With artificial intelligence (AI) taking center stage, business leaders across all industries are discussing (and considering) the implications of automation and personalization.
The excitement levels have risen and are being felt from conferences such as this year’s CES and SXSW to boardrooms around the world. Executives are beginning to see the opportunities that AI can bring when differentiating their offerings, personalizing their services, designing their products, and optimizing their operations.
With this interest comes the biggest question: How do we implement AI in a natural way, both for our people and our customers? Executives looking to explore AI within their organizations are faced with three major challenges: their own understanding of AI, a shortage of data skills, and the identification of this technology’s business value.
Personal AI Knowledge and Understanding
One of the biggest challenges we see executives face is their own personal knowledge regarding the potential and application of AI. This lack of understanding is influenced heavily by the film industry; Hollywood and the media have painted a pretty grim picture of AI and the inevitable doom it will bring to jobs and skills.
To qualify the impact that AI can have on business transformation accurately, executives must separate fact from fiction by studying what AI is, what it can really do, and how it helps build more meaningful products, services, and customer experiences. This understanding will allow business leaders to use AI as a superpower, shaping both the future of their organizations and their jobs. And as a bonus: it will also allow them to educate others (colleagues and consumers alike) on the deployment, benefit, and usage of AI-empowered interactions.
The US will face a shortage of about 140,000 to 190,000 Data Scientists by 2018.
Data Skills Shortage
McKinsey predicts that we will face a shortage of about 140,000 to 190,000 Data Scientists by 2018 — and that’s in the US alone. This doesn’t include the imminent (and similarly severe) shortage of about 1.5 million Data Strategists with Data-Driven Decision Making skills. Facing a serious challenge indeed, executives will need to dedicate time and resources to acquire and retain talent in a market with such high demand for these skills.
Spotify, Stitch Fix, and Birchbox have not only attracted laudable data science talent as startups, but they have also managed to use data and these skills as a differentiating factor within their product offering. Contemporary examples such as these are very telling of the strategy, messaging, and values that business leaders will have to portray to intrigue future employees. Those executives looking to attract great Data Scientists and Strategists will have to prove a strong commitment to data and the (r)evolutionary role that it will play in their organizations.
Identifying AI ROI/Value
A contextual understanding of AI and its relevance combined with a strong point of view and a good Data Science team (to execute on your vision) is only half of the battle. The biggest, ongoing obstacle that executives face is their ability to architect a process for applying data intelligence to their business and decision making. Executives require a strategic approach in order to invest in the right AI projects.
By infusing AI into the organization and quantifying return, decision-makers will be able to focus investment in specific areas. We built the AI Value Pyramid to help executives do exactly this — cultivate a strategic approach that shows the impact of investing in the transformation of their organization into one that puts AI first.
I built the AI Pyramid to help executives infuse AI into their organization and quantify returns.
The AI Value Pyramid
The easiest way to infuse AI into your organization (and justify the investment) is by identifying recurring activities that can be automated using Machine Learning and the subsequent reduction of costs. In almost every large business, there are processes that require a human mind to take in data and perform a very simple, mental calculation. These simple tasks require various levels of human judgment and are not suitable for automation using traditional, hard-coded rules. Methods in Machine Learning, however, thrive in this space. They consume data and approximate human reasoning to either greatly reduce the human intervention needed or eliminate it completely — thereby freeing up those human resources for more valuable activities.
Automated Product Classification for a Large US Retailer
Traditionally, eCommerce retailers have had to write complex and time-consuming rules to categorize and tag images in their Content Management Systems with human intervention. Over time, all of this human effort creates a rich, historical dataset of tagged content that can be used to automate the process. We applied a Machine Learning approach to build an intelligent model that automates the categorization and tagging of images. In the case when the model was not very confident of its classification decision, it would request human intervention. This significantly reduced the time required to classify products and probed human intervention only when required.
In the second layer of the pyramid, we push executives to be a little more adventurous with AI. Identify opportunities across your service journey that impact your business at scale. And look for critical moments that can be optimized further through the application of Machine Learning.
Building Rich Customer Analysis through Website Behavior for a Large US Retailer
Prior to applying a Machine Learning approach, this retailer employed a persona-based marketing approach on its website. We delved deep into one segment — “Millennial Moms” — and analyzed low-level web behaviours such as what they browsed, searched, clicked on, and eventually purchased. We identified that, despite the fact that they all belonged to one single persona, what these users were really looking for was quite different. There were richer, more refined segments that these users needed to be segmented into. By conducting a simple A/B test for 50% of the audience with richer personalization, we proved that the Machine Learning approach increased sales and reduced churn.
When your organization is ready to make a stronger commitment to data, you need to focus on acquiring the right data. Building an AI roadmap that embeds the right receptors across your customer journeys allows you to better understand your customer and explore how he/she uses and interacts with your service. This rich data will allow your data science team to dig deeper and find meaningful patterns that you can use to engage with the consumer in new ways. Something like this simply wasn’t possible before Machine Learning.
Architecting Meaningful Business Solutions for a Retail Ecosystem
With retailers’ declining sales in a post-Amazon world, mall operators are looking for efficient ways to analyze and invest in the right stores at the right locations in the right regions. After all, agile operations are key contributors to stores’ success — not to mention that of the mall experience, as a whole. So when a top-three, global mall operator came to us with rich data (that they had been sitting on for a while), we went into experimentation mode with the power of Applied Machine Learning. A few weeks later, we put together a strategic roadmap of business solutions that not only helped the mall operator decipher which stores were more successful than others but also enabled the retail ecosystem as a whole. The mall operator was able to see how stores affected each other and take action accordingly; it was able to integrate the digital solutions necessary for seamless experiences (e.g., mall-wide gift cards) and, perhaps most exciting, it was able to use the data to predict and plan for day-by-day behaviour — making the environment come to life.
The rise of ubiquitous interfaces like Amazon Alexa and Google Home have attracted a lot of executives to the chatbots space. This is an area that we now see trending across incoming client requests. While building a chatbot can definitely aid in reducing cost, increasing efficiency, and enhancing insights about your customers, the right context that accentuates the existing customer experience is integral to its success. With chatbots on the rise, minimal consumer understanding, and a lack of brand discovery across platforms, building a truly meaningful chatbot requires deep insight and a thoughtful execution.
Building a Rich, Voice-Powered Brand Experience for Patrón
While Patrón enjoys great brand equity and dominates the premium tequila market, consumers were unaware of the myriad of ways it can be used in other cocktails. Research suggests that Amazon Alexa is most often used in the kitchen or living room — prime real estate for the tequila producer. So, we built an Alexa bot that leverages an extensive site catalog of expertly curated cocktails to guide customers in making the perfect drink, all while educating them on the diversity of tequila.3 Needless to say, people took to the utility of the experience. Patrón saw a 38% increase in traffic (109% in the case of cocktail pages) with over 27,000 people currently using the skill on Alexa devices.
The higher you’re able to elevate your organization in the AI Value Pyramid, the more mature it becomes in an AI-first world. At this stage, your organization should have a mature data practice — an existing process for acquiring data, translating it into insights, and delivering it in execution — and a strong customer feedback loop. The combination of these is what enables you to identify journeys of differentiation that you can automate.
World Class Automation Experts at Amazon
The core of Amazon’s focus has always been increasing the speed of delivery for their customers. The organization uses its data chops to identify the behaviours that will remain constant in the future and optimize those experiences as much as it can. For example, customer service is a hub of these types of behaviours. From being able to recommend the right products and reordering favourites with Dash buttons to speeding up delivery with one-click ordering, drones, and fulfilment centre blimps, Amazon is constantly trying to optimize and automate its customer support.
Applying AI to the automation and differentiation of current processes is great, but the genius of AI is realized when you can use all the incredible insights that you’ve uncovered to recognize opportunities, validate needs, and spin off new business initiatives. This requires an advanced and experienced data practice that’s able to glean insights from multiple sources of data and formulate those insights into rich implications applicable to the industry or business at hand.
Discovering the New from the Regular at Spotify
In 2008, Spotify launched officially as a music streaming platform that allows users to search and create their own playlists via the Spotify web player. Fast forward to 2012 and Spotify’s most differentiating feature is something else entirely: serendipity. “Discover” represents highly tailored recommendations for users looking to discover new artists and playlists based on what their communities are listening to. This serendipitous discovery is made possible by the sheer power of data and the behavioural patterns that data reveals. And, as we’ve seen with the brand’s end of the year campaign, Spotify is continuously looking to apply its rich customer understanding to the improvement of its services.
Spanning cost reduction, increased efficiency, enhanced insights and customer engagement, and (new) business automation, the AI Value Pyramid is a set of guidelines for executives looking to invest in AI but unsure where to start. With an existing understanding of AI’s potential and value, along with a strong Data Science team, executives from across various industries can use these guidelines to uncover and explore opportunities for infusing AI into their own organizations. These leaders will be the ones that spearhead AI’s shift from potential to execution, educating their peers and audiences along the way.