Email Marketing Data | Why Open Rate Stopped Being a Metric

Email should be the easiest channel to measure. You know exactly who you sent to, you control the links, and the recipient is already identified. Yet email reporting is dominated by a number that is now roughly half fiction.

Open rate is broken, and it is not coming back #

Email opens have never been measured directly. They are inferred from a tiny invisible image — a tracking pixel — loading when the message is displayed. If the image loads, the platform assumes a human opened the email.

In September 2021 Apple introduced Mail Privacy Protection. When enabled, Apple Mail pre-fetches every image in every message through a proxy, whether or not the recipient ever looks at it. The pixel fires regardless. The open is recorded regardless.

The scale of it

Roughly half of all tracked email opens now come from Apple Mail Privacy Protection. Independent measurements in 2026 put it near 49–50% of opens across large samples. That is not a margin of error — it is most of the metric.

Three consequences follow, and each one breaks something people still do:

Your open rate went up in late 2021 and it was not a win. Any year-on-year comparison spanning that period is meaningless.
Open-based segmentation is now actively harmful. “Send to everyone who opened but did not click” selects a group that is roughly half Apple proxies who never saw the message. Suppression based on non-opens is worse — it removes genuinely engaged non-Apple readers while keeping inactive Apple ones.
Open-triggered automations fire on nothing. Any workflow whose entry condition is an open is triggering on Apple's servers, not on a person.

One thing MPP does not affect

Clicks are unaffected. Apple pre-fetches images; it does not click links. Click data remains genuine, and it is the foundation of everything below.

What to measure instead #

Metric	Why it works	What to watch for
Click rate clicks ÷ delivered	Requires a deliberate human action. Unaffected by MPP.	Some security scanners click every link in a message. A subscriber who clicked every link within one second is a scanner, not a lead.
Click-to-delivered, not click-to-open	CTOR uses opens as its denominator, so it inherits the whole problem.	If your platform reports CTOR by default, change the report.
Conversion rate from email to key event	The only measure that connects email to the business.	Requires consistent UTM tagging on every link in every send.
Revenue per recipient	Normalises for list size and makes sends comparable. The best single email metric for ecommerce.	Needs order data joined back to the send, which most platforms do natively for Shopify.
List health unsubscribes, spam complaints, bounces	Leading indicators of deliverability collapse, which is the real existential risk in email.	Spam complaint rate above 0.1% is a warning; above 0.3% is an emergency.
Deliverability inbox placement, not delivery rate	“Delivered” only means the server accepted it. It says nothing about the spam folder.	Needs seed testing or a deliverability tool. Most teams have never measured this at all.

Why your email platform's revenue differs from your CRM #

Email platforms apply their own attribution, and it is usually generous.

Attribution windows. Klaviyo defaults to a 5-day click and 5-day open window for revenue attribution. That open-based half is now unreliable for the same reason as above, and it will credit email for purchases it merely coincided with.
Last-touch bias. Email frequently arrives at the end of a journey created elsewhere. It is very good at harvesting existing intent, which makes it look extremely efficient under last-touch models.
Overlap with everything. A recipient who saw a Meta ad, searched your brand, and then clicked a promotional email will be claimed by all three systems.

The honest framing

Email is nearly always the highest-ROI channel by last-touch attribution, and this is partly real and partly an artefact of where it sits in the journey. The list was built by other channels. Email converts it. Both facts matter, and a report that credits email with the whole value is quietly writing off the acquisition that made the list possible.

Marketing automation adds its own complications #

Once you move from broadcast sends to automated journeys — HubSpot workflows, Klaviyo flows, Salesforce journeys — a few new measurement problems appear:

Flows and campaigns are not comparable. An abandoned-cart flow converts at a spectacular rate because it targets people who already chose a product. Comparing it to a newsletter is comparing harvesting to sowing.
Attribution can double inside one platform. If someone receives a campaign and a flow before purchasing, some platforms credit both.
Lifecycle stage changes are timestamped, not versioned. Most CRMs record when a contact became an MQL but not the full history, which makes cohort analysis harder than it looks.

The compliance layer #

Email is the most heavily regulated marketing channel in South Africa. Section 69 of POPIA prohibits unsolicited electronic direct marketing unless the recipient has given prior consent or is an existing customer under specific conditions, and every message must identify the sender and offer a working opt-out.

This is not only a legal matter — it is a data quality matter. A list built without consent produces spam complaints, which destroys deliverability, which means your legitimate email stops reaching the inbox and your reporting quietly becomes worthless. See Consent and privacy.

Common questions #

Should I stop reporting open rate entirely?

Stop reporting it as a performance metric, and stop using it for segmentation, suppression or automation triggers. It retains one narrow use: a sudden collapse in opens across all clients can indicate a deliverability problem worth investigating. Keep it as a diagnostic, remove it from the results slide.

Can I tell which subscribers are Apple Mail users?

Some platforms flag MPP opens or let you segment by client. Where available this lets you calculate a cleaner open rate for the non-Apple portion. It is useful for trend continuity, but it does not restore the metric — you are now measuring a shrinking, self-selecting subset of your list.

How should I report email to a board?

Revenue per recipient, or leads generated per send, against list size and send frequency. Add unsubscribe and spam complaint rate as guardrails, so nobody is tempted to buy short-term revenue by sending more often. Do not put open rate on the slide.