When you ask your phone for directions, artificial intelligence (AI) serves up the optimal route.

When you browse Netflix or Amazon, AI surfaces movies and products that fit your personal interests. Facebook surfaces ads in your newsfeed based on your mood, and it’s likely you’ve received medical treatment that was either diagnosed or treated with the help of AI.

There’s no arguing that AI is already ubiquitous in the daily lives of consumers.

Artificially intelligent systems have been pioneered by companies like Netflix, Amazon, Google, and Tesla to collect data and use it to compel behavior. Companies like IBM have captured our hearts and imaginations with AI systems like IBM Watson who can outperform any human at the trivia game show Jeopardy.

There’s a lot of buzz around Sales and Marketing AI, but what is real and what is hype? What can we actually expect from AI in the sales and marketing technology in 2018? And how do AI and human-driven processes work together?

AI for Sales and Marketing Tech: High Expectations Meet Reality

Despite Hollywood depictions of AI creating a dystopian future where humans are slaves to robots, the general public tends to disregard this notion. But unless you’re directly working on AI, your ideas of AI likely come from watching IBM Watson beat the Jeopardy world champions in 2011 – or when, in 2016, Google’s AlphaGo defeated a world champion at the ancient Chinese game of Go.

Even if you have not heard of either of these, you probably have at least had some lonely conversations with your AI assistant, SIRI.

With these awe-inspiring displays of superhuman intelligence, it’s no wonder Marketing teams will spend $2 billion a year on AI tools by 2020.


But if it seems like we haven’t realized all the tantalizing, fantastic possibilities we imagine this new technology to bring, that may be because expectations are high … perhaps unrealistically so.

At a recent Marketing Technology conference, I informally surveyed a group of marketing leaders I happened to be seated with during lunch. I was curious to know how AI developments might influence their future investments.

Responses included:

  • “AI will manage customer relationships through predictive marketing.”
  • “Sales and marketing will be able to align around the right buyer personas.”
  • “We’ll see a faster, smarter, more personalized buyer journey!”
  • “It will optimized sales interactions, increasing revenue and reducing wasted research time.”
  • “Personalized content creation at scale.”

To me, these expectations echo the hype surrounding the possibilities of what AI can do … rather than the reality of what B2B AI is actually capable of today.

We’re already on the path of AI adoption – but we’re not that far down the path!

adoption curve of B2B artificial intelligence

You may remember seeing Everett Rogers’ model in a college class. This famous chart from 1957 is still relevant in explaining the adoption of new innovations today.

The B2B business community is only just now entering the “Early Adopters” period. The primary development and experimentation of AI marketing tools today are still being pioneered by large firms like Google, who are assuming more risk and less short-term results as products mature. Innovations tend to show greater ROI when the majority of the population has adopted the technology – when ample research and development has taken place, and the product or process has been fine-tuned.

According to a new report from Narrative Science, 38% of enterprises are already using AI technology in some form, and that number is expected to nearly double to 62% in 2018 – with the expectation that marketing departments alone will spend $2 billion annually on AI tools by 2020.

I believe that AI will change the way we do business in the future; however, companies should temper zealous expectations with the following reality check before considering AI investments.

The effectiveness of AI depends on the quality and quantity of data being ingested

Perhaps the most enchanting part of watching billion-dollar AI systems like AlphaGo defeat World Champion go player Lee Sedol in 2016, is how seemingly easy the AI makes it look.

lee-sedol-loses-alphago-artificial-intelligenceSedol, intensely focused, often took several minutes to strategize and contemplate his next move during his turn. As soon as Lee made his move, AlphaGo almost instantaneously made its move – and it was Lee’s turn again before he could blink.

AlphaGo went on to beat Lee Sedol three of four games, with Lee winning the final match. This kind of seemingly super-human intelligence have owned the AI spotlight for decades – and fueled the perception that AI systems are inherently intelligent.

This could not be farther from the truth – especially in B2B applications.

What we did not see in the GO match were the enormous datasets used to train the machine learning algorithms over time. Before defeating Lee, AlphaGo had to lose tens of thousands of times against other players – and itself – before even understanding how to devise a strategy.

The same principle applies to the AI tools that are making their way into your sales and marketing systems.

First, in comparison to AlphaGo, the algorithms in your sales and marketing tools are relatively immature. Second, and more important, your database contains only a fraction of the universe of customer and market data needed to meet the optimistic expectations of my lunch mates, cited earlier.

1. AI systems need data quality

Accurate data is critically important for all sales and marketing functions. Even when bad data creeps into our systems, humans are often able to recognize and account it … in ways that a computer can’t. As automation becomes machine learning, and machine learning feeds artificial intelligence, errors and inaccuracy can move downstream to affect entire systems – and decisions.

Sales and marketing data decays 30% per year due simply to personnel moves; in addition, simple error, bias, and duplication contribute to a big bad data problem.

Most companies aren’t investing enough in solving for this challenge, and until they do, AI based on internal data will only be marginally effective.

2. AI systems need data quantity

Large datasets are critical for accuracy. If a dataset is too small, anomalies and variations can appear – incorrectly – to be patterns, and lead to incorrect conclusions. Consider that Google collects data from 40,000 search queries PER SECOND … just for your Maps app to serve up for directions to the nearest gas station.

For sales and marketing, this means gathering robust amounts of data through your sales and marketing systems – everything from fit criteria to behavioral and usage data to intent data.

We’re getting there, too – but we’re not there yet.

The success of machine learning algorithms is entirely dependent on the quality and size of the datasets being used to train them.

AI requires a human touch

Organizations get better at data collection every day – but we still struggle to find context and interpret actionable insights from raw datasets.

The human experts who engineer, refine, and augment data to add layers of value and understanding for the end user – or for training ML algorithms – are at least as important as the data or the algorithms. Additionally, humans have access relevant information that doesn’t exist digitally or which can’t be collected in a legal manner.

The day may come when data integration is so good, when our algorithms are so advanced, that creating meaning becomes an AI function in the B2B world – but we’re not there today. Until then, AI can work in collaboration with sales, marketing, and other business functions, providing analytical predictions and requiring human feedback in return for refinement.
The rise of AI-Assisted Sales and Marketing

We need the help of AI to analyze patterns in large datasets that humans can’t decipher; and AI needs humans to interpret, connect the dots, and introduce non-digital signals and interpret unstructured data. And so the next few years will see Marketing and Sales professionals increasingly empowered by the rise of AI.

Sales and marketing professionals will be able to assume a more strategic and scientific approach to their role, thanks to workflow optimization and deep analytics enabled through AI.

terminator 2

Hollywood’s dystopian view of the world, where artificial intelligence – usually in robot form – enslaves the human race (Skynet, Matrix, Terminator), will remain limited to the silver screen.

By 2020, I believe machine-learning algorithms – the foundation for AI – will have advanced enough that sales and marketing teams will be less driven by specific technology features within a product – and more focused on the accuracy, availability, legality, and interpretability of the data that will feed the AI systems they will direct.

In short:

  • AI is already part of everyday life, but B2B sales and marketing application is behind the adoption curve
  • Accurate, robust sales and marketing datasets are needed to capitalize on AI
  • AI will not replace human sales or marketing functions; rather, it will empower them
  • The importance of specific features in sales and marketing tools will take a back seat to the availability, quality, and accessibility of the datasets used to feed them

Considering the pace of adoption in the B2C community, AI will be a competitive B2B advantage second to none.

But before making any serious AI investment, sales and marketing teams have to recognize that highly integrated, actionable data embedded throughout their sales and marketing systems is necessary for trustworthy results.

[cta id=”14577″ color=”green” size=”full” align=”center”]

Phillip Sundal
About the author

Phillip Sundal

Phillip Sundal currently works for DiscoverOrg as the Product Marketing Manager where he oversees various marketing programs, manages the go-to-market strategy for new product launches, and continuously gathers feedback from the marketplace to guide the product development roadmap. Phillip has a B.S. in Marketing and Social Theory from the Carson Business College at WSU.