Assessing the best time to send email

Latest posts | Feed | By Mark Brownlow

various clockfacesIn Jane Austen's Mansfield Park*, Mary Crawford says "I cannot be dictated to by a watch."

Unfortunately, email marketers dare not share that opinion.

We fuss over subject lines, offers, calls to action and targeting, but sometimes a lack of response is simply a matter of timing.

By the time the recipient checked her inbox, your carefully crafted email was buried by Facebook alerts, business memos and a reminder to pick up a bottle of wine on the way home.

Timing matters.

At this point in any article on "the best time of day to send email," it's traditional to say "it depends" and advise testing to discover the best time for your particular audience.

Which is very sound advice, but let's see if we can come up with something more.

You've probably been through the brainstorming process to draw out a "best guess" time to send. And testing may not be an option if you haven't the time or a big enough list. So where do you go from there?

1. Look at your past results


As noted before, there's a lot of insight already sitting in your campaign reports.

How about taking 6 months' worth of emails and graphing time of send against response and seeing if any patterns emerge?

Specifically, are certain send times always associated with a higher response?

2. Exploit benchmark data on open patterns


If you always send emails out at the same time, you have nothing to compare. But here's an idea...

MailerMailer's half-yearly email marketing metrics report includes a graph comparing opens per hour with time since send.

You'll find the original graph in the latest report, and it looks something like this:

open rate graph

As the numbers are drawn from thousands of email campaigns, you can treat this graph as a typical open rate curve.

Generate your own curve using your campaign stats and compare it to the MailerMailer benchmark. Suppose you see this:

open rate graph

The deviations are telling you something. Is there a timing issue at play?

For example, my newsletter's opens/hour follows a similar curve until around 13-14 hours after the send, when there's a bump.

My email goes out at 1.30pm Eastern Time. That's 7.30pm for most of Europe and that bump represents European readers opening their inboxes the following morning.

The lesson? Even though the newsletter is buried a little in inboxes, it still gets an "unusually" large amount of opens in the European morning.

I should delay sending the newsletter out to Europe until, say, 9am local time so my email is even higher in the inbox when marketers arrive for work.

3. Segment by time of response


That simple example highlights the fact that any "optimal" send time for your whole list is actually a compromise.

It's not "optimal" at all.

Optimal is when you send out each email at the best time for that individual recipient.

At least one ESP provides that functionality in their system, sending out each email based on when the recipient opened and clicked on previous messages. They report big response improvements as a result, with one customer lifting net revenue by 40%.

That kind of functionality is not accessible to all, but you can still apply the principle to your list. Instead of defining segments according to what recipients open and click on, you can look at when they open and click.

So you might be able to build segments like...
  • "checks email between 6am and 10am EST"
  • "checks email between 10am and 1pm EST"
  • "checks email between 1pm and 4pm EST"
  • "checks email between 4pm and 9pm EST"
  • "checks email between 9pm and 2am EST"
  • "checks email between 2am and 6am EST"
...and time your emails to each segment accordingly.

If you have data on recipient location, you can segment by timezone. So if your "best guess" time to send is 9am, you can ensure that people in California, Connecticut, Cologne and Canberra all get the email at 9am local time.

Even without geographic data on your subscribers, you can do some timezone segmentation by matching TLDs (the last bit of the email address) to a timezone. For example, people with .de (Germany) or .fr (France) addresses are likely to be on Central European Time.

[Here's a list of TLDs and their associated country or region.]

4. Make the question obsolete


An alternative option to second guessing the best time of day to send out your emails is to do away with the question entirely.

Eh?

Control of timing becomes less critical to success where the recipient either determines the time of send for you (a concept also applied in the one-to-one model described above). Or where the recipient is determined to seek your emails out, irrespective of when you send them.

Both those concepts are explained in the New Email Marketing series, specifically this post from last year.

5. Think beyond email


Clearly timing is an important issue in email marketing. But many of the concepts involved apply equally to other channels.

Do you, for example, apply as much care to the timing of your blog posts, Tweets or Facebook updates as you do to your emails?

6. One problem


Nearly all articles on the best time to send out email assume that every email leaves the delivery system simultaneously and reaches the recipient instantaneously.

Limitations imposed by both sending and receiving systems mean this just isn't the case, unless your list is relatively small.

So if you have a "best time to send," when do you actually set the email campaign to go? Do you know how long it takes to get all the emails out the door? And should the "best time to send" be at the beginning, middle or end of this time period?

I don't have answers here, but you need to ask the questions. (Love to hear how you deal with this issue.)

This very problem is, however, yet more reason to segment by response time.

If you break up your list into smaller segments, each getting emails sent at different times, then you have a much better chance of delivery times matching intended send times (there are less emails for the systems to get through at any one time.)

Coming up: the best day to send your emails (which also, of course, affects the best time! Fun, isn't it?)

OK, over to you - thoughts?

*Austen novels are a key source of online marketing insight.

More on timing | Tags: ,

Permalink | June 02, 2009 | 12 comment(s)
Get posts like this: as an RSS feed | biweekly email | via Twitter

12 Comments:

Another great post. I have to say that when I send out my emails to my list, I never wrote the times down or tracked the results. But have been aware of the different timezones. After the results then I start to think about when is the best time to send out emails.
I guess that tracking and making a graph of the results will gradually eliminate the question "When is the best time to send out the emails?"
I still think that after tracking and having a good idea of when there's no right time to send out emails. It all depends of your list and your campaigns.
By Blogger Erasmo Velez, on 02 June, 2009  
 

Thanks Erasmo.

Another point I didn't mention is that the "best" time can change of course, depending on what your email is all about.

A restaurant advertising lunchtime work specials probably wouldn't want to send the email after lunch. But if the next email pushes post-work cocktail specials, after-lunch might be OK...
By Anonymous Mark Brownlow, on 03 June, 2009  
 

Excellent post Mark. It is very important to mention that if you have a very large list, they are not all going to go out at the same time.

Thinking about time zones is key as well.
By Anonymous Michael, on 03 June, 2009  
 

Great article. Tons of informative information for me to use with our company. Thank you!
By Anonymous Restoration Media, on 04 June, 2009  
 

Thanks Michael, RM!

Mark
By Anonymous Mark Brownlow, on 05 June, 2009  
 

"A restaurant advertising lunchtime work specials probably wouldn't want to send the email after lunch."

Case in point: I received a 'One Night Only: TV Dinner Takeaway' promotion for the Championship League final from Gourmet Burger Kitchen the day after the match took place...so not much use to me really.
By Anonymous Gareth Cutter, on 05 June, 2009  
 

Excellent Gareth! Or maybe they were just 364 days early.
By Anonymous Mark Brownlow, on 05 June, 2009  
 

Mark - Great considerations when thinking about time of day testing. A/B split testing for some of my clients has revealed major differences between sending at one time versus another.

I find that most marketers typically just choose a time and run with it. Taking the time to test can make a big difference in your results.

Another interesting find: One particular client found that not only did Sunday garner more conversions for her business (versus sending out first thing Monday morning), but also gave her a head-start on fulfillment, leading to faster delivery. Just another coincidental perk uncovered! Like the smart marketer she is, she plans on continuing to test to see if a different time on the weekend works even better.
By Blogger Kristen Gregory at Bronto, on 08 June, 2009  
 

Kristen, very intrigued by your example. It's not just about response, but related topics like the practicalities of handling that response. Excellent!
By Anonymous Mark Brownlow, on 08 June, 2009  
 

Tracking emails is a process just like tracking errors in a production environment. Every element (copy, offer, format, timing, audience segmentation, etc.) can be tracked/graphed. The key is:
What data needs to be tracked?
How do look at the data?
What positive action can be determined?
What data tracking is irrelevant and should be eliminated?
By Anonymous Roger, on 18 June, 2009  
 

Pure360 can decide for you from each recipient's open history. If there is a pattern we will send the message at the best time to open, if there isn't a pattern we will release it straight away.
By Anonymous Andy Thorpe, on 19 June, 2009  
 

Very informative. Thanks.
By Anonymous Madi, on 26 October, 2009  
 

Comments closed for this post