Open rates are broken. Here’s what email teams should measure instead

Leadership still asks: “What’s the open rate?”

The campaign goes out. The dashboard updates. Someone asks whether opens improved.

For marketers and CRM teams, it’s a familiar reporting habit. Open rate is easy to present, easy to compare, and increasingly difficult to interpret.

Did more people pay attention? Did software load the email’s images? Did your tracking coverage change?

One percentage cannot answer all three.

Apple Mail Privacy Protection and similar auto-opens make this a measurement problem first. Consent and tracking rules are another reason to rethink the metric.

Your next campaign report should make it easier to answer a more useful question: what did recipients actually do?

An email can register an open without being read

Open measurement usually relies on a remote image, often a tiny tracking pixel. When that image is requested, the email platform records an event and labels it an “open.”

The event records an image request. It cannot establish that someone read the message.

With Apple Mail Privacy Protection enabled, remote content downloads in the background regardless of whether the recipient engages with the email. That download can register as an open without a person reading anything. Apple explains how Mail Privacy Protection works.

Some other email clients also use image preloading and proxies. Their behavior varies, so avoid treating every proxy request as evidence of the same thing.

The practical consequence: a higher open rate can reflect changes in email apps or privacy settings alongside changes in audience interest.

That weakens decisions based on opens alone:

• Which subject line won?
• What time should the next campaign go out?
• Which profiles count as engaged?

Filtering suspected auto-opens may help. The remaining events still cannot prove that someone read your email.

Consent and tracking rules are another reason to rethink opens

Person-level open measurement used to optimize campaigns, personalize messages, or build profiles can sit under a different consent standard than an ordinary email marketing opt-in.

If your program reaches audiences where those rules matter, treat that as another reason to stop optimizing on opens — not as a reason to rewrite your whole consent stack in this article. For the France and Italy detail, see our guide to email open tracking in France and Italy.

“No open” can send an automation down the wrong path

Search your automations for “no open → resend,” “no open → win-back,” and “no open → mark inactive.”

Each rule assumes the absence of an open tells you something dependable about the person.

Consider what can happen:

• An unnecessary resend: someone reads with images blocked, or receives an email without open tracking. Your automation sends the same message again.
• A missed win-back: automatic image downloads keep a dormant profile looking active.
• An incorrect inactive label: a profile without measured opens gets suppressed despite recent purchases or genuine replies.

Our recommendation: give automations an explicit way to handle unavailable measurement.

“Tracking disabled,” “suspected automatic open,” and “no recorded event” should not collapse into a single engagement status.

Use relevant purchases, replies, and filtered clicks to inform the next step. Add frequency limits and stop follow-ups when the intended action is complete. Review inactivity over a period that fits your buying cycle.

Missing measurement should never be the sole reason you declare a customer inactive.

What to measure instead: a practical checklist

Choose metrics around the job each campaign is meant to do.

• Delivery and list health. Report sent and delivered counts, delivery rate, hard and soft bounces, complaints, and unsubscribes. Remember that “delivered” generally means accepted by the receiving server; it does not guarantee inbox placement.
• Clicks, with care. Count unique clickers after filtering suspected automated activity. Review unusual bursts across multiple links, and exclude unsubscribe links from positive engagement. Define the denominator and show tracking coverage. Avoid making click-to-open rate your replacement headline: it still depends on opens.
• Completed actions. Pick the outcome before sending: a purchase, registration, completed signup, or qualified enquiry. Use a consistent conversion window. Label revenue credited to email according to your attribution method; a purchase following an email does not by itself prove the email caused it.
• Engaged profiles. Count unique profiles with meaningful activity over a defined period, such as 30, 60, or 90 days. Specify which events qualify: filtered clicks, genuine replies, purchases, or relevant product activity. Exclude automatic replies and count each profile once.
• Coverage of tracked sends. For each tracking metric, show the share of delivered emails where that tracking was enabled and permitted. Report exclusions for suspected automation separately. Coverage tells you how much of the campaign was instrumented; it does not guarantee every human interaction was observable.

For example, if open tracking was enabled on 6,000 of 10,000 delivered emails, coverage was 60%. Any open rate needs that context. The measured group may also differ from the rest of the audience.

Apply the same discipline to click measurement. Changing the event you track does not remove the need to review how you collect and use it.

Give leadership a scoreboard they can act on

Lead with the campaign’s intended result: purchases, qualified replies, registrations, or another completed action.

Then show delivery, filtered clicks, engaged profiles, and measurement coverage. Keep open rate as supporting context if leadership still finds it useful.

Here’s a simple way to brief the change:

“We’re updating email reporting because opens include automatic activity and measurement covers only part of the audience. We’ll lead with completed actions, delivery, and meaningful engagement. Every tracked rate will show its coverage.”

Before the next reporting cycle:

• Agree on one primary outcome for each campaign type.
• Document denominators, attribution windows, and filtering settings.
• Mark changes to tracking coverage or filters on the report.
• Compare periods with consistent definitions.
• For important marketing campaigns, consider a randomly selected group that receives no campaign email to help estimate the additional outcomes the send produces.

A drop in reported opens after a tracking or filtering change may reflect reduced visibility. Investigate that before treating it as weaker demand.

How Joryio handles opens by default

Joryio filters known bot and automatic opens out of the headline open number by default — including patterns that look like Apple Mail Privacy Protection prefetch. Reported opens are tighter as a result. That is intentional: less noise in the open column, not a claim that opens are suddenly trustworthy.

Even then: don’t optimize on opens. Prefer conversion events and clicks (with care). Use filtered opens only as supporting context when someone asks “did people open?” — never as the campaign scoreboard.

If your reports still lead with opens, move completed actions and filtered clicks to the top of one campaign report. That operator posture is what Joryio is built around.

Should I turn off open tracking?

Not necessarily. If you keep it on, treat opens as noisy supporting context, show coverage, and stop using “no open” as the sole automation trigger. Where person-level open measurement needs its own consent review, handle that separately — including in our guide to email open tracking in France and Italy.

Does switching to clicks fix it?

Clicks can provide stronger evidence of intent, but automated link loading can generate activity too. Person-level click tracking may also need its own consent assessment, depending on how you use the data.

Treat clicks as a better signal than opens — still not a complete substitute for completed actions.

Start with one report and one automation

You can begin this week.

Choose one campaign report and move its business outcome to the top. Then review one automation that treats “no open” as a reason to resend, reactivate, or suppress someone.

This article provides general information, not legal advice.