Some form of Artificial Intelligence (AI) project can now be found in most companies, with the remainder seeming to at least have several projects in a planning stage. If Marc Andreessen’s now famous “software will eat the world” Wall Street Journal essay was on its way to being true when he wrote it seven years ago, the “eats the world” statement is equally true today regarding AI.
AI, or at least something claiming to be AI, can be seen throughout the consumer and business landscape. Are the recommendation engines that power Amazon and Netflix suggestions AI? They seem to be. In Netflix’s case, they made a move a few years ago to have their “you might like” recommendation based on analysis of content you’ve liked, instead of only on what you’ve watched. It’s up to the Netflix subscribers to determine if this made their recommendations sharper. On the Amazon front, in 2016, the company open sourced the AI framework of its recommendation system through a project named DSSTNE.
While there is machine learning involved in these recommendation engines, because they are so ubiquitous we rarely think about what is behind them. To most people, they are recommendation engines and not much thought goes into dissecting exactly how they work. This is the normal cycle for AI’s application into an existing technology. Often AI is most valuable when it is an ingredient, enhancing existing products and technologies.
To this point, as the current wave of AI-related innovation has increased, not surprisingly, the first applications have been related to new products or product enhancements. AI assistants, chatbots, and AI’s contributions to autonomous cars may be the most discussed. Investment in AI technologies supports this focus on innovation with the Brookings Institute reporting that investment in self-driving cars alone equaled $80 billion in the 2014-2017 timeframe, with even more substantial investment expected in 2018.
Business strategists can tell you AI has been sending waves of change throughout entire organizations for decades, and by extension, throughout our lives. For example, predictive analytics companies have benefited greatly from machine learning to assist organizations in a broad array of areas from credit scoring and fraud detection to determining which leads in a database are most likely to provide higher lifetime value.
Innovation fueled by AI is nothing new. AI projects have been active for decades, officially beginning at Dartmouth College in 1956. It’s not that they’ve been hiding away in academia all this time. It’s a common practice that as AI projects mature, they are folded into technology. The recommendation engine examples show this. Following this reasoning, it’s likely that in the near future, we’re likely not to hear when AI is present in a technology. For example, today, you are likely to read how AI enables advanced fraud detection or improved chatbots. In the not too distant future, it will just be assumed that anytime predictive analytics and proactive personalization is needed the best of breed will include AI.
AI has proven effective in breaking technology barriers. What would have seemed impossible even five years ago is now a reality. Take the example of product and content personalization. Not too long ago, digital personalization was only possible on the internet, where tracking past visits and purchases were used to deliver more personalized experiences. If the user of a mobile or IoT device desired a “personalized” experience, they needed to set their preference themselves, a task that even with the best of intentions often went incomplete. Jump to today, the ability to understand what end users are doing in the physical world throughout their day is now used to proactively predict the most appropriate experience to deliver to the end user. The results are increased usage, decreased churn, and higher conversion rates. This illustrates the next stage for AI – KPI multiplier.
As the inclusion of AI becomes more prevalent, how it is seen in organizations has evolved once more. Yes, it will continue to be a driving force of innovation, but more is being asked of it. This is a good thing for both the development and business sides of the organization. A project without a clear connection a company’s strategic vision or existing core business is in danger of being a distraction inside and outside the organization.
When measuring how project efforts match with stated results, KPIs are one of the best tools. Well-defined corporate KPIs with department and individual KPIs rolling up to them can keep a company: focused on its quarterly numbers; its product roadmap on track; or a long-term strategic initiative moving forward. So how can AI be assimilated into the KPI process?
There are two approaches to leverage AI to improve KPIs – making the most of what you have and developing with purpose.
The first is utilizing existing AI projects. We’ve all seen cool technology that in hindsight begs the question: “but how can we use it?” Begin with an innovation audit. Take inventory of what is being developed or has been developed. What does this enable? And, how does it tie in with the company’s current KPIs and long-term goals? This will likely require someone other than the developer or project manager to help with this assessment. I suggest a face-to-face or web conference to discuss the capabilities and goals of the project. This is a positive communication about how the project can be leveraged. Communication is critical. You are looking for existing and new ways to leverage innovation, not to assess the value an individual brings to the organization.
The second approach for leveraging AI to improve KPIs starts at the planning stage. Many companies have an understanding they should embrace AI projects, but they wonder with which ones to start. Begin by first looking at your company’s key KPIs, particularly your company’s revenue-related KPIs. Now, ask: What can AI do to improve these KPIs? Looking at the real-world predictions example above, there are several KPIs that may be addressed. Increased time using an app supports a revenue KPI if the app developer receives revenue from in-app purchases. Delivering content that is more relevant to the user based on granular personas supports a conversion rate KPI, which is often used to measure users moving from freemium to premium subscriptions. And, reaching a user at the right moment in their day makes overall usage more likely and reduces the chances the user will churn, which obviously is a closely watched revenue impacting KPI.
For organization’s who look at AI as a vital part of their products and how their business operates, it is likely to be not only a KPI multiplier, but soon a revenue multiplier.
