Putting Big Data to Work: What’s Going on with Gmail Opens?

Every month, the Big Data team here at Constant Contact looks at the previous month’s email open trends in order to learn more about what devices our customers’ subscribers use to view emails and how those devices have changed over time.

The graph below shows the relative percentages of email opens among our customers’ email subscribers on four device types:

  • mobile (which includes all smartphones)
  • tablets (devices like iPads and Android tablets)
  • desktops (good old-fashioned personal computers)
  • other (which includes gaming consoles, set-top boxes and automated email reading tools)


Spotting trends

One of the advantages of Big Data technologies like Hadoop and Hive is their ability to uncover trends in vast data sets, and you don’t have to be a numbers geek to appreciate these trends. Let’s first look at the data up to October 2013 and then have some fun interpreting the unusual data after that time.

First, there are three clear trends in desktop, mobile and tablet usage in the first 10 months of the year:

  • Desktop computers are being used less and less – from nearly 60% of opens at the beginning of the year down to 50% in October.
  • Mobile and tablet opens are increasing in that time period. This tells us that the usage is shifting towards these portable devices and away from the desktop.
  • The ‘other’ category is relatively flat in this period – which makes intuitive sense, since smart televisions and gaming consoles are not an obvious first choice for reading email.

The fun starts around November 2013. We started seeing a slight uptick in opens in the ‘other’ category. Specifically, we look at the User Agent string that the subscriber’s browser sends when reading an email. It tells us the type of device used to open the message. All devices, including browsers, gaming consoles, set-top boxes to name a few, send this string as a way to help identify which device opened a particular document.

When we took a closer look at the ‘other’ category, we found that it was the truly unknown devices that were on the rise. By February of 2014, the ‘other’ category jumped to 11 percent of opens, and we were quickly able to determine why.

Impact of Gmail image caching

As we’ve previously addressed, Google’s Gmail decided late last fall to cache images, including images that many email service providers use to track opens. The User Agent string associated with these cache reads was not consistent until February, when it became clear that the new ‘unknowns’ were really the Google download tool running on its image proxy servers, accounting for 11 percent of all email opens that month. During this same time period, the graph shows that while the ‘other’ category was increasing, the desktop category was decreasing. This also makes intuitive sense since the image caching only affects users of the Google Gmail client, not users of a generic client such as the Apple mobile mail reader.

Email continues to be the one of the most fascinating areas of marketing in terms of analytics. When big players such as Google make a change to their system, we get to see firsthand how it affects their user base and, in turn, affects our customers. The image caching development presented us with an opportunity to learn a little more about Gmail and those who use it while still being able to report accurate overall open rates to our customers.

Trend spotting and analysis such as this only scratches the surface of what is possible with Big Data, and we continue to aggregate and analyze data and help our customers become better small business marketers in the process.

Open rates can reveal a treasure trove of information on how customers interact with your business. How have you used open rates to better craft future campaigns? Tell us in the comment section.

Leave a Comment