Whether a company is sending purchase confirmations via API, segmented promotion emails with tailored messages, or newsletter promotions, every company aims to keep their customers close. Keeping customers close means keeping them engaged with your company, drawn to your brand, aware of your value, and remembering this value when it’s time to buy.
To do this, a company needs to:
Help customers when it counts most. Use transactional messaging to help them recover a password, get a receipt, or learn when a package has arrived.
Reach them regularly. Find the right rhythm of marketing communications to keep your brand top of mind.
Be relevant to their interests in the moment. Finely segment messages that align with their wants and needs, and prompt them to buy.
Artificial intelligence (AI) enhances the activities above because the technology helps complex systems to adapt, learn, and perform—even as the broader dynamics of human communication continually change.
We’ve been on an exciting journey here at SendGrid as we’ve been building artificial intelligence into our platform.
SendGrid is well positioned within the customer communications space to realize the promise of AI, for 2 reasons:
Super big data: At over 1 billion emails per day, the sheer volume of sending activity and resulting data on our platform makes the latest approaches to machine learning possible. To train robust models, you need the kind of scale that SendGrid has, with over 1 trillion emails sent by our customers so far.
Email brain trust: Among our Delivery, Compliance, Support, Customer Success, and Expert Services Teams, we employ over 160 highly-trained people who focus on optimizing our customers’ email deliverability and engagement. Our Expert Services Team alone has over 50 years combined experience in the email industry.
We’ve given a name to our growing collection of AI capabilities: Adaptive Communication Engine, or ACE. With ACE, our platform adapts to the unique communications needs of individual businesses. Customers large and small benefit from SendGrid ACE, which is included with every pricing plan.
Without realizing it, it’s easy for senders to sometimes take actions that can hurt their sending reputation with ISPs, and thus negatively impact their deliverability.
While we already offer our customers a wealth of resources, 24/7 support, and Expert Services, we wanted to develop technologies to sense and address sending aberrations in real time. The latest capability in ACE, which we call neural protection, helps customers avoid deliverability pitfalls by using the power of neural networks.
Neural protection increases your delivery rate by identifying problematic sending patterns.
The technology uses multi-layered network monitoring to spot the ways you may be inadvertently inviting scrutiny from ISPs. In doing so, neural protection protects your sending reputation. Neural protection continually inspects outgoing messages to identify sub-optimal sending patterns by using machine learning on the complex and ever-changing ways that ISPs monitor and penalize senders.
This continuous monitoring of all senders on the platform works in tandem with technology that facilitates senders by intelligently pacing their sends and scheduling retries for each ISP as needed, with attention to the ISP’s unique rules. This intelligent facilitation ensures maximum inboxing.
As we began our AI exploration, we started by asking ourselves how AI could help us, for example, to:
Identify and block spammers who are attempting to open a SendGrid account
Encourage and guide all customers to use sending best practices
Deliver alerts and insights to customers when an aspect of their email program needs their attention
Optimize programmatic and campaign sending based on users’ likelihood to engage
We have taken important steps on our AI journey; for example:
Innovations from our internal hackathons led to the first application of neural networks in our platform. It was used to classify newly signing-up customers based on over 60 inputs, ranging from the recency of their domain creation, to their chosen type of payment instrument.
Our adoption of ensemble methods such as Random Forest has led to improved prediction. Ensemble methods use multiple learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms alone. Random Forest helps correct for the problem of overfitting: when a statistical model describes random error or noise instead of the underlying relationship.
Our adoption of TensorFlow, an open source Python library, has accelerated our progress through its rapid calculation execution and its robust visualization tools. Tensorflow is now the main system we use to build and train our neural networks to detect and decipher patterns and correlations. First released by the Google Brain team in 2015, TensorFlow has been used by computer scientists globally in everything from speech recognition in Google Translate to the identification of signs of Parkinson’s disease in medical scans.
And we’re still early in our journey―we foresee our Adaptive Communication Engine evolving to serve a variety of needs in the realm of customer communications.
With SendGrid ACE, we’re harnessing neural networks and machine learning to help you keep your customers close. ACE enables you to send with even greater confidence, supported by a platform that simultaneously understands both the real-time global dynamics of digital messaging, and your own unique sending needs.
