Email Open Rate Calculator

How it is calculated

Open rate is unique opens divided by emails delivered: opens ÷ delivered × 100. Delivered rather than sent, because bounced messages were never eligible to be opened.

An open is registered when a tracking pixel loads, which means the metric measures image loading rather than reading.

That distinction has become the central problem with the metric, because a significant share of opens are now recorded by software rather than people.

The same thing with real figures

12,000 subscribers with 180 bounces leaves 11,820 delivered. 2,600 unique opens is a 22% open rate.

Of those 2,600, an unknown share are machine opens, mail privacy features that pre-fetch images on the recipient's behalf, registering an open whether or not the message was read.

In lists with a heavy Apple Mail share, that inflation can be 30% or more of reported opens, so a genuine 22% might report as 30%.

The rate can therefore rise without a single additional person reading anything, which is why year-on-year comparisons across the point that privacy protection was introduced are meaningless.

The catch

Machine opens broke open rate as a measure of interest and, worse, broke it unevenly, lists skew differently by client mix, so two businesses with identical engagement can report very different rates.

Automated opens also corrupt anything built on top of the metric: open-based segmentation, re-engagement triggers and send-time optimisation all now act partly on machine behaviour.

Applying it

Move engagement segmentation from opens to clicks and to purchases. Both require deliberate human action and neither is inflated by image pre-fetching.

Keep open rate as a deliverability signal rather than an interest signal. A sudden collapse still indicates something is wrong with inbox placement, which is a use the inflation does not destroy.

What open rate is still good for

Relative comparison within a single list over a short period remains useful, because the machine-open share is roughly constant across sends to the same audience.

A subject line test run on the same list in the same week is therefore still readable, both variants carry the same inflation, so the difference between them is real even if the absolute numbers are not.

What has gone is the ability to compare across lists, across time periods spanning a client change, or against any published benchmark. Businesses still reporting open rate as a headline metric are mostly reporting their subscribers' choice of email client.

Benchmarks published before 2021 are describing a metric that no longer exists in the same form, so comparing a current rate against them measures the change in tracking rather than in performance.

Where a platform offers machine-open filtering, turning it on makes the series internally consistent going forward and does nothing for the historical comparison.

Segmenting the report by mail client, where the platform exposes it, at least shows how much of the list is subject to pre-fetching and therefore how much the headline figure is inflated.

Sender name consistency matters more than most subject line work, since recipients scan for who sent it before reading what it says.