Every e-mail marketer worth her salt knows how to calculate an overall messaging open rate: messages opened ÷ messages delivered. It's probably the easiest calculation you can make that still produces valuable information for your organization.

That's why I was a little confounded when we first started using our new e-mail system, and it would spit out a report like this:

An overall open rate of almost 33%? Fantastic! That makes my day. I need to go tell it on the mount... wait. Almost 3000 messages delivered, but only 600 opens? That sounds suspiciously like 20% to me.

What our system reports – and I suspect others, as well – is an average of averages, or what I call an "Unweighted Open Rate" (as opposed to the traditional "True Open Rate"). It's generally much higher because it doesn't take into account total messages sent, just the open rate of each message: the 12.6% rate for a message sent to 2300 is brought way up by the 65% open rate for a message sent to 350 because it's just adding 12.6 to 65 and dividing by 2.

Here at NTEN, we find it valuable to keep track of both numbers.

True Open Rate

The traditional measure of open rates – total messages opened ÷ total messages delivered – should be used to gauge the overall health of your e-mail program. This is the number you can use to compare the relative robustness of your messaging program with the results reported by the "eNonprofit Benchmarks Study" (or Convio's "Online Nonprofit Benchmarks Study" or MailChimp's "Email Marketing Benchmarks by Industry", or...)

Just bear in mind that almost every industry survey will report slightly different numbers. Because the comparisons will never be exact, the most important thing to watch is your True Open Rate over time. If this number goes down drastically, or even just slowly and steadily, you should be a little worried. Probably. That's where the "Unweighted Open Rate" comes in.

Unweighted Open Rate

The average of all the open rates of every message you send is a measure of how well you segment your lists.

We very rarely send to our entire file; even our newsletter goes out in 6-8 different editions now, tied to job type. In fact, most of the messages we send out go to the folks we know to be already active in our community, because we know they're more likely to open... and take action. It's not cheating, it's segmentation!

The better you target your messages – and customize the content, the subject line, the tone, the images, the landing pages – the better the open rate will likely be for your individual messages – and the higher your Unweighted Open Rate. I've segmented down to as few as 18 people with a custom message – a bit extreme, but worth the 60% completion rate in the end.

We like to go beyond the average to look at other statistical measures, as well. Say you have a set of messages with these open rates:

[14,28,16,22,28,25,12,28,24,28,29,57,26,28,14]

The (unweighted) average/mean is about 25%, the median (the number right in the middle of the set) is 26%, and the mode (the number that comes up most often) is 28%. You can use this data to pretty safely say that when you send out a message, you can expect an open rate in the neighborhood of 26-28%. (I say this because if you change the high outlier (57) to, say, 15, the average sinks to 22%, but the median and mode stay about the same.)

As I said before, this number is likely higher than your True Open Rate. If you only think in terms of absolutes, you're selling your e-mail program short: comparing a True Open Rate of 16% to individual messages can be misleading.

The Unweighted Open Rate numbers help you focus on all of the poorly performing messages – the 14s, the 12s – as well as the good ones, like that 57, without having to worry so much about the 22s or 31s.

What was different about those messages? Figure out what's successful and what's not, and you'll go a long way to bringing up your overall average – without the need to examine every message you send out in detail.

A good program of segmentation also earns you the ability to hit those lists that are least likely to respond, the names that have sat in your CRM for 5 years and never opened a thing, without killing your True Open Rate. It's always worth a shot. Just don't try it too often.

What are your tracking around open rates? We're always looking for new metrics.