Email Benchmarks by Industry: How Do You Compare?

Every year several organisations publish tables of email benchmarks by industry, and every year they disagree substantially — sometimes by more than the difference between the best and worst industries in either table.

That is not sloppiness. They are sampling different populations, defining terms differently, and in the case of open rate, measuring across a period when the measurement itself changed.

The four reasons they disagree

First, sample. A benchmark drawn from one platform's customers describes that platform's customers — their size, their sophistication, their price point.

Second, definitions. Some report against delivered, some against sent. Some count unique opens, some total. A two-point difference can be entirely definitional.

Third, consent standards. A population with a high share of double opt-in lists will show higher engagement than one without, and nothing about industry explains it.

Fourth, and specific to opens, the measurement changed. Figures collected before and after privacy pre-fetching became widespread are not the same quantity, and most tables do not say which period their data covers.

What a published benchmark is good for

Order of magnitude, and only that. If your click rate is a tenth of every published figure, something is wrong; if it is within the same range, you have learned nothing about whether it should be higher.

They are also useful for a specific negotiation: showing someone outside the team that the numbers they are used to from another channel do not transfer. An open rate that would be alarming on a website is normal in an inbox.

Building the benchmark that is worth having

Take your own last twelve months, split by send type — flows, newsletters, promotions — and calculate the median and the range for each. That gives you a target that accounts for your list, your consent method, your cadence and your audience, because it is made of them.

Then set the alerting on the range rather than the median. A send below the tenth percentile of its own type is worth investigating; a send below the median is half of them by definition.

Rebuild it annually rather than continuously, so a slow decline does not quietly become the new normal by moving the benchmark down with it.

The one external comparison worth making

There is a case where outside data genuinely helps, and it is not the industry table. It is comparing your own segments against each other and against a similar programme you have direct knowledge of — a sister brand, a previous employer, a peer who will share real figures.

That comparison holds constant the things published tables vary, because you can ask how the list was built and how often they send. One honest conversation with someone running a comparable programme is worth more than any table, and it is available to almost everyone and used by almost nobody.