Cohort Analysis for Ecommerce: How to Track Customer Retention Like a Pro

Overall averages lie. A store with a 30 percent repeat-purchase rate might have January customers at 45 percent and July customers at 15 percent. Without cohort analysis you would never know your summer campaigns were attracting low-quality buyers, or that a product change in June quietly hurt retention.

Cohort analysis is the lens that turns blurry averages into sharp, actionable insight. Here is how to run it.

What is cohort analysis?

A cohort is a group of customers who share a characteristic within a time window. The most common one is first purchase date: everyone whose first order landed in a given month forms that month's cohort.

Cohort analysis then tracks how each group behaves over the following periods. You can measure:

• Retention rate: what share of the cohort bought again in month 2, 3, 4, and beyond.
• Revenue per customer: how much cumulative revenue each cohort generates over time.
• Average order value trend: whether AOV rises, falls, or holds across repeat purchases.
• Payback period: when cumulative profit from a cohort passes its acquisition cost.

Building a cohort analysis, step by step

1. Export your order data. You need a row per order with customer ID or email, order date, order total, and ideally acquisition channel and first product.
2. Find each customer's first purchase date. That date determines their cohort. A first order on January 15 puts them in the January cohort.
3. Calculate time since first purchase. For every order, work out how many months have passed since that customer's first order. The first order is month 0, an order weeks later is month 1, and so on.
4. Build the table. Rows are cohorts by first-purchase month, columns are the periods (month 0, month 1, month 2). Each cell holds the share of the cohort that bought in that period.

Here the March cohort retains better, 35 percent in month 1 versus 28 percent for January. That could be a product improvement, a stronger campaign, or a seasonal effect. In a blended average, the improvement would be invisible.

How to read retention curves

Retention curves start at 100 percent in month 0 and decline. The shape of the decline tells you a lot.

• Flattening, levelling off in the mid-teens to mid-twenties percent: healthy. You have a loyal core, so grow it with loyalty programs and subscriptions.
• Steep drop below 10 percent by month 2: a product or experience problem. Fix the post-purchase experience and investigate product quality.
• Slow decline that never flattens: gradual disengagement. Add winback campaigns and improve your email flows.
• An uptick after a dip: seasonal or campaign-driven repurchase. Lean into the trigger and schedule more at that cadence.

Beyond time: other cohorts worth cutting

• Acquisition channel. Group by how customers found you, such as paid social, search, organic, email, or referral. This reveals which channels bring customers who stick around. Often the cheapest channel has the worst retention.
• First product. Group by the first product purchased. Gateway products that lead to high retention deserve more acquisition budget; products that attract one-time buyers deserve less.
• Discount versus full price. Compare customers acquired on a discount against those who paid full price. Discount-acquired customers often show meaningfully lower lifetime value because they wait for the next sale.
• First order value. Group by first-order size. Higher first orders often correlate with higher retention, because the customer showed stronger intent.

Turning findings into action

Cohort analysis only matters if it changes what you do. The most common findings and their fixes:

1. Month 1 retention is low. Your post-purchase experience needs work. Add a post-purchase email sequence, include usage instructions in the box, and follow up a few days later with a satisfaction check.
2. Retention varies wildly by cohort. Something changed. Did you change ad targeting, launch a product, or run a big promotion? The variation is your biggest clue.
3. One channel retains far better than another. Shift budget toward the high-retention channel even if its cost per acquisition is higher. A customer who buys four times is worth far more than one who buys once.
4. Discount customers never return at full price. Lean less on discounts for acquisition, and test value-add offers such as a free gift instead of a percentage off.
5. Retention flattens at a loyal core. Focus on moving one-time buyers to a second purchase, which is where most churn happens.

The biggest retention drop is almost always between the first and second purchase. Fix that transition and the whole curve lifts.

Where automation fits

Building cohort tables by hand in a spreadsheet works, but it is slow, easy to get wrong, and rarely kept up to date. It is a natural job to automate: generating cohort reports across time, channel, product, and discount, and flagging when a trend shifts along with a likely cause. That is the kind of analysis StoreWiz is being built to run. What is live today is the free store audit; the autonomous platform is in active development.

Key takeaways

• Cohort analysis reveals retention patterns that overall averages hide.
• A healthy curve flattens in the mid-teens to mid-twenties percent by month 4 to 6, showing a loyal core.
• Cut cohorts by acquisition channel to find which sources bring lasting customers.
• The biggest drop is between purchase one and purchase two, so focus your effort there.
• Discount-acquired customers often carry lower lifetime value than full-price ones.
• Review cohorts monthly and investigate any big change between them right away.

Frequently asked questions

How much data do I need? At a minimum, about six months of order history with roughly a hundred or more customers per monthly cohort. Twelve months is better because it captures seasonality. If your monthly cohorts are small, use quarterly cohorts to keep the groups meaningful.

Monthly or weekly cohorts? Monthly works for most stores. Use weekly cohorts if you have high order volume and want faster feedback, or if you run frequent campaigns and need to isolate each one. Weekly needs more data but reacts sooner.

What is a good month 1 retention rate? For ecommerce, somewhere around a quarter to a third of customers making a second purchase within the first couple of months is good. Higher is excellent and common for consumables and subscriptions. Much lower points to a retention problem. Frequency varies by category, so a supplement brand should retain better than a furniture store.

Can I do this without a dedicated tool? Yes, in a spreadsheet. Export orders, find each customer's first purchase date, calculate the month offset for each later order, and build a pivot table. Expect a couple of hours to set up and about half an hour to update monthly. A dedicated tool automates it, adds visualizations, and can slice by several dimensions at once.