Finding Your Best Time to Send Email

Are you struggling to determine the best time to reach your email recipients? Welcome to the happy club. Identifying your best possible send time is no easy task, and studies that claim to have identified the ultimate delivery hour can often make matters worse, especially if other email marketers listen.

Best Time To Send Email

Earlier this month, MarketingSherpa provided some survey results data in which respondents were asked when their most effective and least effective days to send email. The data indicates that there really is no definitive time or day for email delivery, but nearly half of those surveyed found Tuesday and Wednesday to be the most effective.

Email Send Times Study

To this study’s credit, it does not declare a winner. However, studies like this can be a bit misleading not because of the questions they ask, but due to the answers we derive. As I’ve mentioned before, relying on numbers from other email marketers can be dangerous.

Are Tuesday and Wednesday the best possible delivery days for your email program? Maybe, but maybe not. Is there a “best send time” for your email program? Absolutely. Is your witching hour the same as mine or any other email marketer? Probably not.

Our challenge is to find the unique day/hour/minute that is right for our unique program. Here are a few thoughts on how to overcome this test.

A Note on Transactional Email
If you are primarily sending transactional email or messages that are triggered in response to a customer action (product purchase, list subscription, content-for-contact quid pro quo exchanges), the following thoughts do not really apply. The best time to send transactional messages such as “welcome”, “thank you” or “confirmation” emails is most often immediately as triggers should provide instant gratification for their intended recipients.

The following recommendations apply primarily to recurring messages to an opt-in list. Before you consider how to approach selecting your best time to send, also know that the majority of opens (51.7% to be exact) will likely occur within the first six hours of pushing the “send” button.

Applying the Scientific Method

The most common advice you will likely receive about finding your best day and time to send emails will probably sound something like this:

Test, test … and test some more you big silly! (tee hee)

But what does that really mean? Let’s bring it back to the old school – I’m talking your 4th grade science class in which you likely were introduced to the wondrous Scientific Method. Here is a quick recap:

  1. Ask A Question (What is the best day and time for us to send email?)
  2. Perform Background Research (We’ll cover this in a minute.)
  3. Construct a Hypothesis (If we send at {INSERT DAY/TIME} to {INSERT AUDIENCE}, our open/click rates will improve)
  4. Conduct an Experiment (Conduct an A/B split in which one randomly selected portion of the list receives the message at {INSERT DAY/TIME}, and the other half of the list receives the message at a Control time.)
  5. Analyze Results (Draw a conclusion with test results data.)
  6. Report Results (Decide to conclude this test or formulate a separate hypothesis.)

Research to Inform Our Test

If we are looking to understand how our email subscribers have behaved historically, there are two primary sources of data: email behavior and other online behavior.

Previous Email Behavior
Using my own email list as an example, here is a graph of open rate by day of the week and another that shows open rate by day part (early, morning, afternoon and night). You’ll note that the Email Email, which was started in January, has only been sent on the last day of the month without a set time for delivery. This has allowed me to gather a fairly random sampling of data, but you’ll see that the email has never been sent on a Monday or after 5pm Phoenix time.

Email Open Rate Results

Given the limited data and delivery exposure of the Email Email during certain days and times, the following is a bit presumptuous. However, the data leads to specific questions that can be answered with a testing program such as:

  • Is Thursday truly effective as a day to send? Should I test sending out on the last day of the month vs. the last Thursday of the month?
  • What impact would sending at night have on open rate? Should I test sending during the lunchtime sweet spot vs. an early evening hour?

Time of Day HistoryThe graphic on the right shows opens for an actual client email campaign. The line graphs show the volume of opens over time. This particular client has performed day of week and some time of day testing in the past. We found that Tuesdays were most effective and the late morning to early afternoon time frame provided the highest rate of opens.

However, when you look at the three graphics to the right (pulled from MailChimp) you will notice a trend by the audience to crack open the email early in the morning around the 6am to 7am time frame. Despite the fact that this client has conducted time of day and day of week testing previously, it is always a worthwhile consideration to revisit the testing process. Two questions arise from this cursory analysis:

  • Will open rate and/or click rate change if we send the message early in the morning?
  • What is the impact of segmenting our list by “time of open/click” behavior? In other words, will our overall email metrics improve if we split our list into two segments: early birds and the lunch crowd?

Other Online Behavior
If metrics from previous email sends are limited or non-existent, we can rely on website and social media behaviors to help us design a hypothesis. Please note that the following considerations all rely on the assumption that your audience is apt to open email around the same time they conduct other online activity such as browsing the web, shopping online or scanning their social media accounts.

Here are a few examples in which we can rely on the findings from other online data sources to help us find our best time to engage with email subscribers.

The two graphs below were compiled from Google Analytics data from a client. You can see that desktop visitors are prone to visit the site during the mid-day hours and mobile visitors are viewing pages at night. Once again, if we are to infer that the device used to browse the web is the same device used to check email and both behaviors are likely to occur around the same time, we could come away with the following question:

  • How will email metrics be impacted if we split our list by device usage and send desktop recipients messages at 11AM and deliver messages to mobile recipients at 6PM?

Google Analytics Time of Day

The chart below was pulled from Zuum, a social media analytics utility that is capable of showing what days and times Facebook users are most likely to interact with your Facebook posts. While much of the engagement metrics are a reflection of the quality of the post itself, this data can be used to at least give us an idea of when Facebook connections are online and ready to interact. In the graphic below the size of the circle represents the volume of posts and the depth of color indicates engagement rate (dark blue = high engagement).

  • From this data, you may decide to conduct an A/B/C split test in which email metrics are compared for the send times of Sunday at Noon vs. Tuesday at 8PM vs. Friday at 4PM.

Facebook Time of Day Analysis

Finally, if you know that there is a heavy overlap between your email subscribers and Twitter followers, consider using a tool like Tweriod. This tool helps us to find when our Twitter followers are most active. The following graphic shows Twitter activity from my most recent 1,000 followers, which begs the question:

  • What will the impact of a early morning vs. afternoon test have on email metrics?

Twitter Time of Day Analysis

Once you have determined your hypothesis, actually running the test should be quite simple. Most email marketing delivery platforms allow for simple A/B testing by subject line, from name, content and time of day. Here is an example of the A/B split interface from ExactTarget.

ExactTarget AB Testing Interface

Applying the “Set Expectations” Method

Outside of monitoring and testing metrics from email subscriber behavior data, perhaps the easiest approach to ensuring your program utilizes the best send time is to A) tell the audience when they will receive messages or B) ask them when they would like to receive messages. For instance, many email programs will define their delivery schedules upon sign-up. With the latter option, many organizations empower subscribers to pick and choose when they would like to receive email.

The following is an example from OpenSky, an online marketplace for women seeking to buy and sell jewelry, clothing, home decor, cosmetics and products from other categories (a kind of Etsy competitor). Upon signing up to participate in this online marketplace, OpenSky customers can select the days of the week on which they would like to receive updates and control other account-related alerts. For OpenSky, time of send is likely less of a concern as the customer is in total control.

Email Preference Center

Time of Day Subscribe OptionIn a similar and much less complicated fashion, email marketers who deliver on a weekly or monthly basis can ask subscribers when they would prefer to receive emails. Here is a “practice what you preach” example from my own subscriber form on the right.

No matter what method you choose, know that there is indeed a “best” time for you to send email. However, that time is not determined by an industry study. It should be uncovered by recipient behaviors and preferences.