A few months back, LinkedIn noted that after repeated user complaints they'd undertaken a review of their e-mail notifications process, and significantly reduced the amount of e-mails they were sending as a result. Personally, I can vouch for this - like most users, I was getting daily updates about every single group I was in, whether they'd been updated with new content or not. Since the change, however, I'm seeing a lot less - LinkedIn says they've now reduced e-mail sends by over 50%, while complaints about e-mail notifications have dropped a massive 65% overall, a great result.
But LinkedIn's not done - today, via their official blog, LinkedIn has detailed how they're using a new, data-fuelled model to help refine and improve the quality and quantity of e-mail notifications they send users. The new system, called Air Traffic Controller (ATC), uses learning algorithms which take into account member interactions on the platform "to better determine the right frequency of communication you get from LinkedIn, at the right time, and through the right channels".
It works like this...
Coming in for Landing
LinkedIn explains the ATC algorithm by first noting how the system determines relevance and individual preferences, based on on-platform actions.
From the official announcement:
"Our first priority is to determine the right balance of mobile notifications and emails. ATC will help us understand the best time for you to hear from us and which channel you prefer; be it email, push notification or SMS, as well as determining the right amount of messages we send you. We're doing this by paying close attention to your communication-setting preferences and by building intelligence around how you interact with LinkedIn. For example, in the past, we sent an email for every connection invite you received. Now, if you receive a handful of connection invites in a short period of time, our platform will automatically roll that up into a single email."
That example's pretty basic, but you get the idea - ATC will factor in how you indicate your preferences to the system, based both on what you input in your settings and how you interact on the network. And while the first point makes logical sense, the second - 'building intelligence around how you interact with LinkedIn' - is a bit more vague and isn't examined in much detail in the initial documentation.
The three areas of focus for ATC are:
- Volume - LinkedIn's reduced e-mail volume by more than half and is looking to refine this even further. The example provided on this front is that less e-mails will be sent to those who log-in to the site more regularly, as they've most likely seen all the relevant updates on the platform.
- Frequency - LinkedIn's putting daily and monthly limits on notifications.
- Quality - LinkedIn will use ATC to determine each individual's preferences to deliver more targeted content - this measure will be dictated by each user's activity on LinkedIn.
Through this new system, LinkedIn's hoping to better personalize each users' individual experience and subsequent e-mail outreach profile. While little detail is given as to how each action on LinkedIn will influence ATC's understanding in this respect, you'd assume that the system would be refined over time, enabling it to zone in on how each user needs and wants to be provided with updates relevant to them.
As noted in the introduction, LinkedIn's e-mail notifications have long been problematic, and have caused many headaches over the years, so it's great to see LinkedIn taking this feedback on board and working to develop more responsive, intelligent systems. It's also interesting to see LinkedIn taking a more algorithmic and data driven approach to this problem, something they've not done a lot of in times past.
The update also aligns with LinkedIn's recent changes to their on-platform messaging functionality, which has received mixed responses from regular users.
Main image via LinkedIn