What Mailchimp’s deliverability numbers do and do not tell you

Mailchimp's campaign report is accurate about what it measures. The problem is that three of its headline numbers do not measure what people use them for. Sort out which is which and the report becomes a decent operational tool.

Metric by metric

Two of those deserve expanding, because they cause most of the wrong conclusions.

Delivery rate is acceptance, not inbox

When Gmail answers 250 2.0.0 OK, it has taken custody of the message. The folder decision happens afterwards, inside Gmail's filters, and no receiver reports it back to the sender. So a campaign showing 99.4% delivered can be entirely in Spam, and nothing in any ESP dashboard — Mailchimp's included — would show it.

The only way to see folders is to send to mailboxes you can read: a seed set across Gmail, Microsoft, Yahoo and independent hosts, checked per sending domain. Treat the result as directional (seed sets are small) but binary signals are reliable: "Microsoft put this in Junk" is a fact about that path and that message.

Opens are inflated, unevenly

Apple Mail Privacy Protection pre-fetches images through a proxy, so a message can register an open that no human saw. Corporate security gateways do the same. The inflation is not a constant — it depends on what clients your audience uses — so you cannot subtract a fixed percentage and move on.

Consequences for decisions you may be making:

• Do not define an "engaged" segment by opens alone. Use clicks, replies, purchases or site visits where you can. If you must use opens, keep the window short and expect the segment to be larger than reality.
• Do not A/B test subject lines on opens if the difference is small. The noise can exceed the effect.
• Do not report an open-rate improvement as a deliverability improvement. They are not the same measurement.

Complaints undercount, and still matter most

A complaint appears in Mailchimp only when the recipient's provider has a feedback loop and reports it. Gmail does not pass individual complaints to senders; it aggregates them into the spam rate you see in Google Postmaster Tools. So your Mailchimp abuse figure can be near zero while Gmail users are marking you as spam steadily.

This is why Postmaster Tools is not optional: enrol each sending domain, and watch the user-reported spam rate. Google's published bulk-sender threshold is under 0.3%; treat sustained readings above 0.1% as something to fix this week. Note that the domain reputation chart no longer exists in Postmaster Tools, so spam rate and authentication pass rates are the signals that remain.

The numbers Mailchimp quietly handles for you

• Cleaned contacts. Addresses that hard-bounced or complained are removed automatically and cannot be re-added by importing them again. The cleaned list is an accurate record of your worst data; export it and purge those addresses from your CRM too.
• Pre-send filtering. Platforms of this size run their own abuse checks on campaigns before release, which is why an account can be paused or a campaign reviewed. That is the shared pool being protected, not a bug.

What to measure outside the platform

1. Per-provider inbox placement, from seed testing, per sending domain and per stream.
2. Gmail user-reported spam rate and authentication pass rates, from Postmaster Tools.
3. Blocklist status for the sending domain, with a current list of lists — SORBS shut down in June 2024, and Validity's Heatwave blocklist launched on 3 September 2026.
4. Authentication from real headers: dkim=pass with your domain, dmarc=pass with header.from= on your domain. The platform's green tick tells you the records resolve, not that alignment is right.
5. Revenue or replies per send, which is the only metric that is not a proxy for something else.

What not to do

• Do not set targets on delivery rate. It will sit above 98% while placement collapses.
• Do not re-import cleaned contacts. Mailchimp will refuse, and on other platforms you are manufacturing complaints.
• Do not compare a delivery percentage between two platforms. They count different events.
• Do not treat an open-rate drop as a deliverability incident without checking placement and clicks — a mail-client change can cause it on its own.
• Do not wait for the complaint number to move. By the time it does, the spam rate has already moved.

MailPilot does not integrate with Mailchimp, and we will not imply otherwise. It measures the things a campaign report cannot: real per-provider inbox placement through a seed-inbox network, SPF, DKIM and DMARC verification per sending domain, warm-up over OAuth (Gmail, Outlook) or SMTP and IMAP (SendGrid, Amazon SES, Mailgun, Postmark, custom SMTP), and a reputation score per mailbox every day.