How AI can be used in deliverability to inbox better

How AI can be used in deliverability to inbox better

Join Chaitanya Chinta from Pepipost to learn how AI can be used in deliverability to improve inbox

About Chaitanya Chinta

Global Head – Email Business at Netcore, Cofounder & Deliverability Guy at Pepipost

Transcript

Typical usage of AI when it comes to email marketing has been mostly around creating right content, right segments and predict the right time to deliver the message etc. While we are also heavily invested in them with AI for personalization, send time optimization, etc. We’ve also looked at harnessing AI in email delivery, at least as per my limited knowledge, very few people ventured into this area. So let me walk you walk you through on what we have done. So usually, when a when a marketing campaign is scheduled, the list consists of users across the spectrum. Some are super active users, some are dormant users, some are best customers, worst customers, pineapples, and bananas, all of them are part of the mix all Imad email ids are not equal. Right. So but when it comes to NTFS, you know, the final delivery layer. They treat them with no difference. The delivery thresholds are usually configured per domain, per mainstream, per customer, etc. But the larger point is that they are not dynamic in nature. So effectively, your dormant users get delivered at the same throughput and priority as your super active users. And the mailbox providers usually don’t appreciate a sender delivering to inactive users at a high speed. world example to establish to establish the case, right? Suppose a brand XYZ does an email campaign to 2 million users. And right after that they Schedule A reengagement campaign to the users who are on the verge of churning out. The first campaign, which is the regular campaign gets delivered when the default setting, and let’s assume delivering at 1 million emails an hour. And delivery has been proper and all of that all of that is done properly. But the second case, the dormant campaign, also gets delivered at 1 million emails per hour, which most likely might not be treated well with some of the mailbox providers. So MDS don’t understand the user behaviour and adjusted throughputs accordingly, but MTS do understand the mailbox provider feedback, they get it in real time, and some advanced MTS can adjust throughputs intelligently, but that too, is 50% of the story. They still don’t get the context of the data and campaign and adjust robots. So they’re not they’re not supposed to either. So the thought was to build an intelligent layer that understands the context of the campaign, and also real time feedback, ISP feedback and adjust the delivery throughputs. For best of the results. Enter Rahman, our AI engine that sits at the centre of all our air driven technologies. So we have lived, so we have leveraged Rahman’s existing tech stack, and built a contextual email engine just before the MTA. Right, this, this engine just sits just before the MTA, and does the following right, it understands the context of the campaign. So when a campaign when a campaign gets scheduled, it breaks down the campaign into hundreds of micro segments, and each of the segment will have its own priority and throughput setting. And the MTA is, is in complete harmony with this to understand and deliver according to these throughputs. So the MTA is now understanding the context of the campaign and delivering it accordingly. The real time feedback from the mailbox providers, which is by default given to MTA is fed in real time to Romans AI engine. And this feedback is also read for any, you know, adjustment in throughputs, and data priorities etc. It is humanly impossible to keep track of 1000s of IPS and customers and adjust delivery accordingly. Hence, it has to be AI. In the earlier example of 2 million regular campaign and 1 million reengagement campaign to the users on the verge of channel. This engine scheduled best customers in the 2 million campaign with high priority and speeds. And as the campaign progresses towards users who are less active or towards the dormant users, the speeds and priorities are reduced. So also the entire dormant campaign that gets so the entire dormant campaign of 1 million users that gets delivered at a slower pace than the original campaign. So this helps in spreading the load. And also since you’re starting your day, with with targeting best of your users The dynamic reputation that builds can help in inboxing better. So, the entire campaign is delivered in a staggered fashion. So, so, after all this after all this hassle, you know, delivering each of the campaign with different throughputs and you know taking the taking the load of AI and everything right. So there has to be a result, right. So we have tested this on 100 plus enterprise

Video Tagged:

Leave a Comment

Hey there 👋 Can we stay in touch?

Once a week we will send you a list of our latest new posts and videos. Never miss a thing!

We don’t spam! Read our privacy policy for more info.