Email Analytics: The Complete Guide

Open rate stopped being a reliable signal: mail privacy protection loads email content automatically, and an open registers even when nobody saw the message. Measure clicks, conversion and revenue per email instead. Below are eight metrics with thresholds and links to the detail on each topic.

What broke in open rate

The mechanism is simple. An open is recorded by loading an invisible image inside the email. Privacy protection pre-fetches that image on the provider side, whether or not the person ever opens the message.

Two consequences follow. The absolute open rate is meaningless and cannot be compared with industry figures. But the trend is informative: a halving signals a delivery problem rather than a weak subject line.

Eight metrics and their thresholds

Here is every metric ranked by trust. The thresholds in the third column are not recommendations but boundaries where provider behaviour changes.

The metric that matters most

If you keep only one metric, make it revenue per email sent. It links the whole chain and converts the result into money.

Its main virtue is letting you compare the incomparable. A welcome email and a promotional campaign have different open rates and different audiences, but revenue per email brings them onto a common scale.

This is the metric that shows why automated email beats regular campaigns: the gap in revenue per email between them runs into tens of times.

Spam complaints

Of all the metrics, only one has a hard external threshold. A complaint rate above 0.1% — one per thousand emails — starts affecting the deliverability of everything you send afterwards, including transactional mail.

A paradox worth accepting: a rise in unsubscribes is sometimes good. Someone who unsubscribes instead of complaining preserves your sender reputation.

When A/B tests are possible

A valid A/B test needs roughly a thousand recipients per variant and a 95% significance threshold. With a list under two thousand addresses there is nothing to test: any difference will be chance.

The trap almost everyone falls into: tools display a confidence level, and a line such as "winner at 87% confidence" sounds convincing. But it means that roughly one time in eight you will crown a variant that is not better.

List segmentation

Segmentation is built on behaviour rather than demographics. Gender, age and city correlate weakly with how people respond to email, and the data has to come from somewhere — meaning a longer signup form and fewer signups.

Behavioural attributes accumulate on their own: opened, clicked, bought, absent for three months. If you implement only one, make it activity — it affects both campaign results and the deliverability of the entire list.

What to compare your numbers against

The most common evaluation mistake is comparing yourself against someone else numbers. Industry figures are computed with different methods and vary several-fold between sources.

The right reference point is you, a month ago. A second reference is the difference between email types inside your own programme: if automated flows do not beat regular campaigns on revenue per email, something is wrong with them.

How to set up measurement

Technically, measurement needs two things: identically structured tags across every link, and an analytics goal that records the final action rather than the visit to the site.

In Mailvex the tags are configured once for the whole email and applied to every link automatically, so the structure stays consistent without manual work.