One of the most frequent emails I answer from users is:
I was testing your open-tracking, and I know that certain emails weren’t opened, but GMass still reports them as having been opened.
99% of the time this is because the user opened his own sent emails from the Sent Mail folder in Gmail. Doing so triggers the download of the open-tracking pixel, which is how GMass, or any email marketing system, knows if an email is opened. Opening the email up from the Sent Mail folder is equivalent to the recipient opening the email from his/her own Inbox.
Additionally, if a particular email address bounces, and you open up the bounce notification, and the original email is underneath, then an open could be triggered also because you’re now viewing the original email underneath.
Another scenario: you set GMass to just create Drafts for a campaign rather than actually send the emails. You then open each Draft individually to customize it. GMass will count you opening the Draft as the recipient opening the email.
How do we solve this?
I have some ideas as to how to provide more accurate open-tracking results in this particular scenario, but none is a perfect solution. Given that we’re integrating email marketing into a Gmail account, this system is by nature different from a traditional email marketing system like Constant Contact or MailChimp, where the user would never have access to the actual sent email, and therefore couldn’t accidentally trigger an open.
The primary method I can separate opens from the user in his Sent Mail folder from opens by the recipient in his Inbox is by recording the IP address that the user uses to trigger the email marketing campaign, and then treating subsequent opens that originate from that IP address as suspect. Even this seemingly simple solution, however, could lead to invalid reporting data. If in an office, there are 100 people working at their desktops, and the office has an Internet connection such that all 100 people share an IP address, and one person uses GMass to send a mass email to all 100 people, then it would be expected that legitimate opens come from the same IP that the sender used to trigger the mass email.
Therefore the best solution may be to count opens from the same IP as the sender as “suspect” and flag them appropriately, but let the user decide if they are legitimate opens or not. Enhancing reporting in this way is on my GMass product roadmap.