Why your cold email open rates are wrong

An open rate looks like a simple number: how many people read your email. It isn't. It counts how many times a tiny tracking image was downloaded, and a growing share of those downloads are done by software, not people.

How open tracking works

When open tracking is on, each email carries a small, invisible image with a unique address. When the image loads, the sending tool records an "open". Anything that loads the image counts, whether or not a person ever looked at the email.

Where fake opens come from

Apple Mail Privacy Protection. Since 2021, Apple Mail can download remote content in the background through Apple's proxy servers for people who switch the feature on. Every email those people receive looks opened, often within moments of arriving, even if it's never read.

Security gateways and link scanners. Corporate filters such as Microsoft Defender for Office 365, Mimecast, Proofpoint and Barracuda inspect incoming mail before it reaches the inbox. Some of them load images or follow links while they check. Business email, which is most cold email, goes through these filters far more often than personal email.

Instant fetches. An "open" one or two seconds after sending is almost never a person. Real people don't read that fast.

And opens you never see

Open rates are wrong in the other direction too:

• Many desktop email apps, including Outlook in many setups, block external images until the reader allows them
• People who read in plain-text mode never load the image
• Gmail caches images, so a second or third open may not reach the tracker

So the same number is inflated by machines and deflated by image blocking. It can still be useful, as long as you read it the right way.

Gmail is the exception

Gmail loads images through its own image proxy when someone opens the email. Those requests come from Google rather than the reader's device, but they happen because a person opened the message, so they're a reasonable signal of a real open.

How to use open rates

• Compare, don't celebrate. Opens are most useful as a relative signal: template A against template B, or this week against last week.
• Watch for sudden drops. A sharp fall in opens across a mailbox often means it has started landing in spam.
• Measure what matters. Replies, positive replies and meetings booked are the numbers that tell you whether a campaign works.

Tracking has a cost

The tracking image is one more remote element and one more link to a tracking domain in every email, and spam filters notice both. For brand-new mailboxes, or campaigns where you only care about replies, consider switching open tracking off.

How Mailnitrix counts opens

• Any request within a few seconds of sending is counted as a machine open
• Requests from Apple's privacy proxy and known scanners are counted as machine opens
• Gmail image-proxy opens are kept as real opens
• Machine opens are stored separately, so you can still see them
• The first real open is recorded exactly once, even when several requests arrive together
• Tracking can be switched off per campaign, and warmup emails are never tracked

See how this looks in practice on the analytics page.