Open tracking is broken, and what to measure instead

For years, open rate was the first number every outbound team watched. Privacy changes have quietly turned it into noise, and chasing it now optimises for a number that no longer means anything.

How open tracking works, and why it fails

An open is measured by a tiny invisible image in the email. When the recipient loads it, the sender records an open. The whole method depends on that image loading exactly when a human reads the message.

Privacy protections broke that assumption. Some pre-load the pixel for every email whether or not it is read, inflating opens. Others block it entirely, so genuine opens go uncounted. Either way the number stops meaning what it claims.

Inflated, deflated, or just wrong

The result is open data that can be wrong in both directions at once across a single campaign: some contacts counted as opening without reading, others reading without being counted. A metric that is unreliable in both directions is not a metric you can steer by.

The pixel can hurt deliverability

Beyond being inaccurate, the tracking pixel is an image and an external request, which is one of the signals spam filters weigh and which privacy tools increasingly flag. For cold outreach, where reputation is everything, carrying a pixel to measure a broken metric is a bad trade.

Measure replies instead

The reply is the metric that survives all of this. It cannot be faked by a pre-loaded image, it is not blocked by privacy tools, and it maps directly to what you actually want: a conversation.

Optimising for replies also improves the right things. It pushes you toward relevance and a clear ask, rather than toward subject-line tricks that lift a meaningless open number.

Where the distortion comes from

Three mechanisms do most of the damage. Consumer privacy features pre-fetch images through proxy servers, so every email to those inboxes registers as opened within minutes of delivery, read or not. Corporate security gateways open and scan messages before a human sees them, which both fires open pixels and clicks links, so click tracking inflates too. And image blocking, on by default in some business environments, suppresses the pixel entirely for readers who genuinely opened the message.

The mix varies by audience. A list of startup founders on consumer-grade mail behaves differently from a list of factory directors behind a corporate gateway. That is the deeper problem: the error is not a constant you can adjust for, it changes with the very targeting decisions you are trying to evaluate.

What open rate is still good for

One honest use survives: a crude deliverability alarm. If a campaign that tracked around 50% opens drops to 15% overnight with the same list quality, something changed, usually inbox placement. The absolute number means little, but a sharp move in a stable setup is worth investigating alongside bounce rates and spam-folder tests, as part of routine deliverability monitoring.

Even for that job, dedicated placement tests do it better without carrying a pixel in real sends. Treat open rate as a smoke detector at most, never as a performance metric. The numbers that belong on a weekly reporting dashboard all sit further down the funnel.

The metrics stack that replaces it

• Bounce rate: below 2-3% on a verified list; higher means the data, not the copy, is the problem.
• Reply rate: the first real performance signal. In the campaigns we run it sits in the low single digits, with tight niches doing better.
• Positive reply rate: the share of replies showing interest, which separates a provocative subject line from a relevant offer.
• Meetings booked: the number the campaign exists for.
• Unsubscribes and complaints: the cost side, worth watching as closely as the wins.

Every number in that stack is observable without tracking pixels. For what counts as good at each stage, see what is a good cold email response rate and the wider outbound benchmarks.

Common tracking mistakes that still circulate

• "He opened it 14 times, call him now." That is usually one security scanner and a mobile preview pane, not a hot prospect. Fix: act on replies and stated interest, not pixel folklore.
• Click-tracked links in cold email. Link rewriting through a tracking domain is a spam signal and a common reason clean copy lands in junk. Fix: send links bare or drop them from the first touch entirely.
• A/B tests judged on opens. A subject line that wins on inflated opens can lose on replies. Fix: test on replies, accept that it needs more volume.
• Comparing open rates across audiences. Different inbox mixes make the numbers incomparable. Fix: compare replies, which mean the same thing everywhere.

The European angle: the pixel is also a compliance question

In Europe there is a second reason to drop the pixel. Open tracking processes behavioural data about an identifiable person, which puts it inside GDPR territory, and several European regulators have taken a dim view of invisible tracking without disclosure. Cold B2B email itself can be run compliantly under legitimate interest, as covered in GDPR-compliant cold email, but a hidden pixel weakens exactly the argument that your processing is minimal and transparent. Removing it makes the campaign both cleaner and easier to defend. We run campaigns without open pixels as standard, and nothing about steering the work has been lost.

What this changes in practice

Drop the tracking pixel, stop reporting open rate as if it were real, and judge campaigns and A/B tests on replies and positive replies. If you A/B test, test against reply rate, not opens, or you will optimise confidently in the wrong direction. See A/B testing properly.