How Long Should It Take Customers to Buy Again?
The answer could be:
10 days.
30 days.
90 days.
Three years.
Without knowing what they bought, the number is almost meaningless.
That is why average time between orders needs context.
Second purchase timing is usually front-loaded
BS&Co reports that among repeat customers in its 156,110-customer dataset:
50.3% of second purchases happened within 30 days.
76.4% within 90 days. (bsandco.us)
That gives us a useful broad shape.
But not a universal clock.
Product consumption changes everything
A coffee bag can run out.
A serum can run out.
A couch does not.
Klaviyo’s product-analysis guidance recommends using product-specific purchase timing to determine when cross-sell and replenishment messages should happen. (academy.klaviyo.com)
The product creates the repeat occasion.
Order number changes timing too
RetentionX measures time between:
- order one and two,
- order two and three,
- later purchase stages
separately because the intervals can change as customers mature. (help.retentionx.com)
A customer who becomes habitual may begin purchasing more predictably.
Median versus average
Suppose order gaps are:
25
27
28
30
180
Average:
58 days.
Median:
Which better describes most customers?
The median.
Long-tail purchasers inflate the average.
Use both.
Quantity matters
One 30-day supply:
possible reorder around day 30.
Three 30-day units:
closer to day 90.
If your system ignores quantity, the timing will be wrong for stock-up customers.
Segment the purchase cycle
Calculate timing by:
- first product,
- quantity,
- order number,
- subscription status,
- acquisition cohort.
That creates a much more useful picture than one store average.
When should you intervene?
I would usually think in ranges.
Before expected timing
Education.
Around expected timing
Convenient reorder / recommendation.
Beyond expected timing
Reactivation.
That prevents both premature discounts and very late recovery attempts.
Do not chase the perfect day
Purchase behaviour is a distribution.
Not everyone who usually buys around day 40 will buy exactly on day 40.
Think in windows.
A worked example
Median order-one-to-two timing:
52 days.
25th percentile:
31 days.
75th percentile:
78 days.
Now:
Day 20 = early.
Day 50 = normal.
Day 95 = increasingly overdue.
That is already enough to create better lifecycle logic.
The number should lead to action
If you discover that customers normally reorder in 28 days and your strongest reminder goes out at day 60, change it.
If you discover a 90-day bundle and you are sending at day 30, change that too.
Analytics matters only when it changes the decision.