Two Worlds of Email Deliverability

There are really two "worlds," or ways to approach, email deliverability.

World 1: Working with the mailbox providers

Work to send email that is wanted and expected by the recipients — the type of email that mailbox providers want to put into the Inbox.

In this world, the foundation of good deliverability is sending email that is:

• Wanted – you offer something of value to the recipients.
• Expected – recipients remember subscribing or remember their business relationship with your brand.

Mailbox providers don't have a crystal ball telling them whether recipients wanted the email, but they have something better: statistical information on how tens of thousands of their users interacted with it — most importantly the complaint rate and engagement rates. They know what percentage of people click "this is spam," what percentage read the message, what percentage delete without reading, and how long people read it. All of this feeds their algorithms, which infer whether your email is wanted and expected.

In addition to complaint and engagement signals, providers also examine content (some keywords or language contribute to spam filtering) and technical sending details (connection counts, email authentication, warm-up schedules, etc.).

In this world, a sender does not try to "game" the system in the technical details — the email is sent in a straightforward way. The providers' goal is to deliver email their users want and expect; if you craft your program around wanted/expected email, everything else typically falls in line. Delivery problems still come up, but they can typically be solved.

In this world, open rates of 5% to 20% or higher are common. The goal is to reach the Inbox.

World 2: Working against the mailbox providers

Using technical tricks to try to fool spam filters into accepting or delivering unwanted and unexpected email.

This is a world that is easy to end up in. If your email generates high complaints or low engagement and providers are filtering it, there are two ways to react:

• (A) Examine the email program, figure out what is causing complaints or spam-trap hits, and address the root cause to make the email more wanted and expected. This is operating in World 1.
• (B) Avoid addressing the root cause and focus exclusively on technical tricks — e.g., using lots of IP addresses to avoid scrutiny, changing domain names, or putting the entire email content in an image to avoid content filters. This is the path into World 2.

The essence of World 2 is trying to game the system. There is a lot of bad deliverability advice out there, so many people go down this path with good intentions.

The problem: mailbox providers have dedicated teams of very smart people working to identify the technical tricks used to avoid their filters, and to block them. All gaming methods eventually stop working. The tricks listed above are well-known, and providers have specific countermeasures that detect and penalize them.

In this world, open ratios like 0.1% are common, and the common "solution" to a low open ratio is "just send more email." Often the goal here is merely to reach the Bulk/Spam folder instead of being rejected outright.

What makes email wanted/expected

1. Permission is essential. People need to invite you to send them email. Otherwise your email isn't wanted, because the recipient never asked for it.
2. What you send should match what people are expecting. If someone signs up for a "daily deal" and you send a daily advertisement with no deal, that's not what they wanted or expected.
3. Volume should be reasonable and expected. If you never state a frequency and then send five messages a day, you're sending more than people expect (and probably want). Set the subscriber's frequency expectation during sign-up.
4. Permission can't be hidden. Permission buried in unread terms and conditions produces unexpected email. Similarly, a pre-checked checkbox leads to unexpected email, because not everyone sees the checkbox.
5. The email must be identifiable. If a user subscribes to a brand and the email doesn't display that brand, they will likely treat mail from an unrecognized sender as unwanted and unexpected — even with a great subscription process.
6. Clean inactive subscribers. Stop sending to subscribers who haven't opened or clicked after a period of time. A subscriber who has received three emails per week for six months without opening any has essentially "emotionally unsubscribed." Removing them makes your volume, on the whole, more wanted, because your list contains a higher proportion of subscribers who read and want your email.

The mailbox providers' stated objective is to filter and block email that is unwanted and unexpected.

Examples of gaming techniques (which don't really work)

1. Using too many IP addresses (IP snowshoeing). Spreading volume over many IPs so each flies "under the radar." Providers detect this and penalize senders whose IP count is too high relative to volume. Each IP address should send no less than roughly 20,000 messages per day.
2. Using too many domain names (domain snowshoeing). Spammers use lots of "disposable" domain names to escape a bad reputation and hide their identity.

Other techniques used by hard-core spammers include compromising other senders' ESP accounts, DKIM replay attacks, sending from botnets, and abusing open relays — some of which are criminal.

In the context of reaching the Inbox:

• IP and domain snowshoeing alone can be enough to get your email demoted to the Spam folder at many providers — filters are specifically designed to detect it.
• Snowshoeing is sometimes effective only at reaching the Spam folder: it can let some email land in Spam that would otherwise be outright rejected.

The two dimensions

The difference between the two worlds boils down to two continuous dimensions (not binary):

• wanted/expected ↔ unwanted/unexpected
• not gaming ↔ gaming the system

The goal is to move toward more wanted/expected and not gaming.

The trap: if you're operating in World 2, improving only one dimension does not improve delivery:

• Moving only toward wanted/expected (while keeping gaming techniques) → you still get filtered because of the gaming techniques.
• Moving only toward not-gaming (while still sending unwanted/unexpected email) → you get filtered because the email is unwanted. Sometimes this move causes email that was reaching the Spam folder to start being rejected entirely.

You must improve both dimensions together.

Less is more: the counter-intuitive economics

Moving up the wanted/expected axis often means sending less email — and it's possible to make more money by sending less email, because it starts reaching the Inbox. A simplified, hypothetical illustration:

• Open rate for email delivered to the Spam folder: 0.3% (about 95% of recipients never look in Spam).
• Open rate for email delivered to the Inbox: 10% — 30× higher.
• Assume each open has the same chance of producing the desired (revenue-generating) action.
• Suppose improving wanted/expected-ness requires changes that cut send volume 5× but get the mail into the Inbox.

Result: volume ÷ 5, revenue-per-email × 30 → total revenue up 6×.

Making the changes

Improving your position on the wanted/expected axis usually requires a thorough review of your email program, ideally through a data-driven approach: make changes and monitor their effect on complaint rates and engagement.

Deliverability enhancement is an iterative process — as sending history accumulates, the data suggests further improvements. Ultimately it is the mailbox providers who decide whether mail reaches the Inbox, the Spam folder, or is rejected; these practices are about sending email they want to inbox.

Related

• Foundations of Email Deliverability
• Basic IP Allocation
• Advanced IP Segmentation and Allocation