February 3, 2006
Why Email Stamps Are a Bad Idea
(also posted on the Return Path blog)
Rich Gingras, CEO of Goodmail is an incredibly smart and stand-up professional. I’ve always liked him personally and had a tremendous amount of respect for him. However, the introduction of the email stamp model by Goodmail is a radical departure from the current email ecosystem, and while I’m all for change and believe the spam problem is still real, I don’t think stamps are the answer. Rich has laid out some of his arguments here in the DMNews blog, so I’ll respond to those arguments as well as add some others in this posting. I will also comment on the DMNews blog site itself, but this posting will be more comprehensive and will include everything that’s in the other posting.
It seems that Goodmail’s main argument in favor of stamps is that whitelists don’t work. While he clearly does understand ISPs (he used to work at one), he doesn’t seem to understand the world of publishers and marketers. His solution is fundamentally hostile to the way they do business. I’m happy to have a constructive debate with him about the relative merits of different approaches to solving the false positive problem for mailers and then let the market be the ultimate judge, as it should be.
First, whitelists are in fact working. I know — Return Path runs one called Bonded Sender. We have documented several places that Bonded Senders have a 21% lift on their inbox delivery rates over non-Bonded Senders. It’s hard to see how that translates into “bad for senders” as Rich asserts. When the average inbox deliverability rate is in the 70s, and a whitelist — or, by the way, organic improvements to reputation — can move the needle up to the 90s, isn’t that good?
Second, why, as Goodmail asserts, should marketers pay ISPs for spam-fighting costs? Consumers pay for the email boxes with dollars (at AOL) or with ads (at Google/Yahoo/Hotmail). Good marketers have permission to mail their customers. Why should they have to pay the freight to keep the bad guys away? And for that matter, why is the cost “necessary?” What about those who can’t afford it? We’ve always allowed non-profits and educational institutions to use Bonded Sender at no cost. But beyond that, one thing that’s really problematic for mailers about the Goodmail stamp model is that different for-profit mailers have radically different costs and values per email they send.
For example, maybe a retailer generates an average of $0.10 per email based on sales and proit. So the economics of a $0.003 Goodmail stamp would work. However, they’re only paying $0.001 to deliver that email, and now Goodmail is asserting that they “only” need to pay $0.003 for the stamp. But what about publishers who only generate a token amount per individual email to someone who receives a daily newsletter based on serving a single ad banner? What’s their value per email? Probably closer to $0.005 at most. Stamps sound like they’re going to cost $0.003. It’s hard to see how that model will work for content delivery — and content delivery is one of the best and highest uses of permission-based email.
Next, Rich’s assertion that IP-based whitelists are bad for ISPs and consumers because IP-based solutions have inherent technology flaws that allow senders to behave badly doesn’t make sense. A cryptographically based solution is certainly more sophisticated technology — I’ve never doubted that.
In terms of the practical application, though, I’m not sure there’s a huge difference. Either type of system (IP or crypto) can be breached, either one is trackable, and either one can shut a mailer out of the system immediately — the only difference is that one form of breach would be trackable at the individual email level and the other would only be trackable in terms of the pipeline or IP. I’m not sure either one is more likely to be breached than the other — a malicious or errant spammy email can either be digitally signed or not, and an IP address can’t be hijacked or spoofed much like a digital signature can’t be spoofed.
It’s a little bit like saying your house in the suburbs is more secure with a moat and barbed wire fence around it than with locks on the doors and an alarm system. It’s an accurate statement, but who cares?
I’m not saying that Return Path will never consider cryptographic-based solutions. We absolutely will consider them, and there are some things around Domain Keys (DKIM) that are particularly appealing as a broad-based standard. But the notion that ONLY a cryptographic solution works is silly, and the development of a proprietary technology for authentication and crypotgraphy when the rest of the world is trying desparately to standardize around open source solutions like DKIM is an understandable business strategy, but disappointing to everyone else who is trying to cooperate on standards for the good of the industry. I won’t even get into the costs and time and difficulty that mailers and ISPs alike will have to incur to implement the Goodmail stamp system, which are real. Now mailers are being told they need to implement Sender ID or SPF as an IP-based authentication protocol — and DKIM as a crypto-based protocol — and also Goodmail as a different, competing crypto-based protocol. Oy vey!
Email stamps also do feel like they put the world on a slippery slope towards paid spam — towards saying that money matters more than reputation. I’m very pleased to hear Goodmail clarify in the last couple of days that they are now considering implementing reputation standards around who qualifies for certified mail as well, since that wasn’t their original model. That bodes well for their program and certainly removes the appearance of being a paid spam model. However, I have heard some of the proposed standards that Goodmail is planning on using in industry groups, and the standards seem to be much looser than AOL’s current standards, which, if true, is incredibly disappointing to say the least.
Jupiter analyst David Daniels also makes a good point, which is that stamps do cost money, and money on the line will force mailers to be more cautious about “overmailing” their consumers. But that brings me to my final point about organic deliverability. The mailers who have the best reputations get delivered through most filtering systems. Reputations are based largely on consumer complaints and unknown user rates. So the mailers who do the best job of keeping their lists clean (not overmailing) and only sending out relevant, requested mail (not overmailing) are the ones that will naturally rise to the top in the world of organic deliverability. The stamp model can claim one more forcing function here, but it’s only an incremental step beyond the forcing function of “fear of being filtered” and not worth the difficulty of adopting it, or the costs, or the risks associated with it.
Rich, I hope to continue to dialog with you, and as noted in my prior posting, I think separating the issues here is healthy.