Asking for the impossible: Inbox Guarantees

Yesterday, we had a look at Delivery SLAs as a contract term that you are unlikely to get out of an email service provider. Closely linked with the Delivery SLA is the Inbox SLA, more commonly referred to as the Inbox Guarantee. This is a contractual term that many people try to impose upon an ESP that the ESP will make sure that a certain percentage of mail will be placed in the inbox, as opposed to the spam or bulk mail folder.

I have seen this proposed contract provision take three forms. Under one form, the percentage is figured using the complete total sent. This would result in a bounced email counting against both the SLA and the Inbox Guarantee. The second form figures the percentage against the total number accepted. The final form of this proposed contract provision does not specify how the percentage will be calculated. (For internal purposes, ESPs should treat this one as the first form.)

There are just far too many things outside of the ESP’s control to allow for agreement to that contract term. For the first (and third) forms, all of the arguments from yesterday’s post still apply. And there are still more obstacles to overcome getting into the inbox.

Getting an email into the recipient’s inbox on a consistent basis takes doing a lot of things right. It takes the ESP running a clean, tight ship. They cannot be allowing spammers to use their services. This will cause negative reputation problems across their system. One client’s mail can be negatively impacted by another client’s poor practices. It is up to the ESP to police their network and get rid of the deadwood before that becomes an issue.

It also takes the receiving system’s cooperation. Sometimes things go wrong over there. Yesterday, I mentioned someone accidentally including an ESP’s IPs in their block list. I have seen it happen more often that the wrong URL will get put into the filter instead. This is generally an outgrowth the last paragraph. There is something in common (like a “powered by” or “sent using” button at the bottom of the email) that has remained in the same in all of a “bad” client’s email’s that is also the same as what is in the “good” client’s emails. ISPs generally know enough about that that they try to avoid listing those buttons in order to reduce false positives. But, it happens. Now, all of the ESP’s mail to that provider is getting sent to the spam folder.

It also takes the recipient’s cooperation. Things that go wrong at the ISP generally start by things going wrong at the recipient level. ISPs do not generally pick IPs or domains to filter at random. They select what they will filter based upon user feedback. The provider will then tune their filters to that feedback. Mail which has the same characteristics as the mail complained will get sent to the bulk folder.

A lot of this also remains in the client’s hands. If the client is not sending mail that people want, then the ESP’s hands are tied. They cannot force the mail through, nor can they force it into the inbox.

And then we get to the question of monitoring. This is generally specified by placement statistics from a delivery monitoring company, or if the client is trying to be cheap, the client’s own seed accounts. Now, don’t get me wrong, I think that delivery monitoring companies are great for getting general ideas. What they are not great at is figuring out the real percentage of mail that is delivered, received, and placed because they are not looking at those numbers — only the ISP can see those statistics. All of the delivery monitoring services set up seed accounts which the client mails to and looks at the results from. But those results can change, sometimes even mid-send. That is just a limitation of the system. I have seen some who have attempted to get around this limitation by seeding with user names beginning with different letters throughout the alphabet. But, mail is not always sent alphabetically by user name.

So, what good are they then? Well, they will help you generally figure out if there are problems in need of resolution. But, were I running an ESP’s delivery team, I wouldn’t agree to allowing something that will give you general ideas determine whether we are meeting a contractual obligation to get 99.99996% or more of email into the inbox. The numbers are just too tight for that. There are too many things that can go wrong.