Introduction: Why AI is important for brand marketers
Artificial Intelligence and Machine Learning have dominated headlines this past year, promising to disrupt every industry, from Media, Advertising, all the way to Law and Medicine.
Structurally, two of the largest companies in the world, Google and Microsoft, have realigned their groups around artificial intelligence research, tools, and applied AI, and are focused on evolving their primary platforms, that serve over 1 billion users, to apply artificial intelligence to all products and services. Not far behind them are Apple, Facebook, Amazon, TenCent, and Alibaba who are all investing heavily in AI to enhance their entire ecosystem of products.
This is the beginning of the next arms race where AI becomes core to consumer engagement and interaction models.
Almost all consumers have a pocket-sized AI in their pockets. The future of intelligent systems is even brighter as millions of sensors become smarter. In fact, research into quantum computing has the potential to unlock an exponential leap forward in terms of processing capabilities.
“Brands have no choice but to evolve with this new AI first landscape.”
The most important takeaway here is that all the above companies own the consumer engagement layer. In other words, they shape how consumers interact with technology and brands. So, as a brand marketer, we have no choice but to evolve with them and adapt to these emerging behaviors. But how? How do brands leverage AI, machine learning and data to stay relevant in this new AI dominated marketing landscape?
The beginner’s guide to AI.
The first step is to understand AI’s evolution and the different types of AI that brands can leverage to uncover insights and engage consumers.
Artificial Intelligence was originally conceived in 1950, and we have seen examples of AI popping up culturally over the years. From IBM’s Deep Blue victory over a chess grandmaster in 1996 to 2011 and IBM’s Watson dominating jeopardy BEFORE the system was connected to the internet. In 2016, Google’s Alpha Go algorithm beat the world’s best player in GO. Every week, we see new examples of AI “learning” new skills from conquering competitive video games to learning how to walk in a virtual component.
1. Three areas driving this new golden age of AI
1. The decreasing costs of graphical processing units (GPU) for parallel computing
2. Explosion in structured and unstructured data
3. Advancement in algorithms
“AI will influence every corner of the consumer landscape.”
As Moore’s Law inverts and computing becomes cheaper and smaller, AI will influence every corner of the consumer landscape. Initial predictions show the market for AI-driven products and services will jump from $ 36 billion in 2020 to $ 127 billion by 2025*. (*Source: BofA Merrill Lynch Group Research Estimates — 2017 the year ahead: artificial intelligence; The rise of the machines.)
2. Three types of artificial intelligence
Type 1: Artificial Narrow Intelligence (ANI) aka Weak AI — focuses on one task, e.g., Alpha GO.
Type 2: Artificial General Intelligence (AGI) aka Strong AI — Computers that exhibit human level intelligence, e.g., Samantha from the movie Her.
Type 3: Artificial Superintelligence — an intellect that is smarter than any human in every possible field, e.g., the Avengers’ nemesis Ultron
Everything we discuss centers around ANI. AGI and Super Intelligence are still beyond the horizon, and frankly, the verdict is still out if we should even try to reach those states.
3. The Subsets of AI
Let’s dig a little further into how AI works. AI has multiple subsets and classes, each leading to different use cases.
a. Machine learning is the first and most well-known layer. Machine learning uses algorithms to parse data, learn, and make a prediction about a specific use case. The more data we provide, the better prediction engines we will create.
Within machine learning, there are over 15 different techniques from association rule learning, decisions trees, random forests, Bayesian networks, reinforcement learning to genetic algorithms.
But the one machine learning technique that is behind the latest boom in AI performance is Deep Learning.
b. Deep learning uses Neural Networks — multi-layered data structures and algorithms modeled off the human brain — to take a “data up” approach to learning and prediction. Unlike the other machine learning techniques, Deep learning doesn’t require researchers to define features or weigh the data. The Neural Nets do it for you. And this is the true advantage of deep learning over the other machine learning techniques: it’s scalable. Instead of human’s classifying data inputs, researchers can feed massive amounts of data into the system. This has led to advancements in sentiment analysis, image processing, and speech translation, which has provided consumers with powerful services like Google Lens and products like Google Assistant, Siri, and Alexa.
But even deep learning has its own subset of different types of neural nets like convolutional neural nets, recurrent neural nets, gated recurrent units, auto encoders, etc. We won’t go into the details of what they do, buts it’s important to know that each of them has specific use cases. When you are building your data offering, you should understand which nets are optimal for what service you want to provide.
Now that I’ve given you a high-level overview of the types of AI and techniques and why Deep Learning has transformed AI, let’s look at how we can apply AI to marketing.
We’ve broken this down into three core marketing territories: Consumer Engagement,Brand Strategy, and Innovation.
1. Consumer engagement
a. Consumer insights
AI is not bound by volume or memory capacity and can be taught to learn an almost infinite number of different types of structured and unstructured data sets. Structured data sets can be 1st, 2nd or 3rd party data in a relational database that is highly organized and searchable. Unstructured data is the boundless data from the open web. When the data is from social platforms and IoT devices, it can uncover deep cultural insights into consumer behavior. Machine learning’s raw power with the right structured and unstructured data sets allows marketers to segment target audiences in almost unlimited ways to uncover behaviors, perceptions, attitudes, occasions on a micro level.
Besides machine learnings ability to scale massive unstructured data sets, its other advantage, and maybe more important, is its ability to uncover unarticulated, unprompted consumer needs. Using predictive algorithms, researchers can link these findings to behavioral outcomes. The key is to develop the right process framework that provides a structured taxonomy to unstructured data and identifies personality traits that are linked to audience triggers and behaviors. There is a myriad of ways to segment, so having the right plan and creative partner in place is vital.
As we mentioned before, the arrival of deep learning and different types of neural nets led to a punctuated evolution in image processing and speech translation. This opened the door to new use cases and offered brands leverage to enhance their creativity and design. With target audience insights, brands can leverage machine learning to parse and classify massive external unstructured data to understand what type of messaging and creative elements appeal to their targets. This will help inform design elements to use for owned and paid properties.
Mondelez, for instance, used an AI creative director to create an ad for their Clorets Mint Tab brand. With their agency partner, they trained it against a database of deconstructed ads from winners of awards shows to teach it what award-winning work looks like. AI’s potential to learn and create ideas for specific categories points to a future where creativity will be based off algorithms that are created by humans.
c. Experience design
The ability to parse unstructured data introduces new data sinks to inform content strategy, hierarchy, and design layout. The key is to identify the right experiences within your domain to parse and identify content and structure. Another key element is working with the right image processor API to identify the right images and design aesthetics to provide insight into what your experience should look like from a design perspective.
Example: Air Canada leveraged machine learning to personalize the copy and images on their Daily Deals Work by analyzing their customers based on their email behavior. Although specific to one channel, this could have been easily implemented via multiple channels and used to inform the content on their website.
c. Brand advocacy
Another key strength of machine learning and natural language processing is the ability to identify and align areas of creative opportunity with influencers. With unstructured social data, brand marketers can identify key influencers around growth areas as well as understanding the type of content that is resonating with their followers. This will help brands target the right influencers with the right content to increase awareness and diffusion.
Example: Kia* used machine learning to identify which social media influencers to work with for the 2016 Super Bowl. AI was used to understand which influencers matched specific personality types. Once identified, the car company activated an influencer strategy to enable the spread of the campaign.
d. Anonymous Personalization
True 1:1 personalization can only exist with your existing customers and the right real-time dynamic creative engine to deliver those personalized messaging. But how do you deliver personalized messaging to target audiences visiting your brand experience? The combination of machine learning and proprietary data can help you identify anonymous users visiting your site by their behavioral patterns. With these insights, you can now target these users with the right messaging to bring them into your database.
2. Brand Strategy
Brand strategy is about developing the right brand platform that can maintain a leadership position in your category. There are multiple factors that go into developing a strong platform, including a strong brand voice, purpose, promise, attributes, messaging tone and perhaps most importantly, a relevant value proposition. But more importantly, they need to stay ahead of market forces and disruptions from emerging players. Before AI and machine learning, evolving and reinventing positioning was constrained to your proprietary data and market intelligence data sources, which were constrained themselves by how fast they could publish data.
The beauty of AI and Machine learning is that you can now add unstructured data sources to map your category competitors to your positioning over a specified time-set to identify gaps and areas of opportunities. The key is to create a category driven taxonomy to understand how consumers within your category perceive your brand, understand your value, and whether they align with your messaging. This works on the product level, too. What you are seeking is to understand how your product has been perceived over time to identify growth and development opportunities.
Another brand strategy use case for AI and machine learning stems from the advancement of Natural Language Processing and the explosion of voice activated devices and conversational bots. This combination has resulted in brand disintermediation where AI systems step into the personality of the brand and can facilitate customer interactions and drive behaviors. For instance, with the Alexa Dominos skill, customers can place an order, pay (account linking), and track their order. Dominos has successful stream-lined the entire customer experience through an AI system. The key, however, is to identify which behaviors are optimized to be handled by assistants and which behaviors need human interaction.
The drawback to having assistants like Alexa, Siri, or Google Assistant own the customer experience is that they now own the information exchange. The key workaround, although a less scalable solution, is to apply machine learning to your owned properties or create new intelligent experiences that can perform the same tasks and capture data. For instance, Bank of America created their own Erica bot that lives within their app and uses predictive analytics to help their customers save money.
3. Corporate Strategy
Similar to how AI can transform brand platforms, companies can leverage AI to drive value across their portfolio, conquer new markets, enhance their pipeline and even transform culture.
AI is key for c-suite members who are looking to automate high volume core business processes that are comprised of routine and recurring work. The systems can identify patterns that ultimately lead to cost reduction and increases in productivity.
The goal is to identify automation targets such as highly repetitive tasks that have a low demand for human judgment. Processes that are built on decision trees are an ideal automation target for intelligent systems.
Automation through artificial intelligence will be the initial bridge into AI for most organizations. As automation and productivity increases, organizations will begin to look at how to enhance consumer experiences for both B2B and B2C as well as leveraging intelligent systems to drive the end user experience.
Example: GE has been using AI to reduce cost inefficacies by helping to integrate supplier data so they can compare prices of the same product across business units. Brands can leverage this same approach to help cut costs and redundancies among their vendors.
Mergers and acquisition
With the right domain of data and taxonomy, companies use historical and present state performance data to inform decisions on which new or tangential markets to enter, and which companies, based on specific signals, to acquire. The key is to identify the right data sources that can help inform your algorithms to make the right predictions on which opportunities align with your business model and show the right growth potential.
Product Portfolio Enhancement
Companies can leverage AI systems to identify product enhancement and expansion opportunities within their portfolios. By creating taxonomies around consumer conversations, you can gain insights into new products or identify future directions for features and designs of products within your existing portfolio. Another avenue is to apply machine learning to your own proprietary first party data to understand how your own customers are talking about your products to identify new use cases and/or enhancement opportunities.
Finding and hiring talent can be the most difficult task for any manager or leader within any company. AI’s ability to recognize patterns through reinforcement learning can help managers understand performance predictors on how well new talent will perform and develop the right plans for talent to learn and grow within the pressures of that specific environment.
We mentioned earlier how brands can leverage machine learning to identify gaps and opportunities in their positioning within their category. But what about using machine learning to enter new categories or disrupt your existing category. Machine learning, with the right data strategy and technique, can tell brands where and how to innovate to future proof themselves from unknown entrants. Machine Learning, NLP, and Image processing can uncover insights into three key consumer engagement areas that will lead to new products, services, and revenue streams.
The first is to parse through unstructured category level data to identify consumer obstacles by determining where consumers are unhappy or where their needs are unmet.
The next area is category opportunity enhancement where you can use NLP to identify what emotional and functional areas excite consumers and fertile disruption opportunities for new innovation.
The last area is to understand tangential categories where your brand can disrupt. This is a combination of identifying obstacles and opportunity enhancement within that tangential category. The key is to use machine learning to understand the categories your brand can naturally extend into by identifying relevant behaviors.
Besides identifying opportunities for enhancement and disruption, the most powerful use case for Machine Learning paired with unstructured and structured data is to identify the velocity and impact of emerging trends. The key here is to leverage social and search data where you can discover nascent conversations, observe their velocity, and predict whether they will become a trend by observing their state over a period of time.
D. Consumer State
The best technology fades into consumers’ backgrounds, only to be summoned when they need it. AI will help drive this shift by intelligently connecting with other devices to determine which technology, data, and content should be present at that moment. This means consumers will delegate the remedial less important decisions and tasks to their personal assistants and focus on more important decisions or activities like creating content or experiencing new environments or places via mixed reality.
In the near future, consumers will have intelligent systems serving the role of their proxy for interactions with other intelligent systems. Both Facebook and Google launched Group messaging with bot integration to facilitate tasks. And start-ups like x.ai and Clara launched virtual assistance that automatically schedules meetings between humans and other bots. With Google Lens, Google Assistant will be able to translate real-time surroundings into actions and facilitate tasks with other bots in real-time. This will turn the camera into a search engine and automate the friction inducing tasks involved with thinking of brand, category, and affinity phrases to discover items.
For brand marketers, this means we need to identify the opportunities where we can leverage data and technology to influence personal assistants, and identify opportunities where we can directly empower consumers to create and enhance their present-day activities or invoke emotions through immersive experiences.
This puts a great emphasis on building customer-centric experiences. The key here is building a strategy that identifies what tasks customers have designated to their bots and defining a strategy that leverages reactive data for the bot to achieve contextual awareness.
Regardless of how the future state shifts and evolves, AI will be the underlying layer driving consumer engagement. We see a future where AI systems will become consumers’ proxies, driving decision making and facilitating tasks on the consumers’ behalf. All of these elements will be highly dependent on data and decisioning through intelligent systems via artificial intelligence.
The various subsets of artificial intelligence will continue to be interconnected, redefining how brands approach connecting with consumers. AI makes it possible to know the consumer better than ever before through the ability to scale infinite data sets and segment to uncover micro behaviors and attitudes. If approached correctly, with the right mix of AI subsets leveraged, companies will see their businesses grow.
Let’s block ads! (Why?)