Email Format
by
Email creative testing for ecommerce teams: find which element of your sends is actually driving revenue, then design the next test worth running.
Email testing collapses into subject line testing, because subject lines are easy to vary and opens are easy to measure. Both of those are now the weakest signals available: open tracking has been unreliable since mail privacy protection started pre-fetching images, and the subject line is rarely what decides whether someone buys. The elements that move revenue, the layout, the number of products shown, where the call to action sits, almost never get tested. This skill finds which element is actually driving revenue and designs the next test worth running.
What It Does
- Reads performance by structural element rather than by subject line alone.
- Measures revenue per recipient instead of open and click rates.
- Checks the list has the volume to conclude the test being proposed.
- Separates a creative effect from the segment the send happened to go to.
- Designs the variant, including what to hold constant.
- Recommends the next test, with what it is worth and how long it needs.
Use Cases
- Move beyond subject line tests to the elements that change revenue.
- Find out whether your long template beats the short one for this list.
- Stop testing on segments too small to produce a result.
- Get a variant spec rather than an idea to argue about.
Example Session
What should we test in email next, and what will it tell us?
Stop testing subject lines. Product count is the element that moves this list.
- Element | Variants run | Best vs worst | Revenue effect | Next test
- Subject line | 34 | 4% on opens | Not detectable | Stop
- Products per send | 3 | 1 vs 6 products | $0.41 per recipient | Test 1 vs 2
- CTA placement | 2 | Above vs below fold | $0.09 per recipient | Retest after layout
- Send from name | 1 | Untested | Unknown | Test after product count
You have run thirty-four subject line tests and none has produced a detectable revenue difference, which is the answer: the subject line is not the constraint on this list. Product count is, and it is worth $0.41 per recipient between one product and six. The next test narrows that further rather than repeating it: one product against two, which is where the curve probably turns.
Why This Beats Prompting It Yourself
Open Rate vs Revenue Per Recipient: The Hidden Risk
A generic prompt suggests testing subject lines, because that is what email testing means in every guide written before mail privacy protection. Opens are now inflated by pre-fetching and were never a revenue signal anyway. Thirty-four tests later you have a lot of results and no change in what the list earns.
When You Still Need To Know The List Can Conclude It
Small lists cannot detect small effects, and most email tests are run on segments far too small for the difference being looked for. This skill states the volume needed before the test starts, which is how a program stops accumulating inconclusive results that everyone quietly interprets in whichever direction they already preferred.
Known Limitations
Element-level attribution is strongest where you have varied that element across several sends, and a never-varied element is reported as untested rather than scored. Open-based metrics are used only where nothing better exists and are labeled when they are. It designs the test and does not build or send the campaign.
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/email-format/
SKILL.md
references/element-effects.md
tests/
# one file per test, with the variant spec
# 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/B testing subject lines?
Subject lines affect opens, and opens are both unreliable and not revenue. This scores the structural elements of your sends against revenue per recipient, which usually shows that the element you have never tested matters more than the one you have tested thirty-four times.
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?
Send history with the creative variants used, clicks, and orders attributed to sends. Without order attribution it falls back to clicks and labels the whole analysis as a click optimization, which is a different and weaker thing.
Why stop testing subject lines entirely?
Because thirty-four tests with no detectable revenue effect is itself a conclusion. If the subject line mattered on this list it would have shown by now. Continuing to test it is spending your testing capacity on the element that has already answered.
How small is too small a list?
It depends on the effect size you care about, which is why the skill computes it per test rather than giving a rule. A list that cannot detect a 10% revenue change on a segment is told so before the test runs rather than after.
What is included with the skill?
The skill itself, the element effects reference, the test design format, and the volume requirements. It is free and security scanned. We re-publish the ZIP when the model changes, so download it again if your recommendations start shifting.
Email revenue depends on which revenue you count.
Polar separates flow, campaign and blended revenue in one semantic layer, so the winning format is the one that moved the business, not the one the ESP claimed.
<script type="application/ld+json">{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How is this different from A/B testing subject lines?","acceptedAnswer":{"@type":"Answer","text":"Subject lines affect opens, and opens are both unreliable and not revenue. This scores the structural elements of your sends against revenue per recipient, which usually shows that the element you have never tested matters more than the one you have tested thirty-four times."}},{"@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":"Send history with the creative variants used, clicks, and orders attributed to sends. Without order attribution it falls back to clicks and labels the whole analysis as a click optimization, which is a different and weaker thing."}},{"@type":"Question","name":"Why stop testing subject lines entirely?","acceptedAnswer":{"@type":"Answer","text":"Because thirty-four tests with no detectable revenue effect is itself a conclusion. If the subject line mattered on this list it would have shown by now. Continuing to test it is spending your testing capacity on the element that has already answered."}},{"@type":"Question","name":"How small is too small a list?","acceptedAnswer":{"@type":"Answer","text":"It depends on the effect size you care about, which is why the skill computes it per test rather than giving a rule. A list that cannot detect a 10% revenue change on a segment is told so before the test runs rather than after."}},{"@type":"Question","name":"What is included with the skill?","acceptedAnswer":{"@type":"Answer","text":"The skill itself, the element effects reference, the test design format, and the volume requirements. It is free and security scanned. We re-publish the ZIP when the model changes, so download it again if your recommendations start shifting."}}]}</script>