# Customer retention rate for a reorder business: measure it against each customer&#8217;s own purchase cycle

https://salesbond.app/blog/customer-retention-rate-reorder-business-by-purchase-cycle/
Published: 2026-11-03 · Updated: 2026-11-14

Customer retention rate is the share of customers you still have at the end of a period, `Retention rate = (customers at the end − new customers) / customers at the start × 100`. Repeat purchase rate is the share of customers who bought in a window and bought more than once, `Repeat purchase rate = customers with 2+ purchases / all customers who bought × 100`. If your clients reorder on their own rhythm instead of subscribing, retention mislabels slow buyers as lost, so also check each customer against their own purchase cycle.

The version we use with distributors, service firms and B2B shops measures retention against the interval each customer has shown you in their own won deals.

## Customer retention rate formula, with a worked example

`Retention rate = (E − N) / S × 100`, where S is the number of customers at the start of the period, E the number at the end, and N the number of new customers won during the period. This is the standard form you will find in most guides, for example [Business News Daily’s walkthrough of the customer retention rate formula](https://www.businessnewsdaily.com/16027-customer-retention-rate.html). Subtracting N matters: without it, a strong month of new business hides the customers who left.

A made-up example, so the arithmetic is visible. On 1 January you had 200 customers. During the quarter you won 30 new ones. On 31 March you count 210 active customers. `Retention rate = (210 − 30) / 200 × 100 = 90%`. Twenty of the original 200 did not buy in the quarter.

Those twenty are where the trouble starts, because “did not buy in the quarter” and “left” are the same thing only when buying is a subscription.

## What counts as a “retained” customer when buying is not a subscription

In a subscription business the definition is given to you: retained while the contract is active, lost on the day it is cancelled. The formula was built for that world, which is why most examples you find are SaaS and ecommerce.

A reorder business has to define “retained” itself, and that choice moves the number more than anything else in the calculation. Three definitions we see in practice:

- bought at least once in the period (the usual choice, and the one that produces false losses);
- bought within some fixed window such as the last 12 months, regardless of the reporting period;
- bought within their own usual interval, give or take.

The first is wrong for anyone whose cycle is longer than the period. The second can’t tell a monthly buyer silent for ten months from an annual buyer who simply isn’t due yet. We use the third. A customer counts as retained until they have gone more than twice their own purchase cycle without an order, and as at risk once they pass the cycle itself.

## Where the calendar formula breaks: the customer who buys every six months

Picture a packaging distributor, invented for the example. One client orders pallets of film every April and October and has done so for four years. Run the quarterly formula in Q3: they bought nothing, so they count as lost. In Q4 they return and count as a new customer, unless someone remembers to exclude them from N by hand.

Multiply that by every customer whose cycle is longer than your period and the metric swings with the calendar rather than with the business. Teams respond by lengthening the the period to a year, which fixes the false losses and creates the opposite blindness: a weekly buyer who went silent in February still counts as retained in December.

One more failure. The formula counts heads, so a client that used to order 40 pallets and now orders 4 is fully retained. That is a revenue question, covered separately in [gross and net revenue retention for a reorder business](https://salesbond.app/blog/gross-revenue-retention-pipedrive/).

## Retention against the purchase cycle: on time, late, gone

The fix is to stop asking “did they buy this quarter” and start asking “are they late by their own standards”. For that you need two numbers per customer.

The first is the purchase cycle: how many days usually pass between this customer’s orders. Take the dates of their won deals, compute the gaps, and take the median. Why the median and not the mean? One unusual project order between regular reorders drags an average around, and the median ignores it. Two purchases give one interval, so a customer with one order has no cycle yet; they are a new customer and a different question.

The second is days since the last purchase. Divide it by the cycle and you have a ratio that says how late the customer is by their own rhythm:

- ratio up to 1.0: on track, the next order is not due yet;
- ratio above 1.0: overdue, the customer has passed their usual interval;
- ratio above 2.0: leaving, they have skipped a full cycle and then some.

The thresholds are a choice you make. A distributor with weekly cycles may call a customer overdue only at 1.3 because a few days of noise is normal, while a firm with annual contracts may move leaving down to 1.5 because a renewal half a year late is already gone. What to do with the overdue accounts once you have them is a separate topic: [customer retention strategies for B2B](https://salesbond.app/blog/customer-retention-strategies-salesbond-customer-health-late-reorders/).

Cycle-based retention rate is the share of repeat customers whose days since last purchase are within twice their median purchase interval, `Cycle-based retention rate = (on track + overdue) / customers with a cycle × 100`. It doesn’t depend on the calendar. It moves when a customer goes quiet for longer than they ever have, which is exactly the moment you wanted to know about.

Our six-month packaging client, checked in September, has a cycle of about 180 days and last bought 150 days ago. That is a ratio of 0.83, so the client is on track and simply not due yet.

![Timeline of one customer with on-track, overdue and leaving zones; today, day 150 of a 180-day cycle, is on track](https://salesbond.app/wp-content/uploads/2026/11/customer-retention-rate-reorder-business-by-purchase-cycle-2.png)

## Retention metrics that go with it: repeat purchase rate, historical LTV, average order

A retention rate answers one question: how many did we keep. Three neighbouring numbers answer the ones that follow.

Repeat purchase rate is the share of customers who bought in a window and bought more than once. Say 400 customers bought from you last year and 140 of them ordered at least twice, so `Repeat purchase rate = 140 / 400 × 100 = 35%`. Many reorder businesses lose most customers between the first and the second order, and the retention formula, which only counts customers already there at the start, never sees that cliff.

Historical LTV is the actual sum of a customer’s won deals to date, with no projection in it. It lets you rank overdue customers by what they are worth instead of treating a one-pallet client and a forty-pallet client as the same lost head. The formulas and the CRM version are in [customer lifetime value: the formula and the historical LTV in your CRM](https://salesbond.app/blog/customer-lifetime-value-formula-historical-ltv-crm/).

Average order value: historical LTV divided by number of purchases. A base that is fully retained on heads but falling on average order is shrinking, and the retention rate will not tell you.

Together these turn retention from a report into a list. [Account management by customer status](https://salesbond.app/blog/account-management-by-customer-status-repeat-sales/) shows how that list becomes a weekly routine.

## How to calculate customer retention rate from won deals in your CRM

You do not need a data warehouse, only complete won deals.

1. Export won deals with four columns: organisation (or customer), close date, value, owner. One row per deal.
2. Sort by organisation and close date. For each organisation with two or more rows, compute the gaps between consecutive close dates and take the median. That is the cycle.
3. Compute days from the most recent close date to today, divide by the cycle, and label the ratio with your thresholds.
4. Count the labels. `Retained = on track + overdue`, `at risk = overdue + leaving`, `cycle-based retention rate = retained / customers with a cycle × 100`.

The first time you do this, three things will go wrong, and none of them is maths.

Customers with one purchase have no cycle. They can’t be in the metric, and if they are most of your base, that is your finding: a second-order problem, which comes before retention. Keep them in a seperate count.

Reorders that never became deals. Some teams log a repeat order by editing the old won deal, or skip the CRM because the customer emails the warehouse directly. Then the CRM shows one purchase for a customer who has bought twenty times. One won deal per order is the prerequisite; [repeat orders in Pipedrive: one new deal per order](https://salesbond.app/blog/repeat-orders-pipedrive-new-deal-per-order/) walks through the habit, and the logic is the same in any CRM.

Organisations that are not linked. Deals for “Acme”, “ACME Ltd” and “Acme (new contact)” are three customers with one purchase each instead of one customer with a clear 60-day rhythm. Clean the organisation field before you trust the number.

## Examples: a distributor, a service contract renewal, a B2B shop

All three are made up, chosen to show the same number behaving differently.

The distributor sells consumables to 400 accounts. Most cycles sit between 20 and 45 days with high confidence, because there are dozens of intervals behind each one. Here a monthly calendar rate is almost usable and the cycle version mostly adds timing. Take a customer on a 21-day cycle: at 30 days of silence they are overdue (ratio 1.4), and at 45 days they are leaving (ratio 2.1), while a calendar rate that only asks “did they buy this quarter” still counts them as retained.

The service firm renews maintenance contracts once a year. Cycles sit around 365 days, many at low confidence because a client has renewed only twice. Any retention period shorter than a year is meaningless here, and even an annual rate can lag by up to twelve months. The cycle version turns the question into a date: this contract’s ratio crosses 1.0 on 14 March, so the renewal conversation happens in February. Keep the thresholds tight, since a renewal 60 days late has usually gone to someone else.

The B2B shop sells equipment to small businesses, and of its 1,200 customers perhaps 300 have two or more orders, so a cycle-based retention rate applies to those 300 only. The other 900 are a repeat-purchase question and sit outside churn.

## How Repeat Sales reads retention: Overview, Customers, Health, New

Repeat Sales, the SalesBond product for repeat business that we build, works from the CRM’s won deals on the same basis as the manual method: a purchase is a won deal, the cycle is the median interval once a customer has at least two purchases, and each cycle carries a confidence of high, medium or low. Days since the last purchase divided by the cycle gives the status: on track, overdue above 1.0, leaving above 2.0, with both thresholds configurable. It is computed per line, customer by seller, so a manager sees the account and the rep behind it. Health is the status view, and New lists the customers with a single purchase so far.

Two honest limits. Repeat Sales does not print a retention rate as a single (E − N) / S percentage; it gives you the statuses and you compute the share. And there is no machine-learning churn prediction, only the deterministic rules on purchase dates. SalesBond works with Pipedrive, HubSpot, Salesforce, Zoho CRM, Microsoft Dynamics 365, Freshsales, Close, monday CRM, Copper, Zendesk Sell, Nutshell, Insightly, Attio, Bitrix24 and Kommo (amoCRM).

If you want to see the state of your base before building any of this, start with the free CRM audit. SalesBond looks at the last 30 days of the connected CRM and sends a PDF report; repeat customers and top-10 clients are two of its sections. Seven days later a second PDF arrives showing what changed, if the team has worked on the findings.

[Check your CRM with SalesBond](https://salesbond.app/)

## FAQ

**Retention rate vs churn rate: what is the difference?** They are complements of each other over the same period: churn rate is the share of starting customers you lost, retention rate the share you kept. In the 200-customer example, 90% retention means 10% churn. In a reorder business both inherit the same flaw, because “lost” is defined by the calendar rather than by the customer’s cycle.

**What is a good customer retention rate? Is 80% good, or 50%?** First read the percentage literally: 80% means that 20 of every 100 customers you started the period with did not buy during it. Whether that is good depends on your definition of retained and on the length of your purchase cycle, so there is no number that is good across businesses. A 50% quarterly rate can be perfectly healthy for a base that reorders twice a year. Compare your own rate across periods with the same definition, and compare the cycle-based version across customer segments. A benchmark from a SaaS article is not a target for a distributor.

**How often should I measure it?** Calendar retention once a quarter, with the period at least as long as your median purchase cycle. Cycle-based retention weekly, because it is a list of overdue customers as much as a number, and the list is only useful if someone calls the names on it.
