ISPs reveal deliverability tips for their own inboxes

Deliverability Facts from the horses mouth

This year’s Email Evolution conference from the EEC had a very popular and enlightening last talk; representatives from Gmail,, Aol and Comcast revealed some great details about how their platforms handle reputation and inbox placement.

Spam Traps

I have mentioned before about Yahoo, Hotmail and AOL recycling dead addresses into traps, I didn’t invent the idea it was from research and close sources. have now denied it and AOL said it had happened but not if it still was. also stated that an account would be shut-down after 2 years of inactivity, not the 7-14 months we had previously been led to believe.

Gmail confirmed that there was no recycling but they have not been around for long compared to the rest so have no need for it … yet.


  • Opens – Good, obvious engagement
  • Clicks – no-one tracks clicks, it’s a breach in privacy.
  • Replies – Good, signifies a conversation
  • Mark as Spam – Bad, of course
  • Mark as Not Spam – Good, big noise that the ISP has got it wrong
  • Delete without Opening – Bad, signals no interest but with less offense than a mark as spam
  • Move to a folder – Good, signals value in the email’s content
  • Add to address book – Good, signals value in the sender.

The only new one on me here was adding to a folder but it makes sense, it’s a keeper.

No mention of stars (or pins) though, I’d expected that to be good one. Unless it’s such a short term thing it carries little weight whereas folders (labels) are more permanent.

Inactives said that inactive subscribers do not affect/increase reputation based junking but in an individual’s inbox it may. Subsequently they do not support the notion that kulling the list of inactives will help reputation but added that the people who rarely open your emails are more likely are to take negative actions.

Gmail didn’t openly agree with and they suggest going with a ramp-down for frequency of emails to inactives before you stop sending to them at

Daily ramps down to weekly, weekly ramps down to bi-monthly or just monthly etc. etc.

I was a little surprise that they suggest stopping all together from 3-6 months, however, this may have been more targeted to daily senders but apparently the Gmail marketing teams are advised to abide by this.

I’ve always been of the opinion that there was a level of percentages being applied to volumes sent that would affect reputation, rather than just reducing actual actions that were negative. Senders who have had higher open rates have always enjoyed more content freedom and sending volumes & speeds. This may not be the case in and they may put more weight on actual reactions.

The ramp-down has legs though. We already have a ramp up for new IPs.

If you do a ramp down, don’t just keep the same content, pick and choose and experiment, you need to get an open from these people so they get back in the engaged segment … more to come on that I expect.


The short version is that not being authenticated is just bad!

DKIM plays a big part in most inboxes nowadays but it is rumoured that SPF makes little difference even to who championed/invented SenderID as the evolution of it but it’s best to have at least those two at all times. DMARC is newer and not yet a requirement, most ISPs use it themselves but do not make it a requirement for delivery.

Content Filters

All agreed that there were no longer any “spammy keywords” which their systems looked for to decide on inbox placement. However, certain copy in the subject and content are more likely to result in negative reactions from users and reputation could then be affected.

I personally have found that I can take out certain collections of words on a junked email and get inboxed in and Gmail.

I’m of the belief that either those content filters do exist and the leniency of their filters depend on reputation.

Alternatively actual copy can have reputation, whether it is a word or a collection of words, which could be learned from user actions. This could have been an evolution of the old bayesian filters.

Either way they so it ain’t so.


Deliverability is personal to each inbox

… was the catchphrase, which essentially concurs with any deliverabiltiy experts’ advice:

reputation is one thing but each person uses their inbox differently and gets different emails, so each inbox learns differently so can react differently.

So it’s nice to get something from the horses mouth. From experience, I’m not ready to accept all of it at face value but I will adapt my priorities slightly. I hope this is of use to anyone else.



Further Reading

AOL, Comcast, Gmail, and open up at EEC15” by Massimo Arrigoni
How 4 Of The Top Mailbox Providers (Isp) Determine What Gets To The Inbox” by Al Bsharah

Image from

3 thoughts on “ISPs reveal deliverability tips for their own inboxes

  1. “DKIM plays a big part in most inboxes nowadays but it is rumoured that SPF makes little difference even to who championed/invented it but it’s best to have at least those two at all times.”

    Outlook championed/ invented SPF? Sender ID, yes, but I didn’t think they had anything to do with SPF.


  2. Fair point Kent,
    I normalised SPF and/OR senderID into just SPF.
    Seeing as it is worthy of a comment, I’ll make the edit.
    Thanks for caring enough to give me the heads up Kent, love your work!


Comments are closed.