When we talk about statistics, we talk about probability. Given the situation that Student A is taking engineering classes, what is the likelihood that she will earn an honorary degree? What is the correlation between social norms and absenteeism? If a construction worker thinks his peers are slacking off, how likely he will be absent from his work? The idea here is to predict what will happen in the future based on the data we collect from a sample. We start off with forming a hypothesis, researching existing studies and results, making our own surveys or experimentation, then moving on to interpreting data and concluding our analysis.
Perhaps the most familiar predictive analytics we experience is performed by Google. When we use Google products, such as Google Search, Gmail, and Google Docs, we often see Smart Compose where the computer tries to complete our sentences before we even hit enter. Aside from that, predictive analytics are useful in different areas, such as fraud and risk, as well as marketing and operations.
What the Tech Giants are doing
If you haven’t noticed it yet, Microsoft, Google, and Amazon are all using predictive analytics to their advantage. Because they have accumulated a lot of data, they can optimize the information to future insights. You may also notice how when you use Waze or Google Maps, it often asks if you prefer a faster route than the ones with a higher amount of traffic. This fact alone indicates that the two apps have predictive analytics. Google also has experimented in machine learning where computers can learn a picture and identify the object as a cat. It took them, 16,000 computer processors, to achieve this! It takes a lot to train computers to identify an object from a picture and guess correctly that it is a cat.
In terms of market leaders, Forrester has named IBM and SAS as the two leaders in the Q3 year 2018. IBM in particular promotes SPSS software to calculate data analytics and IBM Watson to apply the knowledge into Artificial Intelligence. Revelwood, a client of IBM used such tools to uncover sales trends in hundreds of thousands of products and tens of thousands of SKUs.
Harvard Business Review covered the many reasons why predictive analytics are useful to businesses. It can predict income and help us adjust our pricing model. It can also predict maintenance time and cost. Moreover, it can go even further with AI and machine learning: it can change the way things are done in an improved way (for example, in diagnosing and treating patients).
Applications in different industries
There are many industries that are affected by predictive analytics: finance, healthcare, video streaming, e-commerce, social network, retail, and travel. We see Robo Advisor in both Wealthsimple, Wealthfront, and Robinhood. As market transactions occur in seconds (or even nanoseconds), it is hard to keep track of what to trade or invest. Thus, Robo Investor acts as the fund optimizer that can keep up with the speed that customers need. It considers your capital size, what other investors are doing, trending news about companies, and stock prices at that moment. Currently, it is used in ETFs and yet to be personalized for customers.
In healthcare, we see Fitbit and Apple Health that try to compile customer’s data. Fitbit has been selected as the covered fitness benefit in 27 US states for forty-two Medicare Advantage plans. Users can benefit from access to health data when they visit doctors. Back then, people have to go through medical check-ups to know how healthy he is. But now with fitness trackers, patients have better access to their health data.
Meanwhile, travel, retail, e-commerce industries are promoting their businesses through strategic sale events. With predictive analytics, you understand when your business thrives and when the sales are low (for example, Christmas might be the best time in business because people are buying gifts). Some e-commerces like Alibaba and Tokopedia use this opportunity to name their own sales day: Single Day, Online Sales Day, Black Friday, Cyber Monday.
Social media like Facebook, Instagram, and Pinterest use customer’s behavior to show relevant Ads. Businesses pay these social media for an Ad placement and then calculate their pay-per-click. It takes some time for the user to familiarize themselves with the Ads and when the need is there, they click on the Ads.
We also see in Biotech that research has been done to re-engineer the structure of the protein as a recovery drug. Numerate for example uses AI to predict how a potential drug will behave both in the lab and body. Predictive analytics can help us discover new medicines and not only that, to reduce errors in predicting outcomes.
Finally, there is also a personalization algorithm that uses predictive analytics. When you use Netflix, you might see “Based on your previous watch” and get some recommendations. As the computer accumulates data about you, they recognize a pattern of behavior. Combined with the behaviors of other like-minded customers, the personalization algorithm will predict the next products you most likely will choose.
Why invest in predictive analytics?
The question then is how we will use innovation in technology in our product and services. Michael Schrage, a research fellow at MIT, wrote on Harvard Business Review how AI can be an assistant, a guide, a consultant, a colleague, and even a boss.
Predictive analytics definitely can be an assistant, a guide, or a consultant for you because they can perform repetitive tasks. That way, we can focus on the critical thinking side: how to improve the business, how to include design thinking, etc. To give you an example, Tesla was automating the car factory using a fluff bot. However, it soon realized that robots are very bad at picking up fluff. They discovered that installing fiberglass didn’t improve reducing noise in the car and so they made the decision to eliminate fiberglass installation altogether. Here, it shows that humans can focus on decision making and critical thinking.
Your company can either be in the Launch, Mature, or Growth stage. If you are at the latter stages, then you already compile a humongous amount of data of your customer behavior. But if you don’t then you can look at industry standards or use information from the Point-of-Sale (POS) payment platform, such as Stripe or Square.
At the core, predictive analytics will give you insights using data that they analyze. IBM tried to predict the probability of white spots being cancer on an X-ray. Google Maps that can predict time to a destination based on Geo-location and the route selection we choose. IFTTT offers automation where smart lights can be turned on once you’re at home.
The intersection of design and predictive analytics
You must notice from your own experience that visualization is the key to deliver insights from data. In 2020, John Hopkins University provides real-time data about COVID-19 using data visualization. Infogram and Datawrapper allow us to create engaging infographics, charts, and tables for social media sharing.
Designing a building or a city can benefit from AI. Predictive analytics help calculate the making a structure and the materials needed. Retail homes also can utilize Smart Home technology, just as Smart Cities would. Architects then can focus on the design and presentation part.
Creatives such as Gal Shir, Fabrizio, Jack, and even Adobe use AI to help designers pick the right combination of colors (or color palette) for their arts. A way to use the platform is to upload a picture and the program automatically selects a color palette that will work in harmony.
Nir Eyal, author of Hooked: How to Build Habit-Forming Products, has a theory that the most successful products shape our habits and that we’re hooked because of the “rewards” given, such as Likes, Follows, etc. He mentioned that businesses should identify who their followers are, what is their current habit, and what’s the common path in habit testing. Habit testing is one of the ways to use predictive analytics to better businesses.
In UX Design it’s all about interaction design and user experience. An interaction by definition means a compilation of different sets of actions that will result in different results. Based on historical data, we can learn more about the customer through patterns of their behavior. It can be as easy as a preference of what a customer would do to join a mailing list. Would a Pop-up be more effective? Or a side thumbnail works best? Maybe some engaging news (such as Free PDF) inside a blog post would work better? Predictive analytics help because it can give insights about this behavior. There is also a tendency for people to go to “Start Here” or “For You” when they first arrive at a blog website. When it’s about mobile apps, it can be about utilizing buttons and shortcuts: which ones do customers press? Which ones do they neglect? If people are not using particular functions, then maybe you want to take it out.
Predictive analytics in an economic downturn
While the economic downturn would pressure business owners to cut costs for efficiency, we need to also think that technology helps improve our products and services. We can create tools to help predict what the customers would do if we give incentives. In other words, predictive analytics would help us to guide customers to buy our products and services.
Internally, we can also use predictive analytics in automation. The automotive industry, for example, is using robots for car manufacturing. The operation manager then can focus on discussing strategies, design of the assembly line, and ultimately making better decisions with the help of technology.
Keep in mind that with more advanced technology, there will be faster computers with more accuracy and cheaper unit cost. Supercomputers for example had a cost of one Giga FLOPS for 19 billion dollars that later reduced to 3 cents only in 2017.