Project Omnivore

Ben at Mailchimp has posted some information about Project Omnivore. This is a predictive system that not only predicts potential abuse, but can also be used to predict poor campaigns. Steve and I had a chance to see Omnivore in action when we were in Atlanta last fall, and were impressed by the accuracy for bad stuff. It seems, however, that Omnivore is useful to predict good behaviour as well.

The reason we built Omnivore was because we wanted to change the way we think about abuse. The project involved so much data crunching that it resulted in some interesting byproducts. Our subject line suggester is one example, as well as the engagement ranking and segmenting tools we mentioned earlier.
[…]Omnivore is learning more every day, and is actually getting good at predicting not just bad behavior, but good behavior too.

Related Posts

AOL EWL: low complaints no longer enough

This morning AOL announced some changes to their Enhanced White List. Given I’ve not talked very much about the AOL EWL in the past, this is as good a time as any to talk about it.
The AOL Enhanced Whitelist is for those senders that have very good practices. Senders on the EWL not only get their mail delivered to the inbox, but also have links and images enabled by default. Placement on the EWL is done solely on the basis of mail performance and only the best senders get on the list.
The new announcement this morning says that AOL will take more into account than just complaints. Previously, senders with the lowest complaint rates qualified for the EWL. Now, senders must also have a good reputation in addition to the low complaint rates. Good reputation is a measure of user engagement with a particular sender.
This change only reinforces what I and many other delivery experts have been saying: The secret to good delivery is to send mail recipients want. ISPs are making delivery decisions based on those measurements. Send mail that recipients want, and there are few delivery problems.
For a long time good delivery was tied closely to complaint rates, so senders focused on complaints. Spammers focused on complaints too, thus managing to actually get some of their spam delivered. ISPs noticed and started looking at other ways to distinguish wanted mail from spam. One of the better ways to separate spam from wanted mail is to look at user engagement. And the ISPs are measuring engagement and using that measurement as part of their decision making process. Send so much mail users don’t read it, and your reputation goes down followed by your delivery rates.

Read More

Delivery case study

Mailchimp published a case study looking at a customer moving from sending in house to using Mailchimp.

Read More

Email related predictions for 2010

As my recent series of posts has indicated, I am seeing a lot of future changes in the email industry.

Read More