One engagement
Ninety days to rebuild your second order, and a holdout to read it.
The Second-Purchase Sprint rebuilds the path from a customer’s first order to their second, on your own Klaviyo account, and measures it against a holdout designed before anything is built. The second order is the gate to every order after it: ninety days rebuilds the path to it and starts the test, and lifetime value is what compounds behind it. It is the one engagement I sell, and the free audit comes first.
Why ninety days
Second orders take weeks to arrive.
A test that ends before they do can’t read them. Ninety days covers the Read, the build and the test. Late buyers still need their full window. Day 90 ends the engagement. It doesn’t move the measurement window.
How it runs
Diagnose, build, test.
- Days 1 to 14The Read
Your repeat rate, your time to a second order and what your one-and-done customers are worth, computed from your own export. Revenue per recipient for every flow. The holdout, designed and registered before anything is built.
- Days 15 to 45The build
Post-purchase rebuilt around your real second-order window, the offer that asks for the second order, win-back for the buyers who drift, and SMS where it earns its place. Built live in your account and documented for whoever runs it next.
- Days 46 to 90The test
The build runs against the registered holdout, unchanged, so the result means something. I watch exposure, exclusions and sample growth, and I don’t call it early. You get the twelve-month plan at day 90. The result is read when the registered window closes, even if that falls after day 90. The difference is measured, not attributed.
The first fourteen days
The Read.
The diagnosis you keep whatever happens next. It is the first $5,000 of the Sprint, and it decides whether the rest is worth doing.
- 01Your repeat rate and median days to a second order, by first product.
From your own order export, not a benchmark.
- 02What your one-and-done customers are worth.
The size of your gap in dollars, with the arithmetic shown.
- 03Revenue per recipient for every flow.
Which flows do the work and which only collect credit.
- 04Your two calendars, side by side.
When you send, against when your customers actually come back.
- 05The holdout design, written down.
Who is held back, for how long and what counts as a win, registered before launch.
- 06A straight go or no-go.
Whether your gap is worth the next seventy-six days.
The guarantee
The Day-14 Exit.
If the Read shows your second-purchase gap isn’t worth fixing, we stop at day 14. You keep the diagnosis, the numbers and the holdout design, and you owe nothing more. Most engagements ask you to trust them for three months. This one has to earn the next seventy-six days first.
Terms
What it costs, and what you keep.
- How you pay
- $5,000 to start, which covers the Read. Then $7,500 when you green-light the build at day 14, and $7,500 at day 45.
- Founding rate
- $12,500 for the first three brands, in exchange for permission to publish your holdout result on the Holdout Registry when the cohort matures, whatever it shows. You pay the first two installments and skip the third.
- What you keep
- The cohort model, the flows, the holdout design, the registered readout and the twelve-month plan. Yours whether or not we keep working together.
- After day 90
- Keep me on at $7,500 a month to run it, hand it to your team, or stop. Month to month.
- What it’s worth
- An illustration, not a client: at 12,000 first-time buyers a year and a $120 order, five points of repeat rate is $72K a year. At that size, a 10% holdout needs about six months of new buyers to read five points. Run your own numbers.
- Capacity
- Four Sprints a quarter, so each one gets the attention a holdout needs.
Fit
Who it’s for, and who it isn’t.
A good fit
- Shopify or subscription brands on Klaviyo.
- Enough first-time buyers to power a holdout. The Read calculates it from your account.
- A product people can plausibly buy twice.
- Someone on your side who owns the account and wants the truth about it.
Not a fit
- A single campaign, a redesign of your templates or a one-off email.
- Too few first-time buyers for a holdout to read. Start with the free audit instead.
- Wanting attributed revenue for a board deck without a control group.
- Looking for someone to send more emails, faster.
Proof, honestly
The best number I have can’t prove itself.
Every figure on this site says how it was measured. This one is real and useful, and it is still only a before-and-after. That is exactly why every Sprint builds the holdout first.
+38%Repeat purchase over about six months · Magic Spoon
Before and after · no control group · relative or points not yet publishedQuestions
What people ask before they apply.
$20,000 is a lot.
It is, which is why you only risk $5,000 of it first. The Read tells you the size of your gap in dollars. If it isn’t worth fixing, you stop at day 14. For scale: at 12,000 first-time buyers a year and a $120 order, five points of repeat rate is $72K a year. That is an illustration, not a client; run your own numbers in the calculator.
We already have an agency.
Good. The Sprint gives them a measured target and a documented program, and the holdout tells both of you what is actually working. When a Sprint ends, your agency can run what was built. It’s documented for that.
Why ninety days, not thirty?
Second orders take weeks to arrive. A test that ends before they do can’t read them. Ninety days covers the Read, the build and the test. Customers who enter late still need their full window, so the result is read when that window closes, not on day 90.
Why a holdout? Klaviyo already attributes revenue.
Attribution credits the email for customers who would have bought anyway. A holdout keeps a random slice of customers out of the program, so the difference between the two groups is the program. It makes the win smaller on paper and much harder to argue with.
How many customers do I need?
There is no fixed minimum. It depends on your baseline, the lift worth detecting, the share held out and the window. At a 20% second-order rate, a three-point lift needs about 5,900 first-time buyers with half held out, 9,300 with a fifth and 16,600 with a tenth (80% power, two-sided 0.05). The Read calculates it from your own account. If your volume can’t carry a holdout, I say so, and I don’t call an underpowered test proof.
We sell B2B.
The same problem shows up as the reorder. Bring it to the fit call. If your account count can’t carry a holdout, I will say so before you spend anything.
Apply
Apply for a Sprint.
Seven short questions. I read every application myself and reply within two business days, with a time for a fit call or an honest no.
Rather talk first? Book a 30-minute fit call.