↖ Back to articles

On Opens & Click Rates

about 1 month ago
TLDR: We believe opens and clicks are a weak measurement of success in email marketing and are working towards replacing it with more sturdy metrics based on web activity/API behaviour.
It's been an interesting journey running an email marketing provider and sending millions of emails out to our customers and subscribers each week.
The interesting part for us being what people are actually doing with the signals (opens and clicks) that they are creating through their sending.
We have some customers using a click or an open as an indication of intent, while others use it as an indication of "who is active", and we also have customers using these signals as ways to move people around their customer journeys (from automation A to automation B).
But, after working on now hundreds of campaigns and seeing all sorts of set ups, I'm no longer convinced that these metrics are worth their current reputation in the industry.
For most email marketing providers, an open is defined as either two things happening: 1) a small, invisible image is opened in the email client, or 2) a user clicks on a link, therefore implying they opened the email to actually get to the link.
At first, this seems like it all makes sense, but once you dive into the data that's actually being logged, it begins to fall apart.
For instance, what happens if a link is opened by the email client as a way to check links for spam or fetch additional data?
For a lot of email clients, this is the default behavior and in our logs, we've noticed this happening a lot more than we would have expected. For example, did you know that Outlook sends all links to Bing for crawling (yes, really) and is the cause of a lot of one-click emails failing which is why the web moved to time-based expiry vs. click-based? Though not confirmed, we also believe this happens with free Yahoo and Gmail users.
And ESPs love treating these "robo-clicks" as real human intent.
After all, ESPs are incentivised to include these numbers in their stats. 
They want to show you maximised open rates, at any cost, so you think your delivery is better than the competitors you're evaluating them against.
This is also why, if you've used click triggers before, you'll notice a lot of inconsistencies coming from your campaigns as email clients are opening specific links and putting users into campaigns they had no intent of joining.
And perhaps the worst part of this, users won't complain. 
If a user is giving an email that is supposed to be used to place them into campaign A, B, or C and the email client automatically clicks A, then no matter what the user responds with they'll get campaign A's sequence. Not great!
We see this is a pretty big problem on the email analytics side and are doing our best to try and brainstorm better metrics than what the industry standard is (which we do, begrudgingly, provide).
The concept that we're currently running with is a new concept called a "TrueView". In Bento, a TrueView is very similar to a ViewThrough conversion as seen in other ad analytics platforms, and it’s only triggered when a customer specifically clicks on an email and successfully hits the website that Bento is installed on.
When a view event (or purchase) is fired back, we'll then look up the last email that they clicked on, in a given period of time set by you, and attribute that view to that email.
So, say you send a customer an email and the attribution window is 1 day. If they commit any event within that day, then that would log as a ViewThrough.
We've also placed some constraints to what type of events are captured to exclude referral traffic and only include Direct + organic search (i.e people who search for a branded phrase for navigational purposes). This ensures that if you're driving traffic from social channels, like via a remarketing campaign, that will not trigger a ViewThrough event for those visitors.
This metric, in practice, provides a lower rate than opens BUT is sturdy and shows the true intent of the visitor. It's also not being triggered by email clients or bots which rarely execute Javascript on the other side of the link.
We're also starting to encourage visitors to use our auto-tagging feature as a way to indicate a visitor’s intent INSTEAD of a link trigger.
We find link triggers to be a flaky mechanism and believe a visitor physically landing on the page to be more worthwhile.
So, if you want a visitor to join a specific sequence based on a click, then send them to 3 different landing pages and automatically tag them based on which page they visited. These don't even need to be different web pages - they can just have different parameters at the end of them OR you can even just add bento_tags="test" to the end of your URLs in the email to tag them when they hit the site.
We're still thinking on this and want to bring something a little different to the email marketing industry, so if you have any ideas please send them over and let us know what you'd like to see.