Enrolling Existing Customers at Launch: Retroactive Points and the First 30 Days
Back-credit past orders or start the balance at zero? A launch-time decision for existing buyers: what retroactive points cost in liability, a 30-day email sequence to your existing list, and a holdout test against customers you don't invite yet.
The short version: Grant retroactive points to existing customers for a capped window of recent orders, not full history, and measure the invite sequence against a holdout group you deliberately did not invite.
Key takeaways
- Back-credit a recent window sized to your own reorder cycle, not all history — full-history crediting pays dead accounts and inflates liability.
- Cost the grant before you announce it: every point granted is a liability at face value from launch day.
- Invite in waves, best recent customers first, so the first wave's activity becomes the social proof later waves see.
- Run the first 30 days as automated flows, and keep a not-yet-invited holdout so you can separate program lift from points-gift lift.
Should existing customers earn retroactive points at launch?
Yes for a capped recent window, no for full history. Retroactive points are points granted for orders a customer placed before your loyalty program existed, and granting them to existing customers solves the launch-day problem of everyone starting at a balance of zero.
The cap is what keeps the decision sane. A customer who last ordered three years ago receives a balance they may never redeem, and you carry it on the books anyway. Set the window from your own data — your median days-between-orders, extended to cover one or two full reorder cycles — rather than a round number you saw somewhere.
What back-crediting costs in points liability
Back-crediting is a cash decision disguised as a goodwill gesture, so price it before you announce it. Every point granted is a liability at face value the moment it lands, whether or not anyone redeems it.
Worked example with illustrative numbers: a shop with 2,000 orders in its 90-day window, a $65 average order value, a 5% earn rate, and $0.01-per-point redemption grants 2,000 × $65 × 0.05 = $6,500 in point value on launch day. What you actually reserve depends on your expected redemption rate, which is why the grant belongs in your cash and liability budget before it goes in an email. The classic failure: crediting two years of orders at once — 24 months of orders at the same monthly rate is an eight-times-larger grant, and the oldest balances sit on accounts that may never redeem. If your ledger, holds, and return rules are not ready for that, build the liability controls before the campaign.
Which customers to enroll and in what order
Segment existing buyers by recency and spend, then invite best-first. Your most recent, highest-spending customers get the first wave because their immediate activity — redemptions, reviews, repeat orders — is what later waves see when they decide the program is real.
A workable order: recent high-spend, then recent low-spend, then lapsed buyers as a reactivation wave. The classic failure: blasting the whole list at once. You get a one-day liability spike, a support queue full of "where are my points," and no visible activity for wave two to anchor on. A 30-day implementation plan gives the waves slots beside the rest of your launch work instead of competing with it.
The 30-day email sequence for existing buyers
Four automated emails, built as flows rather than one-off campaigns. Flows fire on timing and behavior you configure once; a blast depends on you remembering to send it, and its send dates drift away from your customers' actual reorder rhythm. Treat the day counts below as an illustrative starting schedule — set your own send days from your median days-between-orders and your open-rate data, not from this list.
Day 0: the enrollment announcement — credit already granted, one line on how the program works. Day 3 or 4: a balance explainer — what the points are worth and what they redeem for, short enough to read on a phone. Day 10: a reorder nudge for the window cohort, tying their balance to something they would plausibly buy again; the win-back playbook covers the framing. Day 21 to 30: a bestseller or back-in-stock push, since restock alerts reach customers already primed to buy. Check each send against your own revenue-per-recipient from the last campaign you ran, so the comparison is your baseline rather than anyone else's.
Measuring the launch against a not-yet-invited holdout
Withhold a slice of your existing list from the launch — 10% is an illustrative starting point; size it against your list so the group is large enough to read a repeat-rate difference — and compare repeat rates at two reads. Anchor those reads to your own reorder cycle: one and two times your median days-between-orders. Thirty and sixty days is an illustrative pairing for a shop that reorders monthly; if your catalog is bought a few times a year, a 30-day read measures nothing, because almost nobody has had time to reorder yet.
The classic failure: crediting the lift to the program when the points gift caused it. A granted balance pulls a reorder forward on its own; the program's real test is whether customers come back again after the gift is spent. If your list is large enough, run a third cell — invited with no back-credit — to separate the two effects, and fold the results into your ROI calculation instead of declaring victory at the first read.
Frequently asked questions
How long should the back-credit window be?
Long enough to cover one or two of your typical reorder cycles, read from your own days-between-orders data. Ninety days is an illustrative starting point, not a standard — and if your product is bought a few times a year, any window mostly rewards accounts that will never redeem, which argues for a shorter one.
What if customers demand points for older orders?
Publish the window in the announcement itself so the rule is visible before anyone asks. Beyond that, handle requests case-by-case through support with a per-customer cap: goodwill for the customer who emails, not a precedent of full history. A written missing-points workflow keeps those decisions consistent instead of ad hoc.
When do you end the holdout?
After the second read — two of your median reorder cycles — or once the repeat-rate gap between invited and holdout groups is large enough to act on. Then invite the holdout with the same offer: the group earned you your measurement, and there is no information left to protect once you have read it.