Hive’s CTO on Smart Email Throttling: Sending Intelligently for Better Deliverability

At Hive, we send millions of emails every day for our customers. Our goal is to not only make sure they’re delivered, but to help our customers achieve good inbox placement so these emails are actually seen by subscribers.

We often see email marketers wanting to send campaigns to their recipients as soon as possible, which inevitably leads to spikes in the overall volume of emails being sent.

Unfortunately these spikes in volume are often flagged by spam filters, making it look like a trusted server has been compromised. ISPs read this behavior as a sign of bad list hygiene, assuming the sender is sending a large amount of unsolicited email (i.e. spam).

When ISPs see a spike, they delay the delivery of that message to other recipients, staggering its release. This is frustrating for email marketers - as a sender, you want your email to send as quickly as possible, and you shouldn’t have to keep track of ISPs’ limits. That’s where smart email throttling comes into play.

The theory behind smart email throttling

ISPs look at how recipients are initially engaging with your email to determine how they should continue placing it (to inbox or spam). They want to see primarily positive engagement like opens and clicks, since their goal is to deliver the emails recipients actually want to receive to their inboxes.

When an email is sent to all recipients at once, ISPs have to use limited information to determine inbox placement. So if they happen to see a few bounces right away, or if they don’t think the engagement at that point is good enough, they are more likely to determine that the email is “bad” and send it to spam for other recipients. In other words, sending your email to every recipient at once means ISPs can’t use your campaign’s positive engagement to improve its inbox placement.

Related: How to Build a Strong Sender Reputation to Improve Email Deliverability

You may have heard the term throttling used to refer to the intentional slowing or speeding of internet - email is the same. The theory is that sending a predictable, consistently-increasing volume of email over time will avoid spikes and lower the risk of ISPs delaying your email send.

Factor in sending to the most engaged subscribers first to signal positive engagement to ISPs early on, and you increase the potential of improving overall inbox placement.

“The theory is that sending a predictable, consistently-increasing volume of email over time will avoid spikes and potentially improve deliverability.”

The idea is to automatically send the first few messages of a campaign to your most engaged subscribers within the first minute, signalling to ISPs that this email will be positively received and should not be flagged as spam. Sending these emails first with a brief pause before the next round gives ISPs time to evaluate, react, and hopefully place your message into other recipient’s inboxes.

With smart email throttling, this method of sending continues minute over minute, sending the first batches of emails to the most engaged subscribers and gradually sending to less engaged subscribers until the campaign has been sent to all recipients.

How ESPs approach email throttling may be key

The general concept behind throttling is nothing new. Sending email in small increments to subscribers is a well established way to avoid delays, blacklists, and poor deliverability. Almost all ESPs that send on a large scale, like Hive, need to implement this strategy in order to protect their platform’s IPs and sending reputation, and ensure a reliable service.

But combining throttling with a strategic (rather than random) sending pattern to target recipients based on engagement is a relatively new concept, especially as a system that could potentially be built into an ESP platform.

Many of the industry’s top minds have spoken about how sending to your most engaged subscribers first is best practice for email deliverability, but this strategy is usually left to email marketers and done manually. In fact, some ESPs actually make it difficult to avoid sending to unengaged subscribers.

When email marketers spend so much time optimizing their subscriber lists for deliverability, it can be frustrating to see a campaign penalized when it’s sent all at once, AKA sent to some unengaged subscribers first.

How sending email gradually may actually mean sending emails faster

Ultimately, the concept behind throttling is an optimization problem: how do you send email slowly enough for ISPs to give the best possible inbox placement, but also as quickly as possible so the campaign reaches all recipients sooner rather than later?

Funnily enough, sending in this way can actually help your emails hit inboxes faster than if they are sent out in one massive push. As mentioned earlier, when ISPs flag these spikes, they often delay sending on the rest of the campaign. This means emails are sometimes sent over a matter of hours, on a schedule you can no longer control, so you’re left unable to be sure of when or if subscribers will actually see your campaign.

By choosing to send in smaller increments over a controlled period of time, your overall email send time may be condensed quite significantly (potentially less than if you sent in a massive push), and you’ll end up benefiting your sender reputation and general deliverability.

So, does smart email throttling work?

While throttling has been a fundamental piece of our sending strategy at Hive, we've just started experimenting with the concept of smart email throttling. That being said, there are some early indicators that it's working.

Since deploying the system, we haven’t seen campaign or system-level delays by any ISP, even when sending large campaigns to users with very little historical engagement data.

We’ve also seen engagement rates stay consistent across campaigns. This means that engaged recipients stay engaged, and aren’t negatively affected by less-engaged recipients on a campaign level.

In some cases, we’ve seen at-risk or unengaged recipients actually engage more positively over time. This supports the theory that ISPs are more likely to give better placement to messages sent to unengaged recipients if the emails sent to other recipients have been well-received first.

Wrap up

Moving forward, we’ll be iterating on pieces of our throttling strategy to see what moves the needle further. We’ll look at improving engagement predictions, fine-tuning ISP specifics, and allowing customers to optimize for deliverability vs. delivery speed on a campaign-by-campaign basis.

Interested in learning more about how you can improve your email deliverability? We’ve written a comprehensive guide of our advice and tips for better inbox placement. Want to learn how Hive’s baked-in deliverability features can help your current email sending strategy? Send us an email at to chat!