I'm not going to pretend I have an easy explanation for every situation where mail lands in the spam folder. I'm just sharing some experience. I'll also say, if spam could be identified by just three metrics, ESPs could just get rid of their compliance teams.
Luke On Fri, Jun 30, 2017 at 4:27 AM, Stefano Bagnara <mai...@bago.org> wrote: > On 30 June 2017 at 00:04, Luke Martinez via mailop <mailop@mailop.org> > wrote: > [...] > >> I've had senders say they remove all non engaged recipients after 7 days >> and somehow their overall sending volume never changes by more than 1-2% >> for weeks or months. Pretty miraculous that they are signing up exactly the >> same number of recipients who happen to fall out of their insanely >> aggressive seven day engagement window...No spam reports though...High open >> rates..Low bounces...Bad inboxing. Almost as if receivers have systems >> designed to catch this kind of thing. :P >> > > "No spam reports, high open rates, low bounces": isn't this a clear signal > the email is wanted/solicited? Why should a system be designed to "catch > this"? > > Isn't the whole spam/antispam thing about "wanted"/"unwanted" stuff? How > "high openrates, low bounces and no spam reports" can be compatible with > "unwanted" stuff? > - You can get high open rates with unwanted stuff but then you'll get spam > reports. > - You can get low spam reports (because no one open or they already > receive in spam), but then you will get low open rates. > > The only "matching" behaviour I can think is "wanted but not so > interesting" emails and an antispam that looks at the "skimmed" emails: you > get a lot of opens but then the emails are only shown for 1 seconds and > then deleted. They don't mark them as spam, but they don't engage/don't > reply/don't keep the email for later. Maybe Microsoft track this "negative" > signal, and use it a lot, while this signal is not exposed in SNDS data. > > AFAIK no big provider ever published they track "skimming/trashing" > behaviour in their antispam strategy, but this could be a very good > "weapon", generally speaking, and this could lead to the above case (high > opens, no abuses, no bounces, but considered spam... because > skimmed/trashed more than expected). But how can an MTA on the "sender > side" correctly do outbound antispam in absence of data to detect what is > spammy and what is not? > > Once you get "Bad inboxing" you will not be able to do high open rates any > more: most recipients doesn't even know they have a "spam" folder that > could contain false positives unless you tell them to look there. > > Stefano > > -- > Stefano Bagnara > Apache James/jDKIM/jSPF > VOXmail/Mosaico.io/VoidLabs > > _______________________________________________ > mailop mailing list > mailop@mailop.org > https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop > > -- Luke Martinez Team Lead | Email Delivery 520.400.5693
_______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
