Marketing Automation 2.0, Part II: Adding A.I. to Automation

In Part I of this post, I ran through some of the issues with “marketing automation” that make it a misnomer; there’s precious little automation at work within present-day MA platforms.

We’re taught to be skeptical about anything that seems to offer a panacea. But there’s an elegant solution to the problems besetting Marketing Automation 1.0. It’s not some pundit’s hypothesis, either, but an answer that’s being implemented in the here-and-now.

Artificial intelligence is the cure, making one-to-one personalization and account-based marketing (ABM) into actionable realities.

Since all deals are account-based, it’s corollary that future B2B marketing automation will be account-based, demanding exactly the customized engagement A.I. provides.

How does that work in practice?

• First, an AI/deep learning platform gathers customer and audience data at a very large scale.  Mariana’s own neural network imports customer purchase data, then codifies individuals by drawing on up to 50,000 smart data points from internal sources (like a client enterprise’s existing marketing automation platform), external public data and our own proprietary data sources.
• Then hierarchical and pyramidal systems categorize, sort, and prioritize all these data sets simultaneously, yielding accurate, brain-like associations.
• As these disparate data sets are aggregated and cleansed, the AI is able to generate rich profiles of each individual user.
• That results in far better lead generation for sales teams and outbound marketing.
• A.I. allows you to flexibly deploy ABM at scale across all accounts, by removing the cost and manpower limitations of manually-built workflows or other staff-dependent processes. There’s no need to deny any account the benefits of ABM; with deep learning, every account and every deal is viable for 1-to-1 engagement.

It also drives personalized digital experiences for prospects and customers, as AI provides real-time insights and recommendations marketers can use to serve up truly customized content.

Here’s an example: One Mariana client gave both us and a traditional lead-gen service the same list of 5,000 companies, then asked each of us to identify the individuals at those firms in charge of SEM:

• The other service returned 55K names, but only 15% of them, (about 8K in all), aligned with the client’s criteria.

Our deep learning engine recommended 20K names, but its list was 81% accurate, equalling 16K quality leads.

In other words, AI was able to extract  twice as many qualified leads from the same data.

The power to make intelligent associations from a huge number of data points is how AI discerns patterns and behaviors that identify targets, rather than relying on titles and roles.

In fact, old-school lead-gen can be stymied by the fact that many of the people a marketer wants to reach don’t go by the title being used in its persona-building. For example, we found that only 20% of people who are fulfilling the functions of a data architect actually used the title “data architect.” The rest — 80%! –went by titles like software engineer, senior software engineer, development engineer or others.

Solving the issues of 1.0

Deep learning help solve all five of the issues we touched on in our previous post, where the last generation of marketing automation solutions came up short:

• Personas, as marketers have known them, are replaced by real-time personas instantly generated by the AI as it tracks and analyzes internal and external data and the top-of-funnel (or even above-the-funnel) behaviors of prospects and customers. Since they’re built on real, observable data, AI-created personas are far more actionable: AI quickly identifies the best leads, then recommends the right touch points and messaging for each target, not based on generalized archetyping or supposition but on actual data about their proven preferences and actions. Plus, the more engagement an AI has with each target, the more precise its personalization becomes.
• Infrastructure is streamlined, because those hard-coded workflows and all that expensive elbow-grease to maintain them vanish. A deep learning AI has the ability to learn and make ongoing associations without supervision, learning and improving as you feed it oceanic amounts of data, making the entire platform thinner and smarter. It’ll play nice with third-party applications, too, keeping the entire marketing automation stack remarkably lightweight.
• Lead generation becomes far more targeted and effective, thanks to the greatly-improved quality of leads delivered at lower cost (and with a minimum of headaches) by AI. 
• Speed-to-response is shortened almost exponentially, because an AI-enhanced marketing automation system will initiate personalized engagement in real time, closing any “opportunity gap” as it adjusts to each person’s context and needs. Better still, it will do so across the entirety of an omnichannel marketer’s digital continuum (email, social, web, mobile messaging, apps and more) with seamless consistency.
• Big Data (or less-than-big data) sees its most optimal use within an AI platform, which can collect and connect data from all sources, so adding it from even non-traditional and third-party datapoints is elementary. Because AI is able to work with this wider variety of data, it can deliver a richer, more nuanced profile of your target. This ability to assimilate a wide range of data and learn from encounters with customers actually decouples marketing automation from any need for an existing Big Data architecture. So, reaping the rewards of AI-enhanced marketing automation won’t necessarily entail Big Data implementation headaches.

Other advantages?

• Cost benefits happen across the board. Not only has pricey marketing automation programming and administration been removed, but the overall cost per lead declines, too.
• ROI improves, not just because of cost containment but because quality leads drive more (and richer) conversions. Plus, ROI happens faster because those leads are being supplied from Day One.
• Ease of use is a big bonus; as one Mariana client put it, using AI is a case of “set it and forget it” while it does its job, freeing up marketers’ time and energy.
• Content optimization results from having a better understanding of your target, automatically identifying high-performance keywords they’ll respond to, knowing what content gets shared the most, keeping automated inventory of all content and intelligently recycling older content rather than creating new material.

Tomorrow is being automated…today

Companies of all types are already implementing machine learning in areas beyond marketing. “Adopting AI” is a hot meme in martech, the same way “adopting marketing automation” was over the last few years.

One quote that underscores the impact of AI throughout the enterprise comes from Gartner: “By 2018, 50% of the fastest-growing companies will have fewer employees than (they do) instances of smart machines.”

• When Fuze sought to augment its marketing automation with deep learning, it saw immediate results by integrating a full-funnel SaaS AI solution that drive a 10X bump in conversion rates, with an overall 200-300% lift in personalized email Click Through Rate (CTR). Other marketers have seen their cost-per-click in social, search and display advertising cut in half, and a 10-50X improvement in cold email campaigns. (Full disclosure? Fuze is a Mariana customer).
• Facebook recently debuted DeepText, a deep learning-based tool capable of reading and comprehending a thousand posts per second in over 20 languages, enabling it to do things like scan Messenger for words or phrases that can trigger responses: if you type that you need a taxi, DeepText can help send you prompts or links to call one. DeepText will soon monitor comments on Facebook to surface high-quality ones while removing objectionable ones, as well as pointing people toward relevant content. Another next step? Building new deep learning architectures capable of understanding text and visual content together.
• Coca-Cola is driving hard into the realm of AI, according to the company’s Digital Marketing Strategist, Yuri Pereira: “New social platforms are being created every year and, as a result, brands have to adapt to those platforms and allocate new resources to optimize content…What would be very interesting is if, in the future, we are able to create systems that not only determine the right message given the target group but also determine what kind of clusters of people are effectively replying or engaging with the content. This would allow for rapidly adjusting our messaging/targeting to generate more efficiency. In a way, machine learning allows marketers to not only create real-time content but adjust in real-time as well.”
• Netflix has embraced AI and predictive algorithms for years; as far back as 2012, it was estimating that 75% of its subscribers’ viewing selections were based on algorithm-generated recommendations. It also knows not to over-personalize the choices it puts in front of viewers, so it occasionally stirs in a variant title.
• Visit Clickotron.com and you might think some of the clickbait-y headlines there read awkwardly, but they’re impressive when you consider the fact they were all written by an AI. Norwegian developer Lars Eidnes designed a clickbait generator using a neural network that’s able to create eyeball-grabbing headlines formulated to seize on our undying thirst for sensationalism. Our favorite Clickotron headline so far? “Residents Can’t Remember If They Lost Their Wine At The Same Time.”
• Articoolo is an Israeli startup that’s taking a more straight-laced approach to machine-written content: they sell actual content that’s been authored by their platform, which is based on combining AI with natural language processing to emulate our way of thinking as we’re writing text. Gartner predicts that by 2018,20% of business content will be authored by machines.

How you’ll make it happen

The best deep learning systems don’t require a marketer toss out their present marketing automation platform. They’ll partner happily with it, sitting atop a marketing stack and integrating fluently with existing inbound/outbound tools and third-party plugins alike.

Since they’re SaaS systems, companies benefit from ease of integration and scalability, resulting in quick ramp-up so they’re hard at work ASAP.

Follow the six steps you’ll see outlined next to add A.I. to your marketing automation suite, so you’ll be perfectly positioned to reap the game-changing benefits of artificial intelligence.