).webp)
If you find yourself asking questions like:
- Are our marketing campaigns getting better?
- Are we acquiring the right customers?
- Which products bring in the best customers?
Then meet your new best friend, cohort analysis, the secret to understanding which marketing activities are winners and losers, which products to promote, and which customers are your most valuable.
Here’s what we’ll be covering:
1. What is cohort analysis?
2. Why is cohort analysis so important?
3. How to perform cohort analysis
4. How to use cohort analysis
What is cohort analysis?
A ‘cohort’ is a way of grouping users by their behavior and common characteristics. Cohort analysis is a method of analyzing groups of users over time to find out which subsets are the most valuable and why.
A customer cohort analysis is presented as a visualization of a data set, which uses different colors and metrics to show the breakdown of each cohort by the metrics in question.
We can put that into practice and say at college, each new class can be thought of as a cohort. The class of 2021 has many differences from the class of 2022 - start and finish dates, the economies they graduate into, the jobs they’re offered and the income they earn over their careers.
A cohort analysis report can be used to identify how students in 2021 differ from those in 2022 over a specific time period, and what the difference in outcomes is as a result.
Although cohort analysis is well-known in the SaaS industry. It’s fast becoming an important metric for ecommerce businesses too.
An online store could create the following cohorts:
- All customers who bought a particular product
- All customers from a specific country
- All customers who were acquired through a specific marketing channel
It’s important to note that customers can be part of several cohorts at the same time - a Canadian customer who buys a sweater in January is part of a country cohort, product cohort, and date cohort.
The two types of cohorts
There are two types of customer cohorts: behavioral and acquisition.
Behavioral cohorts can be grouped together by the specific actions of users over a defined time-span. An example would be a cohort who subscribed to an email newsletter or followed an account on social media.
Behavioral analytics help you to understand how users engage with your business, which campaigns create your best customers, how to increase your customer retention rate, and reduce churn rate.
Acquisition cohorts are focused on when and how a visitor first became a customer. This cohort type creates a specific group of users from factors like:
- Acquisition date
- Campaign type
- Discounts used
- Marketing channel
- Products purchased
Acquisition cohorts are useful for establishing at which part of the customer lifecycle your users drop off.
Why is cohort analysis so important?
Cohort analysis reports are one of the most useful, but under-used business analytics methods in the ecommerce world.
They help you understand how changes to your marketing campaigns impact your KPIs.
Ecommerce brands that don’t use measurement techniques like cohort analysis tend to judge their marketing campaigns using short-term metrics. For example, campaigns that don’t hit their Return On Ad Spend (ROAS) goals are considered failures and turned off.
However, this approach fails to account for the Customer Lifetime Value (CLV) of each customer acquired.
On the surface, spending $50 to acquire a customer who makes a purchase of $25 looks unprofitable. But if that customer has a CLV of $200 because of future purchases, that’s a profitable Customer Acquisition Cost (CAC.)
Cohort analysis is so valuable because it helps us to understand the impact of marketing campaigns after the first purchase.
.png)
Using the example above, the same marketer could take two completely opposite actions:
- Based on ROAS, they pause the campaign,
- Based on what cohort analysis tells them about CLV, they increase their media spend.
Cohort analysis allows ecommerce brands to move from a short- to long-term outlook and make decisions over a longer timeframe.
In an industry like ecommerce, where a significant amount of the profit you make per customer comes after the first purchase, cohort analysis is a game-changer.
Why isn’t cohort analysis used more?
As cohort analysis is a lagging, long-term measurement method, it takes several months after each new users’ first action for that data to be available. This delayed feedback makes it difficult to use cohort analysis in isolation to make quick decisions.
However, online stores can combine cohort analysis with long-term predictive data like CLV to approximate how different cohorts are likely to perform over their lifecycle.
Another reason that cohort analysis isn’t used as much as it should be, is that while free-to-use tools like Google Analytics offer a cohort analysis report, it’s fairly basic in nature.
The Google Analytics cohort analysis report works solely on acquisition date, so requires customization using segments and filters to uncover more actionable data. While you can do cohort analysis in Microsoft Excel, it gets complicated fast and takes time to learn.
However, paid alternatives like Segments do the hard work for you and show more actionable data.
They help ecommerce brands answer questions like ‘which products helped us acquire our best customers?’ and reveal insights like:
- Which products generated the most new customers? Use this to identify which products new customers purchase the most and use those products heavily in your marketing campaigns.
- Which products have the highest Customer Lifetime Value (CLV)? Find the products that lead to your best, most profitable customers.
- Which products lead to the most repeat orders? Understand which products generate a higher frequency of transactions from first-time customers over their lifetime.
How to perform cohort analysis
There are two main ways to approach cohort analysis:
- By choosing a particular metric, such as Customer Acquisition Cost (CAC) or Customer Lifetime Value (CLV) and then comparing across cohorts, or
- Defining your cohorts, such as by acquisition channel or first product purchased, and then looking at how they differ across multiple metrics
Which method you choose depends on what you’re trying to accomplish. If it’s your first time using cohort analysis, we’d recommend starting with the latter.
Defining your cohorts first will help you get a ‘big-picture’ overview of your business. Once you’re comfortable interpreting the data set, you can dive into specific metrics.
An example is comparing cohorts by acquisition channel. You can see how metrics like CAC, AOV, CLV, and products purchased differ by the marketing channel that was used to acquire your customers.
If you find that your referral channels have a low CAC or a high CLV, that’s a good sign you should allocate a high percentage of your marketing budget to that channel.
And if you see that social media channels are bringing in lots of sales, but they rarely turn into repeat purchases, you can tweak your campaigns to promote different products that are more likely to generate second and third orders.
As well as comparing cohorts by acquisition channel, you can also define your cohorts by characteristics like acquisition date, or by first product purchased.
How to use cohort analysis
While there are many ways cohort analysis can be used to grow an ecommerce brand, we’ve identified six of the most useful.
- Increase revenue and CLV
Group your cohorts by the date you acquired each customer. Measure how much revenue you make from that specific group during the next 6-12 months. Once complete, do the same with CLV.
Look at what you did differently in your best and worst-performing months in terms of revenue and CLV. Answer these questions:
- Could these differences be down to seasonality? If so, make sure you’re promoting the most relevant products at the right time of the year
- What marketing campaigns were running? Aim to replicate high-performing campaigns and tweak under-performers
- What sales or special events happened? If they achieved good results, can you run them again, or create a variation of them to be used on a different customer segment?
- Which new products were released or heavily promoted? If they brought in sales, can you include them in more of your marketing or feature them prominently on your website?
Double-down on the activities that contributed to your best months, and tweak the activities you were running during your worst months. Take a successful campaign and run it with a different customer base using segmentation.
- Discover your best acquisition channels
Create cohorts of customers segmented by the marketing channel that was attributed with their first purchase.
Look for channels that:
- Generate a lot of customers that only purchase once, and optimise those campaigns if their CLV isn’t significantly higher than their CAC
- Attract customers with high CLVs. You want more of these customers, so it’s usually OK to pay a higher CAC to acquire them (providing CAC doesn’t exceed CLV.)
- Bring in repeat customers. These customers are likely to have a high CLV, which is useful to know if you haven’t calculated CLV yet
Track changes to your marketing campaigns over several months to see what impact they have. Use cohort analysis to stop wasting money on marketing activities that aren’t getting the right results.
- Identify your best products
In this context ‘best’ means products that have led to repeat purchases and high CLV customers, not just those that have the highest sales figures.
To find this information, create cohorts based on which product customers purchased first.
Ask yourself:
- Which products generated the most new customers?
- Which products generate the highest CLV?
- Which products lead to the most repeat orders?
Once you’ve identified the top products for each of these questions, audit your marketing channels to make sure you’re promoting these products heavily. Feature them in prominent positions on your website, add a best-seller category, and create offers based on these specific products.
- Analyze seasonal shoppers and discount-buyers
Group cohorts by the date you ran specific season-based or discount-focused campaigns - like Black Friday or 4th July flash sales.
Dig into customer data and figure out whether these campaigns generate high-value customers that will buy from you year-round, or one-off customers who will only buy when products are heavily-discounted. Discount shoppers lower your profit margins, and although almost 90% of Americans will use a discount coupon at least once a year, some customers will only buy if your items are discounted.
If a particular marketing channel generates a high-percentage of discount shoppers, the activity may still be worth running from a cash-flow perspective (providing ROI exceeds CAC.)
When evaluating discount shoppers, it may be better to judge the campaign based on a metric that’s more likely to reflect single purchases, like Return On Ad Spend, rather than a long-term, multi-purchase metric like Customer Lifetime Value.
- Discover which discounts to offer
To figure out which discount codes or flash sales are most effective, group cohorts by discount codes used as a first-time customer.
Rank your discounts by CLV to understand which offers generate high-value customers over their lifetime. It may be worth breaking even or perhaps even losing money on the first purchase in order to acquire a highly-profitable customer in the long-run. This is an example of a loss-leader discount, losing money in the short-run in order to gain a customer that is likely to make further purchases at a profit.
This type of cohort analysis will also show whether your discounts attract repeat customers or one-time buyers.
You’ll understand which discounts you can use to:
- Increase short-term cash flow - those that make money in the short-term but don’t necessarily translate into repeat purchasers with a high CLV
- Acquire high CLV customers - those that help you acquire customers who make repeat purchases of products with healthy profit margins
- Improve your email marketing
Email marketing generates an average of $35 for every dollar spent, and is credited with more than a quarter of all ecommerce revenue.
To analyse your email marketing, group cohorts by whether they’re subscribed to your email list or not. Use metrics that aren’t impacted by cohort size (e.g. AOV.)
Explore the differences between the two cohorts to understand whether your email list is generating higher:
- Active users
- AOV
- CLV
- Purchase frequency
- User engagement
- User experience
- User retention
If your email marketing isn’t generating a higher score in the above metrics than those who aren’t subscribed, you need to work on improving the content of your emails, your segmentation, and analyzing your sending frequency. Dive into Klaviyo’s email marketing guides for a good starting place.
Start using cohort analysis today
We’ve covered what cohort analysis is, why it’s important, and six actionable ways you can use it to grow your ecommerce brand. Carry out your first cohort analysis to understand the impact your marketing campaigns are having after the first purchase, and start making smarter, data-informed decisions.
If you’re looking for a tool that can carry out cohort analysis automatically, then check out Segments’ free 14-day trial, and we’ll do the data-crunching for you.
Other ecommerce guides
Want to learn how to understand and improve other important metrics and benchmarks for your online store?
Read our in-depth ecommerce guides here:
Authored by
Tresl Data Science Team
George Sylvain
George Sylvain is a San Francisco-based DTC expert and co-founder of Social Print Studio, known for transforming e-commerce strategies into success stories. Visit his insights on AI and Shopify at www.georgesylvain.com.
Sharad Thaper
Alex Greifeld
