Machine Intelligence transforming consumer interaction

Cut through the noise of massive data sets and meaningful insights and you reach subtle patterns in real-time enable greater personalization by taking the guesswork out of content curation. Automated processes save you time, lift engagement, reduce churn, and substantially increase the lifetime value of your customers.

Machine Intelligence, technology transforming consumers marketer interaction, like any revolutionary technology, will feel scary at first and delightful once we get used to it; ultimately, it will simply become the norm. Want some evidence? Let’s go back a few years and look at the last big revolution in marketing technology: Marketing Automation. These now-commonplace platforms enabled us to program triggers and schedules that told computers when to send which kinds of emails and push messages at any scale.

Getting marketers to loosen their tight grip on the reins and embrace that level of mechanization was a hard sell, but ultimately, the performance gains sold themselves. Machine Intelligence is a lot like Marketing Automation, except, in this case, the computers are telling themselves what to send and when, and the performance gains are even more remarkable.


The What and Why of Machine Learning

“Think of Machine Intelligence much as you would think of raising a child: the goal is to create something that can make smart decisions when you’re not there.

Practical examples of machine intelligence are rarely as dramatic as self-driving cars or Jeopardy-winning robots. More often, they take the form of Netflix movie suggestions and Amazon product recommendations. Machine Intelligence (synonymous with Artificial Intelligence) is defined as “the science and engineering of making intelligent machines” by John McCarthy, who coined the term in 1955. This definition isn’t particularly helpful since it’s hard to define MI or AI without defining intelligence, which is an immensely complex and subjective topic.

For the purposes of this paper, think of Machine Intelligence much as you would think of raising a child: the goal is to create something that can make smart decisions when you’re not there. And Machine Learning is simply the process by which a machine develops that ability. Like human learning, it involves developing a number of basic skills and then combining and refining them to gradually multiply the power of that intelligence.

While we don’t always exercise it, humans have an immense and unique ability to make those “smart decisions.” We can weigh many different and nuanced variables to determine the right course of action for a given scenario—such as what subject line and what content should go into a promotional email. But we can only see and do so many things at a time.

Machine learning pairs the processing power of computers with powerful, unbiased algorithms. It identifies patterns we cannot see ourselves and applies multiple insights simultaneously to solve problems at a scale and speed that we—humans— cannot physically match.

By approximating human intelligence in machines, we’re seeking a way to “scale up” our own smarts and even improve our accuracy by bringing more variables into the mix.



To further stir up the alphabet soup of buzzwords and acronyms (MI/AI/ML) that have permeated this field, we now add Predictive Analytics (PA) to the mix. Machine Learning encompasses a multitude of topics or “ingredients,” such as supervised learning, unsupervised learning, clustering, attribute selection, and more.

Predictive Analytics is one such ingredient of Machine Learning that uses past or current information to predict the likelihood of a particular event taking place in the future.

The classic example of PA is your credit score, which indicates the statistical probability that you will pay your credit card bill in a complete and timely fashion. This calculation relies on complex statistical models that define the relationship between specific data points (e.g. your balance/credit limit ratio) and specific actions (e.g. a late payment).

An example from the world of marketing would be your email churn rate. Both models operate off of the same business goal: reduce churn. While Predictive Analytics can tell you how likely a customer is to churn, Machine Learning can look at that probability, compare it against other unrelated data, and tell you what specific actions are required to reduce the likelihood of churn. Another distinction is in how those statistical models are updated. In Predictive Analytics, models must be manually updated, whereas Machine Learning can automatically and instantly calibrate those models based on each successive outcome.



Imagine that your goal is to sell clothing to a 34-year-old woman who lives in Austin, reads the New Yorker, and recently bought a car seat. You could probably do a decent job of writing an email that’s relevant to her, but you’d be making a lot of assumptions that might make the email feel off-target or impersonal. Now imagine you’re asked to promote that same array of clothing products to your friend Karen. You know all the same things about Karen, but you also know that she always loves to read and share something funny on her lunch break. You know that she bought that car seat as a gift for her brother.

That is the difference between segmentation and 1:1 personalization. 1:1 personalization is sending an email to Karen, not just people like Karen – 34-year-old women who live in Austin. But even a close friend might not know that Karen’s on-site behaviour fits a pattern that indicates she is likely to consider buying something for herself soon, or that her search history indicates that she’s a budget-conscious shopper with less affinity for brand names, or that her interest in Antarctica was likely an aberration that could be attributed to a pop culture phenomenon and is unlikely to influence her behaviour in the long term.

Now imagine Karen is one of a million people who you and your 5-person marketing team want to deliver content to, and you’ll begin to see the potential for Machine Learning. It’s the power to send relevant, timely, and engaging content to every customer, every time, at any scale. It gives you the equivalent of a million email marketers all crafting individual emails for every one of your customers or subscribers. Welcome to SwiftERM.



As marketers, particularly email marketers, we create a lot of noise. This is the result of 2 dangerous assumptions that have been allowed to perpetuate:

1. Dismal engagement rates (low single digits) are to be expected.

2. And the “That’s ok because they don’t cost much” mentality

While it’s true that we can’t expect every email to drive conversion (in any marketing medium, even the most relevant content only provides a nudge in the right direction), we need to expect better performance from our emails, in part because there is a potentially high cost to sending irrelevant emails.

As you compete for the increasingly fractured and overloaded consumer mind space, each time a message is ignored, deleted, unread – or worse, read and deemed irrelevant – it erodes your brand’s equity and makes subsequent messages that much easier to dismiss.



Cut through the noise with individually relevant messaging delivered when consumers are ready to pay attention.

It’s time to move beyond the “mail merge” and “batch send” methodology. Though the success of every email campaign starts with earning those “opening clicks,” the clicks mean next to nothing if your subscribers don’t engage further. (Plus, when it comes to tips and tricks for writing a juicy subject line, everyone has access to the same cheatsheets.) You need to offer content other people can’t offer or don’t know enough to offer, as well as a related, compelling subject line.

The mythical “Segment of One” has become a reality. Through Machine Learning technology, we can send individually tailored emails to every person, every time. The goal is to send an email that someone who knows the recipient would send because people almost always open emails from people they know because they know the content would be relevant and personalized to them only.

Machine Intelligence enables you to increase engagement rates by:

Making subject lines dynamic based on what’s likely to engage the individual recipient and work well with the personalized content within the email. (And using natural language parsing to ensure that those dynamic subject lines don’t sound awkward or computer-generated.) Optimizing delivery time based on when that person is likely to open and/ or engage with the content within the email, known as Delivery Time Optimization. (Read more about it here). This also helps overcome “position bias,” which makes us more likely to open emails at the top of our inboxes.



The content included in emails traditionally requires editorial curation, and that means making hard choices. Even if you’re just trying to drop in the latest or top trending content, you have to decide what content you can actually build out, knowing full well that you’re just making an educated guess as to what will be relevant to an individual recipient.



Auomatically curate and populate content for EACH user

Machine Intelligence takes the guesswork out of content curation by automatically surfacing and populating the most relevant content for each recipient, based on the conversion goals you define.

Machine Intelligence gives you the power to make your communications more “human” at any scale.

Automatically customize emails with content that is statistically likely to engage each user based on that individual’s behaviour.

Make the most of new and time-sensitive content like a daily deal or coupons, and instantly tap into cultural trends and viral opportunities.

Match users to content based on a deep semantic understanding of the content itself, as well as its real-time popularity among all your other customers.

BONUS: As it learns, you learn. Machine Intelligence will surface insights that allow you to see WHY certain content is relevant and identify broader trends among your content and your audience.



Marketers are perpetually behind the 8ball, trying to crunch data, identify segments, build content, send, test, optimize… all in order to hit a target that might have already moved.

The mandate is almost always to do more with less. When you think about it, it’s almost a wonder that we ever tried to accomplish all that without help.

Automate everything by making your automation smarter

Machine Intelligence can give you back the time you never knew you had by tirelessly building, targeting, and delivering emails on a level no human can accomplish alone. We’re not talking about hard-coded rules-based automation that has to be actively managed by a person.

We’re talking about a system that can learn, adapt, and respond based on your business rules but with minimal human intervention. It’s like a zillion-strong army of researchers, testers, analysts, and writers that never eats or sleeps. Gone are the days of manually segmenting lists. Think of it simply as one master list that the machine slices and re-slices in real-time, into segments of one person. Deploying emails using machine intelligence is like having a constantly running A/B test with a constantly vigilant team of analysts to instantly tweak and optimize the algorithms based on performance data.



“For Email Marketers, and marketers as a whole, Machine Intelligence will give you back precious time
to focus on your core strategy and help you deliver engagement and retention stats that will make you a superhero.”

Whether you’re measuring open rate, CTR, Click-to-Open-Rate, or impact-on-site, success ultimately comes down to money earned vs. money spent. We can talk about relevant content and customer experience and brand affinity all day (and we should, because they are all important), but ultimately, all marketing needs to perform in a measurable way.

We all know we gripe about the yardstick the most when we’re not measuring up. Email marketers feel this pain acutely because of the relatively low barrier for email marketing has often led to a skewed investment/ expectation ratio.


Incorporating Machine Intelligence into your email marketing will deliver significant performance lifts across almost any campaign or metric you track. Once you start sending individually customized messages, the investment can be quickly recouped in performance gains and increased efficiency.

Here’s the better news:

It keeps learning. Your Machine Learning platform will be constantly testing and refining its statistical models and algorithms against new and existing data. It can even add an element of serendipitous discovery to generate new insights about individual users and content once gains begin to plateau.

We’ve talked a lot about the value of automation, but Machine Intelligence platforms are not sentient beings. They still require you to work, and your hands are still on the levers. AutoCAD programs didn’t make architects obsolete; it just freed them up to explore new realms.

For Email Marketers and marketers as a whole, Machine Intelligence will give you back precious time to focus on your core strategy and help you deliver engagement and retention stats that will make you a superhero. So, fight to get your organization on that train while ML is a competitive advantage instead of just table stakes.



Technology continues to change the relationship between customers and brands. As marketers, we must adapt, as we always have. But the core truths about what makes an email campaign (or a direct mail campaign, or even a town crier for that matter) win or lose have remained the same: you need to be more relevant, more (personally) engaging, and more timely than the other guy.

The difference is that to do it at the scale and pace that today’s markets and consumers demand, you need the kind of help that only a machine can provide. But don’t get all teary-eyed for the days of marketing yore. This isn’t some “the machines are coming for your jobs” proclamation. The machines are coming, but for their jobs: the jobs that involve doing repetitive tasks on a massive scale. What Machine Intelligence really means for marketers is that machines are finally good enough to operate without so much oversight from us.

We can stop fighting the losing battle to keep up and get back to focusing on leading, innovating, and creating compelling content… the things that we’ve always done well and that machines can’t touch.

And who’s going to complain if we can also take credit for massive performance gains in our email marketing? Certainly not the Intelligent Machines responsible for them.



What Machine Intelligence really means for marketers is that machines are nearly good enough to operate without so much oversight from us – they have become an extension of us.

Machine Intelligence gives you back time by tirelessly working and building, targeting, and delivering emails on a level no human can accomplish alone. Machine Intelligence makes your emails cut through the noise by keeping your messaging relevant and timely for each user at an unlimited scale. Machine Intelligence takes the guesswork out of content curation by automatically customizing emails with the most relevant content for each recipient.

Machine Intelligence delivers greater performance lifts across your email campaigns through a constant learning process that continually learns more about each user and the types of content they have the most affinity towards.

We hope you enjoyed this article, intended to help improve our client’s profitability. It reflects the care SwiftERM offer. If you haven’t already done so, then please enjoy a FREE month’s trial and let us know what you think. Register, call us on 0207 998 3901, book a call with us or Zoom ID 964 515 7464 


Other articles of interest below:
(Index to all articles here)

Email Marketing Made Simple: A step by step guide with examples
Where Predictive Analytics Is Having the Biggest Impact
“Millennials” are meaningless – and demographics are useless
DMOZ Directory Web Directory