> On 10 May 2019, at 11:40, Michael Wise via mailop <mailop@mailop.org> wrote: > > > The solution ... is to stop thinking that all decisions on the legitimacy of > an email can be resolved at the instant the machine is focusing on that one > email. > > We need to move beyond "Spam" filtering as a one-off, per message task and > start working more actively on techniques of campaign detection, both > outbound, as well as Inbound. And that requires noticing patterns of good > behavior going over a line and becoming abusive.
I’ve been thinking about this a lot. All the modern consumer filters (and the business ones) are actually really good at separating out mailstreams. You don’t need to separate your transactional and your marketing mail on different IPs because the filters aren’t using IPs as the unique mail identifier. They can identify streams. Yahoo’s been able to very effectively separate out streams for almost a decade - and everything I’m seeing from the shared Verizon Media infrastructure says that bit of the filters were kept during the rebuild. Google is incredibly good at it as well and even different marketing streams with the same d= will be filtered differently - even if it’s just “updates” vs. “promotions” tab. Microsoft is a bit of an outlier in that they still seem to put a lot of emphasis on IP address reputation more than domain reputation, but even then I think they are more flexible on the back end and look at content (and recipient pools) than some senders believe. I actually wrote about this last month: https://wordtothewise.com/2019/04/gmail-suddenly-puts-mail-in-the-bulk-folder/ “They measure mailstreams over time, they don’t measure individual sends.” The Yahoo evidence was a client was sending content and marketing to the same COI list over the same IP with all the same domain authentication. Yahoo was putting the content (wanted) in the inbox and the marketing (eh… who cares) into bulk. Even Return Path certification of the IP wasn’t enough to get the marketing into the inbox. This was a client from ’11 or ’12? Quite a while ago, anyway. The Google evidence is more recent. I was reviewing google postmaster tools data with a client. This client didn’t include a FBL identifier in their email, but Google “discovered” one anyway - something like v1 or L3 or something. When asked the client says “Oh, that’s the name of the customer segment we’re sending to.” Basically, from the emails coming in, Google was able to discern that this segment was different than other segments and correctly assign a FBL rate to that particular segment. The consumer ISPs can clearly identify mailstreams already. laura -- Having an Email Crisis? We can help! 800 823-9674 Laura Atkins Word to the Wise la...@wordtothewise.com (650) 437-0741 Email Delivery Blog: https://wordtothewise.com/blog
_______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
