Yesterday, Amazon launched Simple Email Service, yet another addition to their excellent cloud-based infrastructure. Since I’m CTO of an email service provider, a lot of people asked me about it.A bit of background: most know Amazon as the world’s biggest online store, but years of running one of the largest websites ever has allowed them to create massive, scalable infrastructure that is available for anyone to use. Services like S3 (Simple Storage Service) and EC2 (Elastic Computing Cloud) power a ton of websites.
Simple Email Service works pretty much the same way as the rest of the Amazon infrastructure: use a simple API, pay only for what you use (in this case, how much email you send), grow as you scale. And like most of the rest of the Amazon services, the rates are really, really cheap: only 10 cents per thousand emails sent. That’s maybe 20% the cost of similar existing services.
I love Amazon’s services — most of Sailthru’s infrastructure, including our web and database servers, run on Amazon’s EC2 cloud platform. It has its drawbacks (I/O performance is poor, machines can disappear on you randomly, and you’re on your own for managing them), but so does real hardware. By and large it’s awesome for a fast-growing startup, since we can deploy new servers in minutes without up-front costs.
But we have always used our own dedicated machines for actually sending email for a simple reason: you do not get good email delivery from the cloud. Low-cost shared IP space ends up sending a lot of low-quality email (non-opted in lists and borderline-spam), and eventually the entire service provider is flagged as a spammer.
So the email industry will be watching one thing closely: deliverability. Will emails sent by Amazon get flagged as spam, or will they get through to the inbox?
Amazon is doing plenty of smart things for deliverability — gradually ramping up the allowable send rate for a user, applying outgoing spam filters, and making use of the various feedback loops that Internet Service Providers and the email industry have evolved over the years. And I’m sure they’ve reached out to the major providers to make sure the service launch is a success.
But it remains to be seen over time whether SES delivery will be excellent (up to par with top providers), decent, or poor. If it’s poor, it will be only useful for experiments, would-be spammers, and very low-priority email. If it turns out to be excellent, a lot of email providers (including my own) may start using them and lower our infrastructure costs, just as we’ve lowered our costs by using so much else on Amazon’s platform.
I suspect the reality will lie in the middle — deliverability will be decent but not great. If so, SES poses a major threat to existing cloud-based SMTP services that just provide simple sending of emails. These services will have to innovate to compete with Amazon’s dirt-cheap pricing. If past history is a guide, Amazon will likely leave them some room to do so, as all of their other cloud services have focused on infrastructure rather than extensive functionality.
For email service providers that operate at a higher level, managing mass email campaigns, providing reporting and analytics, dynamic content assembly, behavioural targeting, and so on, SES isn’t a threat, since it provides none of these features. Rather, it will be a useful tool that we will evaluate and could quite possibly use.
Ian White is cofounder and CTO of Sailthru, an email service provider startup. Previously, he was lead engineer at Business Insider.