Second Purchase
by
Second order conversion for ecommerce teams: find where first-time buyers stall against your payback period, then pick the cohort offer and the channel that closes it.
Most ecommerce businesses do not make money on a first order. The customer becomes profitable on the second one, which makes the gap between order one and order two the single most valuable window in the lifecycle, and the one most often handled with a generic welcome flow. The window is not the same for every cohort either: a consumable repeats in six weeks and a considered purchase in six months, so one reminder cadence is wrong for both. This skill measures where each acquisition cohort stalls against your payback period and picks the offer and channel that closes the gap.
What It Does
- Measures first-to-second purchase rate and interval per acquisition cohort.
- Compares each cohort's payback point against its actual repeat behavior.
- Finds the point in the gap where the cohort stops converting.
- Sizes the offer against the margin the second order carries.
- Picks the channel that reaches each cohort at the moment it stalls.
- Returns the cohort offer and channel with the expected payback shift.
Use Cases
- Stop sending one welcome cadence to cohorts that repeat at different speeds.
- Find the acquisition source whose customers never place a second order.
- Time the second-order nudge to where the cohort actually stalls.
- Judge a channel on payback rather than on first order CAC.
Example Session
Which cohorts are not making it to a second order, and what should we send them?
Four cohorts from the last quarter. Two pay back, one stalls, one never will.
- Cohort | CAC | 2nd order rate | Median gap | Payback | Action
- Paid search, non-brand | $31 | 44% | 38 days | Order 2 | No offer needed
- Meta prospecting | $46 | 26% | 71 days | Order 3 | Nudge at day 45, 10%
- Marketplace | $22 | 9% | 140 days | Never | Stop acquiring here
- Referral | $12 | 51% | 29 days | Order 2 | Scale acquisition
The Meta cohort stalls at day 45 and converts at 26%, so the nudge goes there rather than at the generic day 30. Marketplace is the finding that matters: a $22 CAC looks like your cheapest acquisition and 9% of those customers ever return, which means it never pays back and no second-order offer will fix it.
Why This Beats Prompting It Yourself
One Welcome Flow vs Where The Cohort Stalls: The Hidden Risk
A generic prompt writes a post-purchase sequence on a standard cadence, usually day 7, 14 and 30. That cadence is right for a consumable and wrong for anything bought seasonally, and it sends the same offer to a cohort that repeats on its own and one that never does. The flow performs adequately on average and misses every cohort specifically.
When You Still Need Payback In The Picture
A second order is worth whatever it does to payback. This skill states 26% conversion at a 71 day median against a $46 CAC that needs order three, which is what makes the day 45 nudge worth its discount. It also names the cohort that never pays back, which is an acquisition decision rather than a lifecycle one and the more valuable finding of the two.
Known Limitations
Cohort payback needs acquisition cost attributable at source level and enough elapsed time for repeat behavior to show, so a cohort under one purchase interval old is reported as incomplete rather than scored. Customers who first bought as guests under another email will appear as new, which understates repeat rates. It picks the offer and channel and does not build the flow in your ESP.
Install It
- Download the ZIP.
It is free and there is no account to create. - Unzip it into your agent's skills folder.
Claude Code reads ~/.claude/skills/, which is hidden by default: the command in the folder block opens it. Other agents scan their own directory, so drop the same folder there instead. - Ask your agent to use it.
Restart the agent if it was already running, then it picks the skill up with no config.
skills folder
Copy
~/.claude/skills/second-purchase/
SKILL.md
references/payback.md
cohorts/
# one file per review, per cohort
# macOS: create the folder and open it in Finder
mkdir -p ~/.claude/skills && open ~/.claude/skills
# Windows: paste in the Explorer address bar
%USERPROFILE%\.claude\skills\Frequently asked questions
How is this different from a post-purchase flow?
A flow runs on a fixed cadence for everybody. This measures where each acquisition cohort actually stalls, compares that against the order at which the cohort pays back, and picks the offer and channel for that specific gap. Some cohorts get nothing, and one usually gets an acquisition decision instead.
Which agents does the skill run in?
Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision.
What does it need connected to work?
Order history with customer identity, and acquisition cost by source. Without acquisition cost it still measures repeat rates and intervals, and says that the payback comparison is missing, which is the part that makes the offers worth sizing.
Why would a cohort get no offer?
Because a cohort converting at 44% within its payback order is already working, and discounting it spends margin on orders you were going to get. The offer goes where the stall is, which is usually one or two cohorts rather than the whole file.
What if a cohort never pays back?
Then it is an acquisition problem, not a lifecycle one, and the skill says so. No second-order offer fixes a source whose customers do not return, and the honest recommendation is to stop buying there rather than to keep discounting into it.
What is included with the skill?
The skill itself, the payback reference, the cohort definitions, and the offer sizing rules. It is free and security scanned. We re-publish the ZIP when the payback model changes, so download it again if your cohort reads shift.
The second order is where a customer becomes profitable.
Polar reads first-to-second purchase behavior against CAC on one definition, so the cohort offer is aimed where it pays back.
<script type="application/ld+json">{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How is this different from a post-purchase flow?","acceptedAnswer":{"@type":"Answer","text":"A flow runs on a fixed cadence for everybody. This measures where each acquisition cohort actually stalls, compares that against the order at which the cohort pays back, and picks the offer and channel for that specific gap. Some cohorts get nothing, and one usually gets an acquisition decision instead."}},{"@type":"Question","name":"Which agents does the skill run in?","acceptedAnswer":{"@type":"Answer","text":"Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision."}},{"@type":"Question","name":"What does it need connected to work?","acceptedAnswer":{"@type":"Answer","text":"Order history with customer identity, and acquisition cost by source. Without acquisition cost it still measures repeat rates and intervals, and says that the payback comparison is missing, which is the part that makes the offers worth sizing."}},{"@type":"Question","name":"Why would a cohort get no offer?","acceptedAnswer":{"@type":"Answer","text":"Because a cohort converting at 44% within its payback order is already working, and discounting it spends margin on orders you were going to get. The offer goes where the stall is, which is usually one or two cohorts rather than the whole file."}},{"@type":"Question","name":"What if a cohort never pays back?","acceptedAnswer":{"@type":"Answer","text":"Then it is an acquisition problem, not a lifecycle one, and the skill says so. No second-order offer fixes a source whose customers do not return, and the honest recommendation is to stop buying there rather than to keep discounting into it."}},{"@type":"Question","name":"What is included with the skill?","acceptedAnswer":{"@type":"Answer","text":"The skill itself, the payback reference, the cohort definitions, and the offer sizing rules. It is free and security scanned. We re-publish the ZIP when the payback model changes, so download it again if your cohort reads shift."}}]}</script>