Walk into a conference session, open a newsletter, or gloss over your LinkedIn feed and you'll be hard pressed to walk away without a new take on artificial intelligence (AI).
When you strip away the rhetoric and hyperbole, you're probably wondering—what can AI do for me? What should I keep my eyes out for in the next few years? How does this add complexity to my tech investments?
I sat down with NextRoll's Senior Data Scientist, Terry Feng, to get to the bottom of how we should be thinking about AI and machine learning when it comes to bringing account-based marketing (ABM) to scale.
Q: Before we dive in, can you give us your simplest definition of AI? What are the various categories we should be thinking about?
In the broadest sense of the word, AI is an umbrella term for an intelligent program or machine that is capable of taking in information and using it to make reasonable judgements or actions.
AI spans many subfields such as machine learning, computer vision, natural language understanding, robotics, and expert systems. The goal is to achieve an Artificial General Intelligence (AGI) that is capable of learning and understanding new tasks as any “intelligent” being would, without any engineering or alteration to the original system. Yes, just like in the movies!
Q: For marketing and sales teams, we've seen machine learning top the list of advancements showing the greatest impact. Can you tell us a bit more about how machine learning works?
The field of AI has hit many important milestones and accomplishments in the recent decade, with AI systems having achieved near-human to even superhuman performance in common human tasks such as speech synthesis and image classification.
In more recent years, we have seen AIs that are capable of competing toe-to-toe with professionals in complex strategy games such as Go, and Dota 2. These achievements can be primarily attributed to advancements in machine learning, the science and study of algorithmic systems that are able to learn from data to perform specific tasks without explicit logic or instructions.
Machine Learning is the primary driver of AI, based around the idea that a computer given access to data and/or ways to interact with its environment, can learn for itself an optimal way to accomplish a desired outcome. Simply put, machine learning enables computers to carry out tasks in a statistically principled way, one that we would consider “smart.”
Within the field of machine learning, there are three main categories: supervised learning, unsupervised learning, and reinforcement learning. Supervised learning is predominantly associated with prediction and uncertainty quantification, unsupervised learning with data mining and information extraction, and reinforcement learning with planning and decision making. Often in complex systems, all three branches of machine learning would be engineered into one system.
Q: What can machine learning do for technology used in marketing and sales teams?
Being at the right place, at the right time and with the right message is the foundation of any strong marketing strategy. Imagine a sales team being flooded with millions of low-quality leads, not all of which are great. You wouldn’t want to put all of your resources against low-fit leads. This is also the case for companies running with an account-based approach. You want to make sure the best-fit accounts are prioritized over low-fit accounts, ultimately making your marketing team more efficient.
How might the team manager with finite capital optimally allocate resources so that ROI from sweat and tears can be maximized? Or for a marketing team looking to change their company image, how might they spread their message to their audience to have the desired impact? In either case, informed decisions need to be made so that the right people can be targeted. Whether the proper audience is addressed will be the deciding factor between success and failure.
The promise of machine learning is to be able to find patterns and insights within data that a human cannot—at least not without considerable effort.
This makes machine learning a particularly intriguing tool for sales and marketing teams, as many of the decisions that are traditionally made by intuition alone can now be fortified by data and intelligence. Accounts can now be properly prioritized, with insights detailing why. You now can precisely select physical locations, industries, web domains and contacts for targeted ad audiences. The customer journey can now be personalized even to a massive audience, with minimal human-induced error.
Q: What’s one myth you’d like to shed some light on? What do you see happening in the space over the next few years?
To the uninitiated, AI often look like black magic, and in some cases, rightfully so! But even the most advanced of machine learning algorithms will not be able to decipher meaningless input. The learner is only as good as the material it learns from. Unreliable data results in unreliable decisions, and as the mantra of data science goes: garbage in, garbage out.
ABM is about sending the customer on a memorable experience. In recent years, advancement in generative text and images are getting ever closer to being indistinguishable from that of a human. It would not surprise me if we start seeing personalized advertisements crafted with the help of an AI for every step of the customer journey in the near future!
Q: We’ve recently released updates to our Account Selection product, including account scoring, account suggestions, and account groups. What’s your simplest explanation for how account scoring works for RollWorks customers?
Across our entire platform, the Data Science Core layer applies machine learning models at scale across the entire spectrum of our three solutions from account scoring and suggestions to dynamic messaging and bidding and budgeting. For instance, our NextRoll infrastructure allows us to operate at over one million predictions a second.
Diving into Account Selection specifically, our newest account scoring product uses machine learning algorithms to determine the distinct characteristics that makeup your current or targeted accounts, uncovering the dimensions that you might not have considered as a strong signal.
It uses these insights to help you grade and prioritize your accounts and to find the needles in the haystack. It is a technology that will help you easily scale your business and increase your ROI.
Q: Since you were an integral driver of our newest features, tell us a little bit more about what a day in the life of a data scientist looks like.
The ultimate goal is to deliver actionable intelligence to stakeholders so they can steer the ship in the right direction. Consequently, building monitoring systems and investigating any anomaly that comes up; conducting analyses to find holes and gaps in the industry and in our product offerings; conjuring up hypotheses and theories and designing experiments and statistical models to test the claims are all part of the day to day.
And as data science is a fast-moving field, keeping up with the latest research is paramount. The research and development of our machine learning systems is integral to keeping our competitive edge and our ability to deliver performance for our clients.
Terry Feng is a senior data scientist on the Feature Engineering team at NextRoll. His primary responsibilities include the research and development of statistical learning models and the optimization of NextRoll’s real-time-bidding (RTB) pricing and targeting algorithms. Prior to NextRoll, Terry designed and developed scalable pipelines and production machine learning systems as a machine learning engineer.
About the AuthorFollow on Linkedin More Content by Caroline Van Dyke