Email Marketing Metrics: What to Track and Why
Fourteen metrics, what each one is actually measuring, and the four that are now proxies for something else entirely.
On this page
The short version
- Four of the fourteen metrics most dashboards show now measure something other than what their label says, and three of those four have opens somewhere in the calculation.
- Report three numbers. Everything else belongs in a diagnostic view you open when one of the three moves.
- A metric with no decision attached to it is not a metric, it is decoration. If nothing would change at any value, stop reporting it.
Email dashboards are unusually generous with numbers, which is not the same as being informative. A default report will show you fourteen figures, several of which move together, several of which no longer mean what they are called, and most of which nobody has ever acted on.
The useful exercise is not adding metrics but attaching a decision to each one. A number that would not change anything at any value is decoration, and removing it makes the rest legible.
What each number is really measuring
| Metric | What it sounds like | What it is now | Decision it supports |
|---|---|---|---|
| Delivered | It arrived in the inbox | The server accepted it — spam counts as delivered | Bounce diagnosis only |
| Open rate | People read it | Human opens plus automated image fetches | Detecting a sudden drop |
| Click rate | People clicked | Clicks over delivered, minus scanner noise | Whether the content worked |
| Click-to-open | Interest among readers | Clicks over a contaminated denominator | Directional at best |
| Unsubscribe rate | People left | Exactly that — undamaged | Cadence and relevance |
| Complaint rate | People reported spam | Complaints from providers who report them | Stop, this is the ceiling |
| Bounce rate | Bad addresses | Exactly that, split hard and soft | List source quality |
| Conversion rate | It worked | Whatever your attribution credits | Only against itself |
| Revenue per email | What it earned | Attributed revenue over delivered | Comparing sends |
| List growth rate | The list is growing | Adds minus removals, which hides churn | Split it, then use it |
Complaint rate is the only row where the decision is binary rather than directional. Above the threshold, nothing else on this table matters until it is back down.
The contamination, briefly
Apple's Mail Privacy Protection pre-fetches images for its users whether or not the message is read, and other providers apply their own proxying and scanning. That puts a variable, invisible number of machine opens into your open count.
The size of the effect depends on what share of your list uses Apple Mail, which you cannot see and which changes over time. So the number is not merely inflated by a constant you could subtract — it drifts for reasons that have nothing to do with your email.
Anything with opens in the denominator inherits this. Click-to-open is the common casualty, because it is often the metric people trust most.
Three numbers, and why these three
Click rate against delivered tells you whether the content earned action. It has a clean denominator and a small, identifiable contamination from security scanners that you can filter.
Complaint rate is the constraint. It is not something to optimise upward or downward within a range — it is a line, and crossing it makes every other number meaningless because delivery degrades.
The third is whatever the programme exists to produce: revenue, qualified enquiries, activations, renewals. If nobody can name it, that is the finding.
The diagnostic set
Everything else belongs behind the report rather than in it: bounce breakdown by reason, placement by provider, unsubscribe by segment, click distribution within the message, device split, time between delivery and open.
Keeping them separate is not tidiness. A monthly report containing a soft bounce distribution invites someone with no basis for interpreting it to interpret it, and the resulting decisions are worse than no decision.
Auditing your own report
- Every metric on it has a decision attached to it
- No metric on it has opens in the denominator
- Complaint rate is present and shown per segment
- The commercial outcome is on it, however roughly measured
- Diagnostics live somewhere else
- The series is marked where privacy protection broke the history
Frequently asked questions
Should we stop tracking opens entirely?
No — keep them as a drop detector, because a sharp fall still means something went wrong with delivery. Take them out of the headline report so nobody makes a content decision on them.
What about list growth rate?
Split it. Gross adds and removals tell you two different things; the net figure hides a list that is churning heavily while appearing stable.
How often should the report be produced?
Monthly for the three headline numbers, per send for click rate and complaint rate. Weekly reporting on a fortnightly programme produces mostly noise.