The skill
Install once in Claude. It computes inbox placement by provider, models engagement decay on clicks, builds a recency-based sunset policy, and parses DMARC aggregate XML.
What's inside
Three files, one job: find the leak before the inbox
Built for email and lifecycle marketers, not deliverability consultants.
Install guide
A short setup walkthrough, plus a no-install option if you cannot add skills.
Plan B prompt
A single prompt you paste into any Claude chat to run the same analysis without installing anything.
How it works
Three steps, from your exports to the verdict and the fix.
Step 1Bring your exports
Google Postmaster or Microsoft SNDS data and seed-test results for placement, subscriber event data with click and conversion timestamps for decay and sunset, or your DMARC aggregate XML.
Step 2Claude runs the analysis
It computes inbox placement by provider and separates it from delivery, charts engagement decay on clicks rather than opens, segments the list for sunset, or parses the DMARC report by alignment, in code.
Step 3Get the verdict and the fix
Placement and risk flags by provider, the cohort where engagement collapses, who to re-engage and suppress, and which sending sources to fix before you move DMARC to enforcement.
The skill works from the exports you bring, it does not connect to your ESP or mailbox providers. It also reads engagement on clicks and conversions, never opens, because Apple machine opens corrupt the open rate.
How to improve email deliverability
Improving deliverability starts with measuring the right number, because roughly one in six marketing emails never reaches the inbox and the metrics most teams watch cannot see it. Delivery rate only means a server accepted the message, and open rate is now mostly machine generated. Here is what actually drives inbox placement and how to fix it.
Why delivery rate and open rate hide the problem
Your email platform reports delivery as messages sent minus bounces, typically near 98.5 percent, which only means a receiving server accepted the mail, not that it reached the inbox rather than spam. Open rate is worse, because Apple Mail Privacy Protection fires the tracking pixel whether or not anyone reads the message, and as of March 2024 that was about 55 percent of all opens. The honest measures are inbox placement rate, click and conversion rate, complaint rate, and revenue per recipient, and the skill reports placement instead of the vanity numbers.
How to measure your true inbox placement rate
Inbox placement rate is the share of sent mail that actually lands in the inbox, and on Validity's 2025 network that was 83.5 percent globally, so about one in six legitimate emails misses it. The skill computes placement by provider, IP and campaign from your Postmaster and seed-test exports, separating delivery from placement so the gap is visible, and flags reputation risk as your spam complaint rate approaches 0.1 percent and again at 0.25 to 0.3 percent.
Why authentication is now a hard gate
Since 2024 the major mailbox providers have made authentication pass or fail. Bulk senders, defined by Google as 5,000 or more messages a day to personal Gmail, must publish SPF, DKIM and aligned DMARC, offer one-click unsubscribe, and keep complaints below 0.3 percent with a 0.1 percent target, and Microsoft applied the same in May 2025. The skill parses your DMARC aggregate XML, reads each source by alignment since a message can pass raw SPF but fail aligned, and tells you whether every legitimate sender passes before you move from monitor-only to quarantine or reject.
Why sending to the unengaged hurts your whole list
Mailbox providers compute reputation per sending domain and IP, and sending to unengaged users trains the classifier that your mail is unwanted, which drags placement down for the whole list, including the subscribers who do engage. The fix is a recency-based sunset, suppress no-click, no-conversion subscribers at about 90 days into a re-engagement series and remove them at about 180, since at least 23 percent of a list degrades every year. The skill models engagement decay on clicks and builds that sunset for you.
What the skill cannot do, and how to stay honest
It is an analysis tool, not a magic reputation fix. AI cannot quickly repair a damaged reputation, recovery takes four to eight weeks of throttled sending, and AI-written bulk copy homogenizes content and trips filters, so this is a tool for analysis, not generation. List-verification accuracy runs nearer 93 to 97 percent than the advertised 99 and cannot resolve catch-all domains, and correlation is not causation in deliverability, so validate any confident read against trended real-recipient data. For the full argument and the evidence behind measuring placement over opens, read Fix Inbox Placement, Not Subject Lines.