Identifying engaged subscribers: repeat opens
At one end of the spectrum you have those who signed up a while back, lost interest and now give any message as much attention as it takes to delete it.
At the other end are those who really value your emails. They look forward to, read, respond to and tell others about them.
Identifying the addresses at each end of this spectrum is an important exercise.
You can target the first group with a reactivation campaign. And the second group are nominally your "best" subscribers: you can thank them, perhaps target them with more email, and/or give them tools and incentives to spread the word even more than they already do.
Most people identifying their best subscribers filter their list according to the end objective. Who spends the most money? Who buys most often? Who attends the most events? Or they may use more complex models of behavior.
Which is great, but there are several problems:
- a lot of us don't have the ability to match an email address with subsequent behavior at a website
- a lot of us send informational emails where there is no such "result" to measure
- not all our best subscribers respond online in a measurable way
- the definition of best isn't just about traditional responses like sales or registrations. What about those who don't buy anything but do a great job of passing our message around?
If you just have basic open and click data for your emails, you can still pull out "engaged" subscribers from your list by playing a little with the numbers. Here some suggestions using a real B2B newsletter as an example.
Repeat opens are the forgotten statistic in email marketing.
We know the open rate has its limitations as it simply tells us whether a tiny tracking image was activated or not.
But most email marketing software and services also record how many times that tracking image was activated by any one recipient. And that's an interesting number.
Because if a tracking image is activated once, it doesn't tell us much. We don't know if they opened and read the email, or if they just triggered an open through a preview they never actually looked at.
We can never be sure if the recipient truly gave the email any attention.
But if an open is triggered many times (as measured by the repeat or total opens figure), it's highly unlikely that this is by accident. One of two things is probably happening:
1. The recipient is returning to the email again and again.
2. The recipient forwarded the email intact to others, who are triggering opens as they get the email.
Both scenarios suggest an "engaged" reader. The former is very interested in your message, the latter thinks your message is valuable enough to share.
Here's how the pattern of total opens might look for an email (based on real data from that B2B newsletter). The graph takes those who opened the email and details what percentage opened that message once, twice, three times, four times etc.:
Over 60% just "opened" the message once. But see how 2.2% opened the email 11 times or more. In fact, one subscriber actually opened the email 175 times. These are the folk we might zoom in on and tag as "engaged" subscribers.
Now maybe that one recipient found a dreadful error in the email and sent it to all her friends so they could share a laugh. Rather than pick out engaged subscribers from individual emails, we should maybe look at averages.
Here's what I did...
First I aggregated total opens across four months' worth of emails (nine issues) to get an average total open pattern (which is actually the graph above).
This is an interesting exercise in itself, because you can compare the average with the total open pattern for individual emails. Here's an example:
In this case, we can see that the holiday campaign was more engaging than the average and we can then explore why.
But I digress.
Having worked out average total open rate patterns across many emails, I calculated an average total opens per opener: 2.3.
In other words...a recipient will trigger an open an average of 2.3 times for each email that they open.
Defining what makes an engaged reader is a subjective decision, but I did it like this:
1. First I pulled out any address who opened an individual email more than five times the average. So anyone who opened any one email more than 11 times made the cut.
2. Of those names, I then recorded those addresses who opened another email more than twice the average (5 times or more).
The result was a list of "most engaged" addresses...the "influencers". Some 3.49% of the total address list fall into this category.
You can do it other ways, too. You might, for example, calculate the average total opens for each address on the list over a period of time and take the top 5% as your "best" subscribers.
Now you've probably spotted the big flaw in this approach. Thanks to image blocking, a lot of "engaged" readers may never show up through an analysis based on open rates. You only record an open when a tracking image is displayed.
Next week, I'll look at some simple alternative measures that don't have this problem and explore some more flaws to the whole engagement approach...
Some great examples, Mark!
An assumption could be made that those with an above average open/render rate are more than just engaged, but email social influencers that forward to friends that then may forward to friends.
It's easier to test that hypothesis with email tools that track open pixel renders by email AND IP addresses, but there are other methods that can be used to validate.
By John Caldwell, on 10 July, 2009
Thanks John. Yes, those with the tools need to look at things like IP addresses. And as marketing becomes more democratic and social, finding those influencers will become ever more important.
By , on 13 July, 2009
I agree, it is definitely important to count the repeat opens when you are tracking your email marketing campaign. While it is often the same user reopening the email, I do believe there is a large possibility...in most cases...that the email is being passed through a social chain.
By Sara Kmiecik, on 14 July, 2009
A very interesting approach. Identifying influencers in this way and then incentivising them and their recipients further still, could be an excellent way of growing a database and in turn a customer base. It might be interesting to analyse the most influencial receipients to identify common traits - age, sex, career, location, source... and then try to target that type in the future. In this way you could extend the reach of your marketing massively.
By Stuart Noton, on 16 July, 2009
Thanks Sara, Stuart...agree that finding those social influencers is critical.
I like Stuart's idea of finding common characteristics in influencers and using that to target similar folk in the future. Lots of potential there...
By , on 16 July, 2009
Tons of great information here! Very useful insight, thanks. It's obviously of the utmost importance to keep people as engaged as possible.
I also wonder/worry however for those who open an email 2 or 3 times, sure there is a level of engagement, but was something too confusing to be understood the first time?
By Jake, on 16 July, 2009
Thanks Jake. Yep, there is a danger that repeat opens are triggered by something other than engagement. Which is why it makes sense to choose a cutoff point that is relatively "high" (like 5 times the average repeat opens). Then you can be more sure that the stat really is a reflection of engagement and not simply someone clicking through their inbox a couple of times...
By , on 20 July, 2009
Comments closed for this post