Ecommerce customer retention: the metrics to track, a worked cohort table and the levers that bring buyers back

Ecommerce customer retention is how many of your buyers order again, how soon and how often. For a consumable such as a supplement, track repeat purchase rate, time to second order and cohort retention by month of first order, then turn them into lifetime contribution per customer. You raise it by getting the product used, reminding before the supply runs out, keeping subscriptions honest, and making shipping and support dependable.

What is ecommerce customer retention?

Retention is the share of customers who buy again. For a consumable it is the business model: a bottle that lasts 30 days is meant to be bought again around day 30, and the ad that brought the first order doesn't have to be paid for again.

An illustrative supplement brand shows why. It sells one bottle for $70, and after product, shipping and payment fees it keeps 60%, or $42 per order. Its ads cost $56 per new customer, so the first order loses $14. Every reorder adds $42 with no ad cost behind it. Here is what one customer is worth as the orders add up, against what it cost to win them:

What one customer is worth as orders add up

Illustrative brand: $70 per order, $42 contribution per order, $56 to acquire

Cost to acquire (CAC)$56.00

After 1 orderstill $14 short$42.00

After 2 orders$84.00

After 2.4 orders (first year)$100.80

After 3 orders$126.00

The customer pays back during the second order, and everything after that is profit before fixed costs. Whether this brand makes money depends on how many buyers reach a second and a third order.

Retention metrics: what to track and how to calculate them

Five numbers cover it. Fix the window for each one (60 days, 90 days, 12 months) and keep it the same from month to month, or the comparisons mean nothing.

For the brand above, $70 × 60% × 2.4 orders gives $100.80 of lifetime contribution per customer. Three habits keep these numbers honest:

• Use the median for time to second order. A handful of buyers who come back after a year drag an average far out; the median tells you when a typical buyer reorders.
• Count LTV as contribution, not revenue. A revenue figure flatters a low-margin product. The contribution margin post shows what to subtract before you multiply.
• Count only what you have observed. A 12-month LTV for a brand that is five months old is a forecast. Label it as one, and replace it with real orders as the months come in.

How to build a cohort table (worked example)

A cohort is everyone whose first order fell in the same month. A cohort table puts one cohort per row and the months since that first order across the top, so you can follow each group as it ages. Export first-order dates and every later order from your store, group the orders by customer, and fill the grid.

Here is an illustrative brand selling a 30-day supplement. Each cell is the share of the cohort that ordered again in that month after their first order:

Read it two ways. Across a row, you watch one group fade: January's monthly reorder share falls from 24% to 13% over four months. Down a column, you compare cohorts at the same age, and that's where problems show. March's Month 1 is 19% against 23% to 25% for its neighbors. Something happened to March's buyers: a stockout that delayed their reorders, a delivery problem, or a new ad angle that brought in people who wanted something else. April's 25% suggests whatever it was got fixed. A single store-wide repeat rate would have blurred all of that into one number.

Cohorts also tell you what to compare. Split them by first product or bundle size, by traffic source and by ad angle. Different ads bring different buyers, and a cheap first order from the wrong angle can cost more than it looks once you see its Month 1.

Bundle size changes the shape of the curve. A 3-bottle buyer doesn't need a second order until the first three months are up, so judging them at day 60 makes them look disloyal:

Buyers with a second order, by days since the first

Illustrative cohort: 1-bottle and 3-bottle buyers

• Bought 1 bottle
• Bought 3 bottles

0%12.5%25%37.5%50%Day 0Day 30Day 60Day 90Day 120Day 150Day 1801 bottle runs out3 bottles run out40%42%

Days since the first order

Show the numbers

Days since the first order	Bought 1 bottle	Bought 3 bottles
Day 0	0%	0%
Day 30	14%	3%
Day 60	25%	8%
Day 90	31%	26%
Day 120	35%	34%
Day 150	38%	39%
Day 180	40%	42%

In this example the 3-bottle buyers end up reordering slightly more often, 42% against 40% by day 180, yet at day 60 they look three times worse, 8% against 25%. Measure each bundle size on its own window: about one supply cycle plus delivery, then a month more.

What retention is worth: LTV against CAC

Put the brand's numbers into the calculator. Say the January cohort, followed for a full year, placed 2.4 orders per customer. With your own data, use a number you have observed like this, not one you hope for:

Calculator

LTV to CAC calculator

What a customer is worth over all their orders, against what it costs to win them.

Average order value$Contribution margin (after product, shipping and fees)%Orders per customer over their lifetimeCost to acquire a customer (CAC)$

LTV : CAC

1.8 : 1

Lifetime profit per customer (before ads)

$100.80

First order after paying for the customer

−$14.00

Orders it takes to earn the CAC back

1.3

LTV here is lifetime contribution — revenue minus product, shipping and fees — not revenue. Use a lifetime you have actually observed, not one you hope for.

Reset to the example

At 2.4 orders, each customer brings $100.80 of contribution against $56 to acquire, a ratio of 1.8 to 1, with the first order underwater by $14. Now change the orders field and watch what retention is worth: every extra 0.1 orders per customer adds $4.20 of contribution per buyer. Lifting the first year from 2.4 to 2.7 orders adds $12.60 per customer, or $12,600 on a cohort of 1,000. You can keep that money, or spend it on a higher CAC to buy more customers at the same ratio. What ratio to aim for, and why "3 to 1" is a convention rather than a rule, is covered in the LTV post linked below.

The CAC side of the ratio comes from your ad accounts and your attribution. In RevvyScales, profit tracking (Growth plan and up) subtracts ad spend from a connected Meta account, COGS and gateway fees from revenue, and shows ROAS and margin by campaign, so the cost and the margin of each campaign's first orders sit side by side (profit dashboard). The repeat orders live in your store, so join the two by customer to build cohorts by campaign.

Customer retention strategies for ecommerce: six levers

Each lever below works on one reason buyers don't come back, and each comes with the number that shows whether it's working.

1. Replenishment timing

A consumable is reordered when it runs out, so the reminder has to arrive before that day, timed from the quantity each buyer ordered. The post-purchase email flow post has the formula and the emails. Then check it against your data: if the median second order for a 30-day bottle comes on day 41, buyers are running out before they reorder, and some of them won't come back.

2. Subscriptions with clear terms

A subscription removes the reorder decision. It only helps if it's honest. Under ROSCA, a seller charging through a negative option online must clearly disclose all material terms before taking billing information, get the buyer's express informed consent before charging, and provide a simple way to stop the charges (15 U.S.C. 8403). State laws add more: California's Automatic Renewal Law, as amended for contracts made on or after July 1, 2025, requires among other things that a consumer who signed up online can cancel online (AB 2863, California Legislative Information). The auto-renewal laws post covers the details. None of this is legal advice.

Some subscribers leave without deciding to: their card expires or is replaced, and the renewal fails. Card updater services exist for that. Visa Account Updater exchanges updated account numbers and expiration dates for stored-credential and recurring payments (Visa), and Mastercard's Automatic Billing Updater helps keep card-on-file and recurring payments working when those details change (Mastercard). Ask your processor whether your account uses them, and track renewals lost to failed payments separately from cancellations, so you know which problem you have. The supplement subscription post covers the offer itself.

3. Onboarding that gets the product used

A bottle that isn't opened never runs out, and a 30-serving bottle taken twice a week lasts more than three months instead of one, so every reminder reaches a buyer with most of it left. Make taking it easy: clear directions on the label and on an insert in the box, a how-to-use email on delivery, a suggested cue (next to the coffee maker, with breakfast), and a check-in after a week or so that asks how it's going. Count the replies that say "I keep forgetting" or "I wasn't sure how to take it". Each one points at a fix.

4. Loyalty that isn't only discounts

Discounts pull orders forward and cut the margin on reorders you would have had anyway. A 15% discount takes $10.50 off a $70 reorder, and with the same costs per order its contribution falls from $42 to $31.50, a quarter of the profit, so the discount has to produce a third more reorders just to stand still:

Contribution from reorders, full price against 15% off

Illustrative: $70 reorder, $42 contribution at full price

100 reorders at full price$4,200

100 reorders at 15% off$31.50 each$3,150

134 reorders at 15% offwhat it takes to match$4,221

Better reasons to come back cost less: a lower per-bottle price on bigger packs, free shipping for repeat orders or subscribers, early access to a new product, and a person who answers when something goes wrong. If you run a sale, make the deadline real.

5. Support that answers fast

A buyer with an unanswered problem doesn't reorder, and may ask their bank for the money back instead of asking you. Answer within one business day, refund quickly when your policy says you will, and track the reasons people contact you. A refund you offer costs less than a dispute the buyer files; the reduce chargebacks post explains why and what else prevents them.

6. Shipping that keeps its promises

Late deliveries break the routine you are trying to build. The FTC's Mail Order Rule requires a reasonable basis for any shipping time you state, shipment within 30 days when you state none, and, when you can't ship on time, an offer to accept the delay or cancel for a prompt refund (16 CFR Part 435). State the window you actually hit, send delay notices before buyers ask, and keep stock for the cohorts about to run out. A stockout at that moment shows up later in your cohort table, like March's weak Month 1. The Mail Order Rule post covers refunds and deadlines.

What to fix first

Start from what your numbers show, not from the lever you like best:

Work in this order: use (is the product being taken?), timing (does the reminder arrive before the supply runs out?), promises (shipping, support, the charge on the statement), and only then offers. Offers come last because they cost margin on every reorder, including the ones you would have had anyway.

A monthly retention review

Retention changes slowly, so a fixed monthly routine catches more than a burst of attention after a bad month. Here is one to run on the first working day of each month:

Monthly retention review

• Add last month’s first-time buyers to the cohort table, with every later order from each cohort.
• Compare each cohort’s Month 1 with the cohorts before it, and explain any outlier.
• Check median time to second order against days of supply, for each bundle size.
• Recalculate orders per customer and LTV from observed orders only.
• Read the check-in replies and support tickets for problems that repeat.
• Check stock against the reorders expected from the cohorts about to run out.
• Review failed subscription payments and what your processor does to recover them.
• Change one lever at a time, and note the date, so the next cohort shows its effect.

0 of 8 done

Write down what you changed and when. Next quarter's cohort table will tell you whether it worked, but only if you can line the change up with the cohort it first touched.