Growth Notes

Replenishment Timing

Tags: commerce concept

Replenishment Timing

Date: 2026-08-16

Predicting when someone runs out and reaching them just before. For consumables it’s the highest-return lifecycle work available, because the purchase is going to happen — the only question is where.

What it is

Replenishment timing is estimating a product’s consumption cycle per customer and prompting a reorder shortly before it ends.

The distinction that makes it valuable: you’re not creating demand, you’re capturing existing demand at the moment it forms. Which is why it’s so much more efficient than acquisition, and why it needs no discount.

Estimating the cycle

Three sources, in ascending order of accuracy:

1. Product-level median. From your own order data — the median interval between repeat purchases of the same item:

select product_id,
       percentile_cont(0.5) within group (order by days_between) as median_days
from repeat_purchases
group by product_id

2. Adjusted for quantity. Someone who bought three of something needs it three times later. A cycle estimate ignoring quantity is wrong for exactly the customers worth most.

product cycle          45 days per unit
customer bought 3      →  expected reorder ~135 days

3. Per-customer observed. Once someone has repurchased twice, you have their interval, which beats any population estimate. Heavy users differ substantially from the median.

When to send

Slightly early, and the asymmetry matters:

too early    ignored, and mildly irritating
on time      good
too late     they've already bought elsewhere — the order is gone

Running out is the trigger for a purchase, not for a purchase from you. Late is far more expensive than early, so aim for roughly 80% of the estimated cycle.

For a 45-day cycle, prompt around day 36. And prompt once, with at most one follow-up — this is a reminder, not a campaign.

Why it doesn’t need a discount

The customer is about to buy. Adding money off is pure margin loss on a sale you already had — the clearest case of cannibalisation in the whole lifecycle programme.

1,000 replenishment reorders at £50

  no discount     contribution  £15,000
  10% discount    contribution  £10,100   ← £4,900 given away

If the flow underperforms without a discount, the problem is timing or product coverage, not price — Discount Impact on Margin.

Where it fails

  • Not all products replenish. Applying it across a whole catalogue trains people to ignore it. Restrict to genuinely consumable lines
  • Quantity ignored, so bulk buyers get prompted three times too early and stop reading
  • Gifts and one-offs. A customer who bought a gift isn’t going to reorder
  • Blended cycles. A site-wide “90 days since last order” prompt is not replenishment timing, and it’s what most implementations actually do — Time Between Orders
  • Stock. Prompting a reorder for something unavailable converts a captured sale into a lost one plus an annoyed customer — Stockouts and Availability

Where it pays

  • Payback Period. Pulling the second order forward recovers acquisition cost sooner, which raises the growth rate at unchanged lifetime value
  • Purchase Frequency — the cheapest of the three value levers to move
  • Repeat Purchase Rate — the second order is the one that makes a customer profitable

For a health, beauty or household category this is usually the single highest-return piece of lifecycle work available, and it’s frequently not implemented at all — or implemented as a generic time-based email that isn’t this.

Hold out a slice to confirm the reorders are incremental rather than pulled forward from a purchase that was coming anyway — Incrementality Testing, Lifecycle Messaging.

Graph View

Backlinks