Through my position here at Word to the Wise I’ve interacted with dozens of companies over the years. Some companies outsource the mechanics of email sending to email service providers, others buy a software or MTA solution from one of the many vendors out there. For both these groups delivery problems are usually issues with permission or user expectations. Technically there are few problems with sending, bounce handling, unsubscriptions and rate limiting. The commercial software, either as created by an ESP or a vendor, typically does these things well.
The last group, those who use a home built system, are a whole different story. They often do no bounce processing relying on the underlying mail transport agent (typically qmail) to do all that work. The problem is that a general mail transport agent handles bounces for a particular email send, but does not have any functionality to handle future emails to addresses that bounce. Consequently the list does not get bounce handled, dead addresses pile up and their delivery rates plummet.
A few weeks ago Derek talked about senders using homegrown email systems and the pitfalls therein. He has a good list of things companies should think of before deciding a home grown system is right for them.
First, there’s a platform’s subtle complexity. Sending an e-mail is a trivial task and it’s easy to think, “How hard can using an in-house system be?” But managing bulk e-mail delivery can be very hard. Million-to-one chances, things you’re never likely to see using your desktop software, show up every day and must be handled gracefully. Plus the major ISPs all have their own standards for acceptable bulk delivery. They have a multitude of concurrency and session limits, extremely variable response codes, and a plethora of other requirements, some of which contradict Internet standards.
A second challenge is scope. A good system must handle bounces, vacation messages, replies, click-throughs, opens, opt-outs, opt-ins, profile management, and forwards to a friend. It must be able to monitor deliverability and reputation and facilitate corrective action where necessary. And it must be capable of collating these results and presenting them in a meaningful manner in real time.
Third is moving goalposts. Once a system is built, it must be maintained. In this young medium, best practices are constantly evolving and your system must run just to stand still. Adding support for systems such as DomainKeys and DomainKeys Identified Mail (DKIM) takes significant time and expertise. We can expect this trend to continue with such requirements as support for SMS, RSS, behavioral targeting, and dynamic messaging.
Finally, there’s expertise. The same Forrester report indicates that most companies have fewer than four employees managing their e-mail marketing programs. The additional requirements of software, e-mail, and deliverability expertise will inevitably require the team to be substantially larger with a commensurate increase in cost. Staying abreast of current developments and industry best practices can be a full-time job.
I have worked with a clients to improve their homegrown MTA and it is possible. It is also expensive to repair the damage done by a home grown MTA, both in terms of the delivery hit and in terms of development effort.