All notifications emails coming out of a Sampa registration process use the "sampa" alias. So if somebody uses a bad email address like "DontExist @yahoo.com" and it bounces, it goes directly into my inbox.
We always had our share of bad emails. Usually 5-10% of people that signup for Sampa use a bad email address. But on the last 4 weeks or so, the number had grown to 10-15% per day of bad email addresses.
I always keep an close eye on it because we had problems with AOL and Rediff blocking all our emails (they thought we were a spam server).
So I went to do a close analysis of one week of bad emails.
Some surprises, some not. We had 4 categories of mistakes:
- People using fake email address, like "fake @fake.com" or "test @test.com"
- People misspelling the domain, like "joe @ yhoo.com"
- People misspelling their alias, like "jeo @ yahoo.com" (intended joe @yahoo.com)
- People using a Sampa address, like "joe @sampa.com" or "joe @sampasite.com".
For #1, there is not much we can do except clarify that the email must exist. So we changed our language on the website. Before it said you must enter a "valid email address", now it says "an existing email address".
For #2, we do a DNS lookup now for every email address. If the guy types xyz@yhoo.com we will flag it, unless, of course, the domain was registered by somebody (at that happens a lot for misspellings of Yahoo, Hotmail, GMail, AOL, etc.)
For #3, there is not much we can do. Yahoo never tells if an email exists during an SMTP session. Nor does AOL. Hotmail does tell you that. Don't know about GMail. Doing that test is hard because I need to find out the MX record for that domain, then do a "special" SMTP conversation to figure it out. It is quite a bit of code and I don't have the time now.
For #4, we just flat out prohibit the use of our own domains, like "@sampa.com", "@sampasite.com", "@brainuse.com", "@inthesphere.com", etc. That is a problem because we cannot use our own work emails to sign up for Sampa (but we have an easy workaround).
What have the results been? We don't know yet because I only have two days of data. I want to compare a full week to see the number of people that registered end up confirming their addresses or not.
I'm the Co-founder & CTO of