What is email open rate? Platform notes and checklist

What is email open rate? A tracked image-load signal, not proof of reading

That means an open is fundamentally an image retrieval event. It is useful, but it is not direct proof that a human read, understood, or even intentionally viewed the email.

The common formula

A typical open-rate calculation is. For example:

8,400 unique tracked opens / 20,000 successful deliveries × 100% = 42%

A reporting system should state whether its open rate uses delivered messages or another eligible population as the denominator, and whether the numerator counts unique recipients or total events. The formula is only comparable when those boundaries match. Always check your platform's exact denominator before comparing numbers across systems.

How open tracking works

HTML email can include a tiny, unique remote image. When the email client requests that image, the marketing platform records an event associated with the recipient or message. That design creates 2 categories of measurement error.

False negatives

A human can read the message without generating the expected image request. Examples include:

• remote images are blocked
• the user reads a plain-text part with no tracking image
• a client or network prevents the tracker from loading
• tracking is disabled

False positives

A tracking image can load without meaningful human reading. Examples include:

• Apple Mail Privacy Protection preloading remote content
• security software fetching resources
• other automated or privacy-preserving clients

So the metric answers "did the tracking mechanism register an open-like event?" rather than "did the person read this email?"

Apple Mail Privacy Protection changed the meaning of opens

Apple Mail Privacy Protection, or MPP, can privately download remote content in the background instead of waiting until the user views the message. Apple says this hides the recipient's IP address and prevents senders from learning the same viewing information from remote content.

Apple's documented privacy behavior means Mail can preload tracking pixels even if the contact has not intentionally opened the email, inflating and distorting tracked-open metrics for affected recipients.

A campaign can therefore show an open event because Apple fetched the pixel, not because the recipient decided to read the email at that moment.

Open rate still has uses

The answer is not to delete open rate from every dashboard. It is to use it for tasks it can still support. Open rate can be directionally useful when:

• comparing similar sends on the same platform with stable tracking rules
• watching a long-term trend while remembering privacy changes
• estimating reach alongside stronger engagement metrics
• using provider/platform filters that attempt to identify automated opens
• diagnosing an extreme change that coincides with a subject-line, sender-name, or audience change

It is weak when treated as a precise measure of human attention.

Stronger signals for stronger decisions

The more consequential the decision, the less you should rely on opens alone. Clicks are not perfect either because security systems can generate automated clicks, but they generally require more interaction with message content than a tracking-pixel fetch. Clicks, purchases, replies, and site activity are usually stronger evidence for a consequential decision than an open alone.

Open rate and subject-line testing

Subject lines influence whether a person chooses to open, so open rate historically became the obvious test metric. Privacy prefetching weakens that relationship. For a subject-line experiment, prefer a hierarchy such as:

1. primary business outcome if the test has enough volume, such as conversion
2. click or downstream engagement when conversion is too sparse
3. open rate as a supporting directional measure, preferably with automated opens filtered where the platform can do so

Be cautious about tiny differences. A 0.4 percentage-point lift in opens can be smaller than the measurement noise created by recipient mix and automated fetching.

Open-triggered automations need guardrails

A flow such as "if opened but did not click, send reminder" can now enroll recipients based on machine activity rather than demonstrated human attention.

MPP can cause open-triggered automations to reach a larger audience and non-open triggers to reach a smaller audience because the event no longer maps cleanly to deliberate human viewing. If the action matters, combine opens with stronger evidence:

• opened and clicked
• opened and visited site
• not clicked and no purchase
• explicit preference or inactivity windows

Comparing open rates across time

Historical comparisons can break when tracking technology or recipient client mix changes. If an old campaign from 2020 shows 24% and a 2026 campaign shows 46%, the difference does not prove your subject lines doubled in effectiveness. Privacy preloading, bot filtering, audience composition, client mix, and reporting definitions have changed. When building benchmarks, record:

• platform
• tracking setting
• MPP/bot filtering setting
• delivery denominator
• date range
• audience/client mix where available

Then compare within the most stable cohort possible.

The formula

open rate = unique recipients with a tracked open / successful deliveries × 100%

Worked example

Campaign A:

• 10,000 delivered
• 6,000 tracked opens
• 120 clicks
• 12 purchases

• Campaign B:

• 10,000 delivered
• 3,500 tracked opens
• 420 clicks
• 64 purchases

If Campaign A had a larger share of Apple MPP recipients or automated opens, its 60% open rate may overstate human attention. Campaign B's lower open rate paired with much stronger clicks and purchases can represent substantially better commercial engagement. The point is not that every open in A is fake. The point is that the open-rate ranking is not enough to rank business performance.

Common mistakes

1. Describing an open as proof a recipient read the email
2. Comparing open rate across providers without checking formulas
3. Triggering high-impact automation from opens alone
4. Ranking campaigns only by open rate after Apple MPP
5. Ignoring false negatives from image blocking or plain text
6. Ignoring bot and privacy prefetching
7. Treating a historical open-rate trend as stable when measurement rules changed

Questions we get asked