This is a guest post by Charlie Carpenter, co-founder and CEO of Kite


Anyone running an eCommerce site – or, indeed, any type of website – will be acutely aware of the importance of conversion. You spend huge amounts of time and often money on generating traffic. It’s no exaggeration to say that maximizing the performance of that traffic is the difference between a website that succeeds and a website that fails.

The average conversion rate for an eCommerce website, as per Monetate Ecommerce Quarterly, is a paltry 2.86%. This means that, for every 100 visitors, less than 3 of them on average are going to convert. 97 will leave without taking any action – which is enough to give you nightmares if you stop to think about how hard you worked to get them on your site in the first place.

Every visitor to your website, every customer, is inherently valuable to your business. You could learn something from all of them if you had time to find out why they didn’t convert.

But for most businesses, it’s at least impractical – and most likely impossible – to analyze every single unique visitor. The time it takes to collect and analyze this data and use it to inform business decisions can be enormous. Having the time and resources for this extensive analysis is a luxury that simply isn’t realistic to most SMBs.

Using this extremely specific data is also not the wisest way to make broad business decisions. It’s like a stand-up comedian hyper-analyzing the reaction of one member of their audience and making sweeping decisions about their whole act. You have to make decisions for the benefit of your whole audience, not just individuals within it.

So where does that leave you?

You know you need to use metrics; you don’t have infinite time to do so, and you don’t want to make decisions on the limited experiences of individual visitors.

So the next logical step is to group your visitors together, in an attempt to identify trends and patterns. Identifying these can make all the difference in understanding the unique issues and problems faced by your website, and removing the obstacles to supercharging your conversion rate.

That’s the exact premise of cohort analysis: a common, and incredibly effective, way to group your customers together, and drive real insight from exploring their collective experiences.

What is a cohort?

Before we delve into the world of cohort analysis, it’s important to briefly explain what a cohort actually is. Basically, a cohort is a group of people who share a common characteristic within a defined period.

But what does that mean? Technically, people born between 1970 and 1990 who put on their left sock first thing in a morning could be considered a cohort. (Of course, it’s a cohort that has little value to your business!)

But there are plenty of other cohorts that would be useful, which might include….

  •     Paying customers in the last 30 days.
  •     Those who added to their cart and then abandoned it, this quarter.
  •     People who have signed up to your monthly newsletter this year.
  •     Those who visited your site from an organic search in the past fortnight.
  •     Visitors to your website after clicking on an advertisement this month.
  •     People who came to your site from social media in the month of June.

The number of cohorts you could analyze are basically unlimited. It’s up to you to identify the common characteristics that define the various ‘clusters’ of your web visitors – and which of these clusters offer the most strategic value to improve your website performance.

What is cohort analysis?

Cohort analysis is all about analyzing those groups or cohorts you identified to identify patterns and trends that unite them.

It’s a form of behavioral analytics that groups data together, instead of examining user experience singularly. The idea is that by looking at the groups we defined above, you can identify trends and patterns. This means that you’re adjusting your website based on the real experience of significant chunks of your audience.

If there’s a particular reason a cohort isn’t performing well, you can reach actionable conclusions and make changes to your website to help improve performance.

Cohort analysis in action

That’s a pretty broad, conceptual overview of what cohort analysis is and how it might work.

But let’s now think of a few practical, real-world ways you could use cohort analysis to enhance your conversion rate. These examples are super-simple, but remember: there’s an almost unlimited number of ways you could use cohort analysis for your own unique circumstances.

Understanding cart abandonment

Ecommerce sellers will be painfully aware that cart abandonment rate can range anywhere from 50-80%. You could analyze that ‘cohort’ of people who added items to their basket but didn’t check out. You might establish, for example, that a high number of them dropped out during the account/registration stage. This could indicate that allowing guest checkout would increase conversion rate significantly. You could then test and validate that theory via A/B testing using a platform like AB Tasty. Allow 50% of your audience to check out as a guest, while the other 50% go through your existing process. Compare the performance of those two groups – and roll out the winner.

Understanding conversion by device

We talk about website traffic as one big label, but obviously, in the mobile age, that’s split across a variety of different devices, including desktop, mobile, and tablet. It seems crazy, but many websites are still grappling with the reality of delivering great experiences across all devices. By analyzing your mobile users as a cohort, you might identify that you have lower conversion among mobile users. You might then consider looking into the mobile-first index to help you correct this issue. (I wrote another article about the mobile-first index here.)

Optimizing site search

Another example of a cohort might be those people who use your site’s own search function. You might be able to identify issues around how they use search – so you can identify and improve conversion issues. Again, the theories and solutions you ideate can be tested and validated using split testing, which can be an ongoing process.

Age and gender

Many modern businesses have diverse audiences, comprising a range of ages, genders, and backgrounds. By grouping demographics together, you might identify that certain groups or genders are underperforming. You could correct course by experimenting with some new content or promotional ideas. For example, if you found that 14-21-year-olds weren’t performing particularly well, you could try reaching out to some influencers to promote your brand.

Conclusion

The marketing optimization expert and bestselling author Jeff Eisenberg hit the nail on the head when he said: “It’s much easier to double your business by doubling your conversion rate than by doubling your traffic.”

We spend huge amounts of time and money on generating traffic – which makes it all the more frustrating when that traffic amounts to nothing.

It’s important that online businesses treat conversion seriously. This includes comprehensive research into why people don’t convert. Realistically, the quickest and easiest way to do this is by grouping your various audience chunks into cohorts – and then using A/B testing to validate and test the solutions you come up with.

Good luck!