Let's track Apple MPP opens


Okay, they're not really opens. They're proxy-based, false positive pre-loaded opens. Don't think of them as opens. Think of them as something you want to get out of the way so as to not inflate your campaign tracking. I know, I know... Apple is not the only one proxying opens today. But they're the big one -- over the past few weeks, Apple MPP "opens" have been 25-30% of my opens tracked for each Spam Resource newsletter. And if you're a typical B2C sender, your percentage is likely even higher.

So let's say you run an email sending platform and you want to give users an option to suppress these opens from tracking. How do you do that?

  • Look for any open that has the very generic referrer (user agent) of "Mozilla/5.0" with nothing else. This is probably low effort, but keep in mind that it is imperfect. In my testing, 95% of the opens I am identifying via this method are Apple MPP image loads. The other 5% are not. YMMV.
  • Look for any open from an IP address owned by Cloudflare, Akamai or Fastly. These seem to be the primary CDN networks used by Apple for the proxy process. (Though keep in mind that there are possibly others.) In my case, the opens all seemed to come from IPs in network blocks owned by Akamai (172.224.0.0/12) or Cloudflare (104.16.0.0/12).
  • Look to Apple for the precise list of IP addresses and network blocks used for their "iCloud Private Relay" proxy service. This should cover all possible IPs used for proxy image loads. I'd suggest being careful here to not suppress CLICKS tracked via these IPs. The private relay service traffic is going to include legitimate, human traffic clicking on and following links. You can't trust that the IP address is the true IP address of the user, but in theory, any traffic generated (beyond image loading) should be considered legitimate. Read more about iCloud Private Relay here.
And remember, for the time being, even with Apple MPP in play, open tracking can still be handy to try to identify the core of your unengaged subscribers. No open = there's probably nobody home. You've still got an opportunity to segment based on negative engagement, which can come in handy in certain scenarios.

I find it interesting that my own personal data shows a lower percentage of MPP false "opens" than what most folks are seeing as a B2C average. Chalk that up to my B2B subscriber base and small sample size -- y'all are a bunch of email nerds who are perhaps more likely to be sitting in front of a computer at a desk while working on solving various email challenges. Quite a different demographic compared to a typical B2C retailer's subscriber base.

(Special thanks to Steve Atkins for the tip about Fastly being one of Apple's network partners.)

Post a Comment

Comments