Most of your file is a decision being made by default. Two short chapters on the shape of a customer file and the stopwatch that misreads it, a tool to select the pool and read the result, and the move.
Up to four in five customers buy exactly once, and a second purchase multiplies a customer’s value several times over. The lever is enormous. So is the waste of pulling it blind.
2 min read · October 26, 2026
The whole file
Inside the one-time buyers
Monday morning
Cutoffs of 30, 60 or 90 days guess at the one event no store can see. The chance each customer is still active, read beside Forward CLV, gives every quiet customer a next step.
2 min read · October 26, 2026
Monday morning
Know which of your one-time buyers are worth converting, free, by tomorrow morning.
The play pays only when it is pointed at the right customers, run as one clean flow, and read against a group that got nothing. Three steps, all on this page, on your own numbers.
Five first-order signals, one point each. Customers scoring three or more are your pool. Build it as a segment this week.
Paid full price on the first order
Discount-first buyers repeat at a fraction of the rate, and mostly on discount.
First order above your median basket
Basket size travels with intent and with second-purchase value.
Bought within the last 90 days
The window where a second purchase is winnable is early.
First order was not returned
A kept order signals fit, and fit predicts return visits.
Arrived organically: brand search, direct, referral
Channel quality persists. Deal-feed and coupon-site arrivals behave like the tail.
The pool, as a Klaviyo segment
Three or more signals
Placed Order equals 1 over all time AND Placed Order at least once in the last 90 days AND Discount Code Used is not set on the first order AND first order value is greater than [your median basket] AND Refunded Order is zero times over all time AND Attributed channel is one of: brand search, direct, referral
The scorecard is backward-looking and unweighted. It will find part of the slice and include some of the middle. It is still far better than sending to everyone.
Klaviyo build sketch
Trigger: segment entry, first order 21 days or older. Random sample split 85 to 15, flow against holdout, tag both. Email one. Wait 11 days, conditional on no purchase. Email two. Exit at day 14. The holdout branch receives nothing and keeps its tag.
Pick the read date before launch, long enough for the offer to expire and late purchases to land. Then put both groups in here. Per-customer value against the holdout is the number that cannot lie to you.
Added per customer, against held-back
+$1.50
$5.30 per customer in the flow, $3.80 held back. Conversion 7% against 5%.
95% range
+$0.13 to +$2.87
Verdict
Scale
The whole range clears zero. The flow added value its customers would not have produced on their own. Scale it to the full pool and re-read next quarter.
A standard two-group comparison on each group’s conversion rate and average order value. Read it on the date you set before launch, not the first good week.
Hand-built selection finds part of the slice. Tacet scores every customer and runs the playbook in your own Klaviyo, read against the customers it held back.
Every customer scored, one-time buyers included.
Forward CLV and the chance each customer is still active, re-scored every night, so the slice worth converting and the quiet customer worth reviving are named rather than guessed.
The second-purchase and win-back playbooks, built in your Klaviyo.
Drafted under your sender, filled every night with the customers worth emailing, timed to each customer’s own rhythm. Your team’s only step is GO.
Read against a held-back group, every week.
About one in five of each playbook’s audience never gets it. The result is the profit the rest made that they did not, with its range. A playbook that does nothing is retired.