The hype around AI and ML has continued to gain momentum as every major industry from marketing to healthcare to manufacturing to transportation to finance to retail, and beyond have started to leverage advances in AI and AI-based applications to improve productivity and performance. Economists have hailed AI as a core enabling technology of the “fourth industrial revolution.1” With all of this excitement, every executive has started thinking about how their business will use AI to not only survive this revolution, but excel amongst their cohorts. According to PwC2, Artificial Intelligence will contribute $15.7 trillion to the global economy by 2030. Clearly, there is a huge opportunity for businesses to benefit from investing in AI. The MIT Sloan Management Review 2017 Artificial Intelligence Global Executive Study and Research Project found that 85% of executives believe that AI will help their businesses gain or sustain competitive advantage.
As with any technology that fuels the public and business imagination, there is a lot of confusion around the terms “machine learning (ML) and “AI.” Many people use AI and ML interchangeably, while others utilize them as discrete, parallel advancements. This confusion ripples into the common understanding of AI and ML, where much of the public discussing these advancements are unaware of the distinctions between the two.. For some, this dilution of the terms is overlooked in favor of creating hyper-excitement for advertising and sales purposes.
This chapter will provide you with working definitions for AI and ML that we’ll carry through the rest of the book, and we’ll also explore how these concepts support the framework of a Lean Startup.
Artificial intelligence (AI) is often used as an umbrella term to describe types of technology that can simulate human intelligence. It’s the ability of a computer program or a machine to think and learn. It is also a field of study which tries to make computers “smart,” where computers work on their own without being encoded with commands.
John McCarthy came up with the name “artificial intelligence” in 1955. Artificial Intelligence is essentially building a machine or computer program to parse data to learn from different signals to make smart decisions on predictive actions to achieve a given goal or desired outcome. Artificial intelligence is the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.
The term “artificial intelligence” has a broader meaning. It’s the idea that machines and computers can complete tasks normally requiring human intelligence. AI as we know it today is symbolized with Human-AI interaction gadgets by Google Home, Siri and Alexa, by the machine learning powered video prediction systems that power Netflix, Amazon and YouTube, by the algorithms hedge funds use to make micro-trades that rake in millions of dollars every year. These technology advancements are progressively becoming important in our daily lives. In fact, they are intelligent assistants that enhance our abilities, making us more productive.
In contrast to machine learning, AI is a moving target, and its definition changes as its related technological advancements turn out to be further developed. What is and is not AI can without much of a stretch, be challenged, making machine learning very clear-cut on its definition. Possibly, within a few decades, today’s innovative AI advancements will be considered as dull as calculators are to us right now.
Let’s define machine learning (ML) since it is such a ubiquitous term these days. Machine learning is a type of artificial intelligence developed around allowing computer system to progressively improve performance on a task by “learning” through a range of statistical approaches. Put another way, machine learning is the development of algorithms that allow for more and more accurate predictions based on the incremental collection of data.
It also worth clarifying that artificial intelligence is not the same as machine learning because artificial intelligence is a broader concept. Machine learning is simply one of the most common and approachable applications of AI. The machine learning concept is about giving machines access to data to enable them to learn from it themselves. AI incorporates a bunch of technologies that include machine learning, deep learning, inference algorithms, natural language processing, neural networks, and computer vision — are starting to show real promise, despite the hype and confusion in the marketplace. Artificial intelligence is at a critical point in its evolution, especially as it relates to marketing automation.
Machine learning works best with large data-sets by examining and comparing the data to find common patterns and explore nuances. Machine learning automates model building for data analysis. The concept behind machine learning is that a computer can learn from the data it analyzes by identifying patterns. Ultimately, this technology can make decisions without humans. Essentially, it gives machines the ability to learn and adapt through experience. Here’s how it works: the system uses probability to make decisions or predictions based on the available data. It then uses feedback loops to find out if its prediction was right or wrong. Predictions get more and more accurate — and the system gets smarter.
An example of a startup that has successfully leveraged ML is IMVU, which works with third-party CRM platforms like Leanplum to help them collect a huge amount of customer data — from user profiles, location, revenue and usage stats, to which features users are interacting with, and much more. But on its own, this data only tells us what happened in the past. Machine learning, on the other hand, uses this information to predict the future for example which users are most likely to churn that need critical attention to keep them engaged to experience the value prop in the product. Data is useless if it doesn’t help you make better decisions in the future to achieve your desired business goals.
Launching and growing a new startup has always been a challenging endeavor – which is why you need to be open to trying new and innovative approaches to increase your odds for success. The chances of being successful can be significantly increased simply by taking a rational and systematic approach to finding the best strategy for running the business. The Lean Startup has been one of the most successful systematic approaches to date that has been widely adopted across the globe, changing the way startups are built and new products are launched.
The Lean Startup is a methodology for developing businesses and products, which aims to shorten product development cycles and rapidly discover if a proposed business model is viable; this is achieved by adopting a combination of business-hypothesis-driven experimentation, iterative product releases, and validated learning. The Lean Startup method teaches you how to drive a startup-how to steer, when to turn, and when to persevere-and grow a business with maximum acceleration. It is a principled approach to get a desired product to customers’ hands faster.
Central to the lean startup methodology is the assumption that when startup companies invest their time into iteratively building products or services to meet the needs of early customers, the company can reduce market risks and sidestep the need for large amounts of initial project funding and expensive product launches and failures.
The same approach to iterative learning by continuously running experiments is practiced by the best growth teams, who take the same systematic testing and tweaking approach to make the business grow as fast as possible. The key is to take action— to try, to fail, to learn, and to win by learning faster than anyone else.
Taking the wisdom of The Lean Startup approach into the golden dawn of artificial intelligence, we can radically improve our chances of successful outcomes. It’s basically running experiments on steroids. A properly instrumented approach to modern artificial intelligence, machine learning and automation combine to offer companies large and small the ability to conduct far more experiments simultaneously. This speeds up the process of finding successful experiments -- some of which you’d never have taken the time to test in a pre-AI world. Incremental experiments that otherwise would have been sidelined for cost or complexity are now valid for observation in the world of autonomous marketing.
Here are three key factors that have helped accelerate the recent advances in AI, which is leading to more major demand side and supply side partners adopting AI to offer more robust paid advertising options to growth teams to get better performance on their paid user acquisition budget.
The price performance of computing power has grown exponentially in alignment with Moore’s Law.3 Exponential means that the computing speed doubles and/or the price drops by half year over year. In recent years, machine learning as one of the key drivers of AI advances has greatly benefited from GPUs (Graphics Processing Unit). GPUs are very performant for conducting vectorised numerical operations which are needed for all machine learning calculations. Google’s TPU (Tensor Processing Unit) is another example where (co) processors are optimized for machine learning problems. With great advances in quantum computing it is very likely that this trend will continue and accelerate. Quantum computers give us the ability to solve complex problems that are beyond the capabilities of classical computers like encryption, optimization, and other similar tasks. Quantum computers can analyze large quantities of data to provide artificial intelligence machines the feedback required to improve performance. Quantum computers are able to analyze the data to provide feedback much more efficiently than traditional computers and therefore accelerate the learning curve for artificial intelligence machines. Just like humans, artificial intelligence machines powered by the insights from quantum computers can learn from experience and self-correct. Quantum computers will help artificial intelligence expand to more areas of growth marketing and technology become much more intuitive very quickly.
Data is the fuel for AI and your own customer data is the most precious asset. There is an accelerated generation and availability of data fueled by the increased use of connected devices like mobile technology and social media. In fact, the number of internet users has grown over a billion in the last five years, more than half of the world’s web traffic now comes from mobile phones. Large amounts of data are essential to successfully do machine learning and therefore achieve a high accuracy of their predictions for growth marketing questions like which of your customers are most likely to purchase or churn in the near future. More users behavior actions are being tracked and shared than ever before, flooding out of the dozens of connected devices we use every day, and it shows no signs of slowing down. Knowledge is power and there’s a lot of knowledge trapped in your internal data as well as external sources. To best unlock that knowledge, you have to consider the type of data you need, where to look for it, how to get it, and how to build the right data models to analyze your business questions. And just as importantly, you need to continually update your data to re-train and enhance the algorithms. There’s certainly a lot that goes into data collection, but it’s worth it. As the lifeblood of AI, data is critical to helping you get the business insights you need to move your business forward.
Algorithms are used for calculation, data processing, and automated reasoning.” Whether you are aware of it or not, algorithms are becoming a ubiquitous part of our lives with the biggest focus on machine learning include data mining and pattern recognition. Algorithms can make systems smarter, for example movie recommendations from Netflix. However, there is a common principle that underlies all supervised machine learning algorithms for predictive modeling. Machine learning algorithms are described as learning a target function (f) that best maps input variables (X) to an output variable (Y): Y = f(X). The most common type of machine learning is to learn the mapping Y = f(X) to make predictions of Y for new X. This is called predictive modeling or predictive analytics and the goal is to make the most accurate predictions possible. The questions being raised about algorithms at the moment are not about algorithms per se, but about the way society is structured with regard to data use and data privacy with data breaches becoming more prevalent. It’s also about how models are being used to predict the future. Algorithms embedded into the technology through which we access so much information could be shaping what information we receive, how we receive it, even how we react to it. And AI might be shaping our behaviour, not as an unintended consequence of its use, but by design. Technology, often aided by AI, is exploiting human psychology to shape how we behave.
Today, artificial intelligence, sensors, and digital platforms have already increased the opportunity for learning more effectively — but competing on the rate of learning will become the key difference between the startups that succeed and those that fail. Companies that embrace AI will be able to test, learn and iterate much faster, raising the competitive bar for learning.
The benefits will generate a "data flywheel" effect4, which is the idea that more users get you more data which lets you build better algorithms and ultimately a better product to get more users. Rinse & repeat — companies that learn faster will have better offerings, attracting more customers and more data, further increasing their ability to learn. This is similar to the Lean Startup premise where every startup is in a constant flux of running experiments to create a feedback loop around “Build-Measure-Learn” using data to answer key questions on whether to preserve or pivot. But when it comes to the specific of a product release or acquiring new users, all that matters is: do we have a strong hypothesis that will enable us to learn? If so, execute, iterate, and learn. We don’t need the best possible hypothesis. We don’t need the best possible plan. We need to get through the build-measure-learn feedback loop with maximum speed. The same applies with customer acquisition, where the goal is to leverage AI to speed up the velocity of experiments at different stages of the customer marketing funnel below to enable startups to learn or fail fast with minimum impact on cash burn rate. The end goal is to figure out how to move new customers further down the funnel faster powered by AI + Data to get smarter with optimizing the right levers.
According to eMarketer report5 “Artificial Intelligence for Marketers 2018: Finding Value Beyond the Hype”, the advent of new algorithms, faster processing and massive, cloud-based data sets is making it possible for companies in all industries to experiment with artificial intelligence (AI). Here are the key takeaways from the eMarketer report on the AI industry trends for marketers:
Investment and interest in AI remains high, though large-scale adoption is happening more slowly. Still, many companies have ambitious plans for AI systems and are looking to them to improve their business operations.
AI technologies—including machine learning, deep learning, natural language processing and computer vision—are starting to show real promise, despite a significant amount of confusion in the marketplace.
A robust ecosystem of prepackaged APIs, open-source software and cloud-based platforms is helping accelerate AI adoption, bringing new capabilities to speed up, scale and personalize marketing campaigns in more economical ways.
Agencies and other consultants are stepping up to the plate, beefing up their technical resources and forging technology partnerships in an effort to help their clients navigate the dizzying array of AI and marketing-tech solutions.
Best practices for marketers include clearly defining business goals, thoroughly understanding the technology, planning for the future, having the right data and using AI ethically.
The AI IN MARKETING 2018 report from BI Intelligence shared the following key insights on the key challenges and opportunities on leveraging AI in marketing:
Digital marketing industry is already focused on streamlining operations and reducing costs, integrating AI takes it even further. Common current uses and applications of AI in digital marketing are cost and ROI analysis for performance advertising on search, social media sentiment analysis, and chatbots for customer service.
Marketers are increasingly incorporating artificial intelligence (AI) tools into their strategies. Over half (51%) of marketers currently use AI, and an additional 27% are expected to incorporate the technology by 2019. This represents the highest anticipated year-over-year (YoY) growth of any leading technology that marketers expect to adopt.
But the rapid pace of innovation is contributing to marketers’ sense of unpreparedness for AI implementation and future use cases. When asked to choose which trending technology they felt most unprepared for 34% of global marketing executives chose AI, the most of any option, according to Conductor.
AI is advancing beyond data analysis and moving rapidly into data generation, as machines get better at automating two basic human senses: sight and hearing. AI technology has now developed to the point where gleaning insights from data-rich media like voice and video is possible, and humans no longer have to manually categorize or describe various types of media.
AI will transform marketers from reactive to proactive planners. The enhanced analytics that AI provides will help marketers more efficiently plan and execute campaigns in three main areas: segmentation, tracking, and keyword tagging.
Programmatic advertising will become smarter and more automated. Implementing AI and getting the most valuable insights depends on a reliable, consistent flow of data to train algorithms and help them learn and improve over time. Programmatic ad buying generates billions of data points, and over the next few years, AI will reduce the manual oversight of programmatic ad campaigns, and help optimize ad parameters in real time.
AI will aid in content creation, but human marketers are still necessary. It’s still early days for marketers to use AI to automatically create editorial content or stitch together the right image with the right messaging for display ads. Machines will help cut down on production time, but humans are needed for their creative juices and ability to inform strategy.
Both of these research analysts reports clearly articulate that AI is at a critical point in its evolution. The future of marketing trends in the world of AI looks extremely bright if you can start to figure out now how to fully leverage it to drive more growth in your business. There is no denying that things will get even more exciting with the adoption to 5G to introduce marketers to retail apps that take foldable displays into account, better-timed and longer ads, and tighter integration between mobile and store experiences. However, the reality is that it will take 5G another five years before it reaches critical mass among consumers in most countries. The user data you can capture across many different touch points on a daily basis is limitless. It’s only going to get faster, cheaper and bigger. And if you’re not leveraging AI to make sense of all the data coming at your organization at such a high velocity, you’re likely to be left behind your competition.
The question is this: how are you going to best leverage AI to give your startup a competitive advantage to take better actions to scale up your user growth efforts to hit your success goals? Today, artificial intelligence, sensors, and digital platforms and a proliferation of data have already increased the opportunity for learning more effectively — but competing on the rate of learning will become a necessity in the 2020s. The dynamic, uncertain business environment will require startups to focus more on discovery and adaptation rather than only on forecasting and planning.
The startups that move fast to adopt and expand their use of AI will raise the competitive bar for learning. And the benefits will generate a “data flywheel” effect — startups learning faster by attracting more customers and more data, further increasing their ability to learn and scale up growth at a faster pace than their competitors. For example, Netflix’s algorithms take in behavioral data from its video streaming platform and automatically provide dynamic, personalized recommendations for each user; this improves the product, keeping more users on the platform for longer and generating more data to further fuel the learning cycle to scale user growth.
There are many exciting ways you can apply the power of AI and ML to streamline marketing processes across the entire customer marketing funnel to help growth teams work smarter by automating in the areas below to help them stand out vs. their competition:
Predicting customer behavior
Data analysis and reporting
Better cross-platform attribution
Creative development and iteration
I’ve found plenty of examples of ways that AI is transforming growth marketing to allow us to achieve things that would never have been possible without it. With AI, you can work smarter and gain a holistic, real-time view of your customers and their relevant interactions throughout the entire journey. AI lets you act quickly on your data and makes it easier to focus on the higher value work by getting fast actionable insights.
However, while the data to support AI is critical, data is nothing without a clearly defined business problem focused on cost reduction, risk reduction and profit. Perhaps the most interesting thing about AI is that, while it can automate and do “work” at greater efficiency, it uses machine learning to “think” and “learn” over time, strategizing, designing, recognizing patterns, and making decisions. If that sounds a lot like a human brain, it’s because deep learning, one of the most important methods of machine learning, is based on the idea of a neural network, modeling the structure and function of the human brain.
Assessing the Maturity of Autonomous Marketing - with help from the self-driving car folks
With ambitions to launch self-driving cars to the public in 2020, Tesla gets a lot of attention in the autonomous vehicle industry. But big automobile companies, start-ups, and tech giants are all working to deliver safe, self-driving vehicles to the masses.
To make sense of where artificial intelligence and automation is at and where it’s going, industry trade association SAE (Society of Automobile Engineers) introduced its autonomy scale. It helps the industry determine and classify different levels of autonomous capabilities for vehicles.
SAE Autonomy Scale
|Level 0||No automation. The driver controls steering, and speed (both acceleration and deceleration) at all times, with no assistance at all. This includes systems that only provide warnings to the driver without taking any action.
|Level 1||Limited driver assistance. This includes systems that can control steering and acceleration/deceleration under specific circumstances, but not both at the same time.|
|Level 2||Driver-assist systems that control both steering and acceleration/deceleration. These systems shift some of the workload away from the human driver, but still require that person to be attentive at all times.|
|Level 3||Vehicles that can drive themselves in certain situations, such as in traffic on divided highways. When in autonomous mode, human intervention is not needed. But a human driver must be ready to take over when the vehicle encounters a situation that exceeds its limits.|
|Level 4||Vehicles that can drive themselves most of the time, but may need a human driver to take over in certain situations.|
|Level 5||Fully autonomous. Level 5 vehicles can drive themselves at all times, under all circumstances. They have no need for manual controls.|
I propose a similar scale for the purpose of evaluating autonomous marketing and marketing automation solutions.
The Lean AI Autonomy Scale
|Level 0||No automation. Marketers manage all tasks with basic tools and CRM systems that provide no real automation, but act as storage repositories for marketing data and results reporting (dashboards or “business intelligence” systems). .|
|Level 1||Recommendation automation. Marketers leverage systems capable of following business rules (defined by the marketer) to make business recommendations for optimizing marketing outcomes. Examples include dashboards with recommendation systems for adjusting marketing spend by channel. The user must make the final step of making the recommended adjustments.|
|Level 2||Rules-based automation. Building on business rules set by marketers in Level 1, Level 2 rules-based automation goes the next step and adjusts marketing campaigns automatically (generally via an application or API) without user intervention or approval. Such systems rely on the user to create the rules; dynamic market conditions shifts on a daily, hourly or even minute-by-minute basis gen render rules-based systems brittle or overhanded.|
|Level 3||Computational autonomy. Systems that use machine learning to observe, learn and improve outcomes based on statistical analysis combined with marketing automation. No intervention is required by the user, apart setting goals or broad-based parameters such as campaign dates or geographies for digital campaigns.|
|Level 4||Insightful autonomy. Systems that understand contextual meaning of user interactions, content, behavior, performance data and more to personalize 1:1 marketing messages across various channels and drive optimal performance for operators.|
|Level 5||Fully autonomous. Level 5 systems build Insightful Autonomy capabilities but generate their own unsupervised tests, creative variations, targeting parameters and more with no ongoing intervention from the marketing team.|
I will go into more detail in the next chapter on how this scale in being applied into autonomous marketing and marketing automation solutions. Most growth teams are in the process of figuring out how to reach a level of proficiency to move from Level 0 to Level 2. However, the biggest challenge and opportunity is to advance from Level 2 to Level 5, to fully reap the full superpowers of artificial intelligence to scale up your efforts into the world of Customer Acquisition 3.0.
1 Professor Klaus Schwab, Founder and Executive Chairman of the World Economic Forum, characterized the Fourth Industrial Revolution by a range of new technologies that are fusing the physical, digital and biological worlds, impacting all disciplines, economies and industries, and even challenging ideas about what it means to be human.
2 2017 PwC’s Global Artificial Intelligence Study: Exploiting the AI Revolution.
3 Moore’s Law is a computing term which originated around 1970; the simplified version of this law states that processor speeds, or overall processing power for computers will double every two years. A quick check among technicians in different computer companies shows that the term is not very popular but the rule is still accepted.
4 CB Insights | The Data Flywheel: How Enlightened Self-Interest Drives Data Network Effects. June 2017
5 Artificial Intelligence for Marketers 2018: Finding Value Beyond the Hype,” October 2017. Author: Victoria Petrock. Copyright eMarketer.