 00:13:40.320
identify the exact time frame and time window 
where you should take initiatives to combat term.  

00:13:40.320 --> 00:13:45.920
What type of initiative you should take obviously 
has a lot to do with the financial picture of your  

00:13:45.920 --> 00:13:51.280
company the brand constraints that you may have 
other initiatives that you may already have on  

00:13:51.280 --> 00:13:55.760
the pipeline. So I can't really tell you, okay, 
people are churning on day 37, so you should do  

00:13:55.760 --> 00:14:00.800
XY Z. Even though if you do reach out to me, I 
will be happily able to take a look and give you  

00:14:00.800 --> 00:14:05.600
more detailed advice on what to do. But I hope 
this was helpful and I hope it at least showed  

00:14:05.600 --> 00:14:12.080
you why cohort analysis is such a powerful tool. 
There's three ways to look at things and just so  

00:14:12.080 --> 00:14:16.960
that you remember them, I'm going to repeat them 
once again. If you look at retention horizontally,  

00:14:16.960 --> 00:14:22.240
you are able to see how one specific cohort 
is retained over time. And then of course,  

00:14:22.240 --> 00:14:26.320
if you look at the average of all of the months, 
you can get a very clear picture of how as a  

00:14:26.320 --> 00:14:31.600
brand you are retaining customers over time. 
If you look at retention cohorts vertically,  

00:14:31.600 --> 00:14:37.360
what you can see is how are you having an 
impact if any on specific stages in the life  

00:14:37.360 --> 00:14:42.720
cycle like how does my efforts affect month 
one retention or month three retention and so  

00:14:42.720 --> 00:14:48.720
on. And then diagonically you can see what the 
specific calendar month looked like across your  

00:14:48.720 --> 00:14:55.440
customer cohorts which typically is helpful for 
evaluating what if any impact decisional events  

00:14:55.440 --> 00:15:00.400
promotions and dynamics as a whole like it might 
be something that's related to politics or like I  

00:15:00.400 --> 00:15:05.840
don't know the economy are affecting your customer 
subscription. I hope this is helpful and I hope  

00:15:05.840 --> 00:15:11.600
you take the time to apply this. Remember there's 
two main scenarios. One is that this information  

00:15:11.600 --> 00:15:16.400
passed through one year and will exit the other, 
which mean you might say it's helpful info,  

00:15:16.400 --> 00:15:19.920
but then you don't do anything with it. And 
the other is that you take at least one thing  

00:15:19.920 --> 00:15:24.640
that we learned in this video and that you try to 
apply it for your own brand. And overwhelmingly,  

00:15:24.640 --> 00:15:29.040
if you do the latter, you're going to get much 
better results. If you found this helpful,  

00:15:29.040 --> 00:15:33.520
please be sure to subscribe and be sure to share 
this with people that might also find it helpful.  

00:15:33.520 --> 00:15:37.920
This is one of the very first videos I'm 
publishing, but I promise you this, this  

00:15:37.920 --> 00:15:44.640
will be one of the best resources that are in the 
world about e-commerce retention, life cycle, and  

00:15:44.640 --> 00:15:49.680
overall subscriptionbased marketing. So, thank you 
for watching and I'm going to see you on the next