Editor’s note: Here’s another post from the great people of TruConversion as part of their follow-up to the heat maps article they published on my blog a while back. Again, the post has little to do with freelance writing, but the analytics components discussed in the article is something that every marketer, even freelancers, need to consider when trying to increase their sales and client base.
80% of all SaaS revenue will come from just 20% of its customers.
The 80/20 rule is still alive.
An increase in retention of just 5% can increase your profitability by 75%!
What if I told you there was a way to find those 5%? And even predict where they’ll come from?
Read on and learn the secrets of using Cohort Analysis for Conversion Optimization.
- What Is Cohort Analysis?
- What Are The Benefits Of Cohort Analysis?
- How Can You Use Cohort Analysis?
- Cohort Analysis vs The Traditional Conversion Funnel
- Examples Of Cohorts (Medical, Sales, Economics, etc.)
- Cohort Analysis And A/B Testing
- Cohort Analysis And The Perfect Customer Persona
Click on the links below to jump to that section on the page
1. What Is Cohort Analysis?
The Conversion Funnel is one of those essential digital marketing concepts. In practice, it allows you to monitor incoming traffic, how much of that is transforming into email subscribers and, ultimately, how your conversions are doing.
And that’s a good thing, It’s a very powerful element. However, it’s main flaw stems from its lack of predictive abilities. Sure, you can look at a current funnel and assume that the Conversion Rates will be the same. But is the data really helping you make informed decisions?
Enter the Cohort Analysis. This splits visitors into groups, based on a common characteristic, and compares their behavior over time.
Cohorts have commonly been used in the medical field, to study the impact of drugs and vaccines, over long periods of time. For example, people born in the same year share what is called a “Birth Cohort”. Within that cohort there can be subgroups formed – Birth Cohort + Location Cohort, Sex Cohort, etc.
A great (illustrated) example of Cohorts can be found on this RJMetrics microsite. It makes it easy to understand what Cohorts are and what their use cases are. Let’s continue and take a closer look at the benefits of this type of analysis.
2. What Are The Benefits Of Cohort Analysis?
We must stress this again: the Booking Funnel is not broken. It’s still a very capable tool and definitely one you should have in your marketing arsenal. If, however, you run more of a SaaS business, Cohort Analysis will provide more actionable data for you to look at and use.
- It’s predictive.
Since you’ll be looking at data over weeks and months, trends will emerge. Just like with a marketing funnel, Conversion Rates will never stay the same over time. However, Cohort Analysis provides a more granular view on how your visitors are behaving on your website.
You’ll be able to predict how certain cohorts will act. For example, you might notice that visitors coming from specific search ads are converting at a higher rate than visitors from a blog post. It only makes sense – visitors in search are already looking for a solution to their problem, while blog readers are not in “buy mode” yet.
On the other hand, looking at costs, you might be able to say that a blog post converts fewer people, but it also has lower overall costs. So it would make sense to create more posts (on your own website + guest posts) and track those over time.
- It’s customizable.
With Cohort Analysis you’re not limited to looking at just one thing. It does get a bit more complicated, if you insert multiple sources and demographics data, but as long as you don’t get carried away, you should be ok.
Here is an example that uses a few elements to get a more complete picture regarding User Retention. It’s tracking the Signup Day, the number of users and the retention over specific periods of time (1, 7, 14, 28 days).
- It’s actionable.
It really depends on the type of tracking software you’ll be using, but Cohort Analysis does provide a lot of answers. You just have to ask the right questions, in order to make informed decisions based on that data.
Bottlenecks and obstacles will be a thing of the past, if you take action after you understand the data.
Imagine you’ve been tracking customer spend over time. You’re looking at a basic Cohort Analysis chart and you notice that monthly product usage begins to drop in Month 3, for all users. It’s not a technical issue, you’re pretty sure it’s not something to do with the time of the year.
You start to realize that users did not get the full benefit of being onboarded properly. That means they’ve probably gotten a basic Welcome Email about how to use the software and that was it. In order to make improvements, you can think about sending triggered emails (when the customer uses the app for 1 week, 1 month, etc.) and asking them for feedback.
In this particular example, it could simply be that they didn’t understand all the benefits of the app, so by month 3 they just gave up using it.
3. How Can You Use Cohort Analysis?
3.1 Predict Future Behavior In Terms Of Purchasing Power
If you’re a gamer, you’ll understand this first example. Looking at the charts above you’ll soon learn that there is a direct connection between the preference towards buying credits and game lag. While registered and basic players still are interested in game credits as lag increases, expert/advanced players are much more sensitive.
In other words, the more gamers have to wait, the worse an experience they get in the game. And that translates to a lower desirability towards purchasing credits towards playing more.
You’re essentially segmenting users, looking at 2 data points and predicting future behavior. What you should be doing next is implementing user surveys. deploying incentivizing campaigns and lowering lag time, to increase overall credit purchases.
3.2 Estimate Acquisition Cost VS Lifetime Value
Things will not always be as simple as the table above makes it seem. But you can extract data from a Cohort Analysis and lay it out in a more convenient table format, to use for your own purposes.
The value that this type of analysis brings you is immense. Being able to know not only the acquisition cost for a user coming from a certain marketing channel, but also its lifetime value, makes it so that you can make smarter choices. Choices related to ad spend, content marketing and so on.
Once you have a good amount of actionable data, you can start working towards cost acquisition optimization. You can offer free trials, decrease initial upfront costs, implement a user referral campaign, etc.
3.3. Improve Lead Generation
Cohort Analysis is a lengthy process. Depending on the type of business you have and the services you provide, you will need a few weeks or months to gather data you can look at. But once you have that, you’re free to dig in and think about improvements.
In the case of a SaaS business, lead generation and trial-to-customer conversions are crucial. Without them, there would be no business. Let’s assume your free trial lasts for 14 days. That means that the users who signed up in Month 1 and stayed in Month 2 are paying customers.
That trial-to-customer Conversion Rate is simply a number. Without additional tools and data, you won’t be able to know if users are about to abandon your product. Framed is a service that uses machine learning to provide you with useful data so you can prevent and reduce user churn.
4. Cohort Analysis vs The Traditional Conversion Funnel
Funnels don’t make it easy to track long lifecycle events. They’re not meant to, as they weren’t built for that purpose. Unfortunately, they’re also pretty bad at tracking split tests and measuring retention (as that tends to happen over weeks and months).
The Cohort Analysis doesn’t entirely replace the Conversion Funnel. In fact, they can work together to better visualize how certain events affect users’ behavior over time. By measuring key events that push leads towards conversion, you can get a feel for what works and what doesn’t. And since you’re looking at all of these across time, you can make changes and tweaks and see how it affects your end goal.
Split tests also function well if you’re tracking a variation from a Free Trial to a Freemium model, for example. The chart above (albeit not based on real data), shows how you can have the same actions side by side and compare them across lead conversion tactics. This is the first step, the next being actually implementing changes to figure out if you’re improving things across your Conversion Funnel. You might be surprised to see that a change you’ve made at the top of the funnel drastically increased conversions at the bottom of it.
The trick here, as with most Cohort Analysis you’ll perform, is to keep it really simple. If you make the mistake of adding too much data, you’ll lose yourself in all the different scenarios. The beauty of this type of analysis is that you can pull up certain elements you’re interested in and ignore others. Time is an important indicator and so is the Conversion Rate. Next, you’ll want to define and track key actions that your cohorts are performing. Once you have those, you’re ready to start optimizing and see how your visitors are acting inside of the Conversion Funnel you’ve implemented.
5. Examples Of Cohorts (Medical, Social, Sales, etc.)
Cohort Analysis is not restricted to digital marketing. Since it originated in the medical field, it’s worth nothing what other types of organizations, applications or business use cohorts.
Does exposure to X (smoking) lead to outcome Y (lung cancer)? This is just one question that can be answered in the medical field, using cohorts. You essentially try to isolate the smoking element as the main cause of lung cancer.
You get 2 groups of people (smokers and non-smokers) and try to have them be as similar as possible. That way, their lifestyle won’t dramatically impact what leads to the outcome.
You then measure those cohorts over a set period of time and analyze the data. Based on that, you can repeat the test and figure out if your initial assumptions were accurate.
Are people who follow a few twitter users at first more engaged in the platform? Twitter was having a challenging time making users who sign up actually use the platform for more than a few days.
Then they realized it was all down to their social environment. If you create an account on a new social platform, there isn’t a lot to do. So now when you sign up for a twitter account, there are popular users recommended you should follow.
Twitter could compare the cohort that didn’t follow anyone with the cohort that had user suggestions. Overall, engagement and retention went up, so the experiment paid off.
Do engaged users care more about your product? And do these users end up paying more and staying longer with the service?
Perhaps the most important type of Cohort Analysis is done at the sales level. You compare traffic sources and acquisition costs. You look at lifetime customer value and retention. You track the Conversion Rate from trial to paying users.
Having early user feedback is essential. It’s worth testing whether listening to your users, asking questions and making constant improvements to the products results in a higher number of monthly customers.
6. Cohort Analysis And A/B Testing
Things aren’t always as simple as you might want them to be. There are 3 main concepts we need to differentiate between: cohort analysis, a/b testing and multivariate testing.
6.1 Cohort Analysis
You use this type of analysis when your customer base is just too big to average all the data and behavior together. You get to focus on specific parts of the funnel – such as engagement or conversion and separate any elements you might thing have an influence on the outcome.
For example, if you implement a new website feature and your overall engagement with it is low, it’s a good idea to dig deeper and have a more granular view on your visitors. Once you look at them through this lens, you might notice that engagement is not actually low for every user.Old users understand your website better and are thrilled about the
Old users understand your website better and are thrilled about the improvement. New users, however, aren’t quite sure what to make of it and are skeptical of trying it out, since they’ve just landed on the website.
6.2 A/B Testing
When testing big changes, you would use this type of analysis. A/B Testing is for checking how two features are different, while Cohort Analysis is for seeing how groups of users are different.
For example, if you’re using a landing page that only has one call to action, testing two different button colors is a simple way to figure out what issues your users might encounter. There are simply too few options to think about user behavior on a very detailed scale. A/B Testing is enough for this purpose.
Lastly, A/B Testing doesn’t require that much time to see initial data, but it’s limited to a more general view of the website visitors.
6.3 Multivariate Testing
You use this type of analysis and testing method when you’re checking a variety of changes at the same time. Multivariate Testing is usually reserved for making small adjustments to a page, rather than big changes.
For example, if you’re testing a few changes to a checkout flow on an ecommerce website, Multivariate Testing will help you look at all the different factors at the same time. That’s both a good thing and a bad one – since you’re analyzing a lot more data simultaneously.
This type of analysis helps you understand a bit of user behavior even when one small improvement or issue is related to another. You get to trace users’ steps and make changes along the whole conversion path.
No matter what type of analysis you use, it’s important to remember that each is better suited for a specific task. In our case, use Cohort Analysis to look at behavior over longer periods of time.
7. Cohort Analysis And The Perfect Customer Persona
We’ve talked about Customer Personas in the past. Their importance in digital marketing cannot be overstated. Now we have another element in the mix: Cohort Analysis. How do the 2 combine? Let’s find out!
Customer Personas are entities built to represent groups of individuals, that share similar traits, personalities and interests. They’re created using customer surveys, in-person interviews, analytics data, forum and chat data, etc. If we look at DataHero‘s article about personas, they’ve included Cohort Analysis as a way of finding more info.
Since we know that Cohort Analysis does well over longer periods of time and it’s also predictive, it not only helps you understand your users better now, but also in the future. That means that you can focus on past triggers, actual behavior, but also predict how they’ll react in a certain situation on certain pages. This gives you an incredible amount of data to play with.
Combine those cohorts with your usual Customer Personas and you’re already one step ahead of the competition. While they’re looking at past actions and realtime traffic, you can keep your pulse on what’s going to happen to specific sets of users.
Imagine knowing that a visitor coming from Facebook will lose interest in your product by month two. Perhaps it’s because the platform offers instant gratification, the targeting might have been off or a combination of those factors. Now you’ll use that cohort data to implement email marketing campaigns to capture more leads. You’ll also deploy remarketing across your entire website. So that no matter what your visitors do, you’ll be able to track it, understand it and take actions that benefit you.
You’ll be able to get a much better handle on customer churn and user retention if you remember to combine Cohort Analysis with Customer Personas. Since most of the tools are at your disposal, it would be a shame not to use this opportunity to provide more personalized marketing solutions for your users.
[su_note note_color=”#F15A24″ text_color=”#fff”]
This concludes the first half of the article about cohort analysis. The second part will be published next week and cover the following topics.
- Cohort Analysis And Conversion Optimization
- Cohort Analysis And SaaS
- Cohort Analysis And Mobile Apps
- The Most Important Cohorts In Cohort Analysis
- Specialized Apps For Cohort Analysis
- Final Tips And Tricks
- Recommended Reading: Articles And Guides
Until then, please share your thoughts about the article below to spark off the discussion!
UPDATE: The second half has been published and can be viewed by clicking on this link.