Hold on, don’t make plans to buy that millionaire island in the sun just yet. While robots can do a lot, they can’t run your business for you – at least not yet. However, artificial intelligence (AI) and predictive analytics can bring you insights, accelerate understanding, and help you get ahead of your business competitors. Sounds good, right? But perhaps you’re wondering, if artificial intelligence isn’t simply robotics, then what is it? And what is the link between AI and predictive analytics?
From Abstract AI to Actionable Business Insight
Basically, AI is the ability of a machine, including computers, to do things and get results without being specifically programmed for those things. One of the most active fields of AI currently is machine learning. This is the ability of a computer to recognize patterns in data, and highlight relationships in the data or take action according to the patterns it spots.
That’s all a bit abstract, so here’s a practical example:
Suppose you have sales data on thousands or more customers and the different products they’ve bought from you. If you look at the raw data, you see nothing but row after row of, well, “stuff.” However, a machine learning program can whisk through the data and see that a popular combination of products bought was A, B, and C, and that 65% of the customers that bought this combination also went on to buy product D.
Aha! Business insight! Now, for instance, you can go back to the 35% that didn’t buy product D and suggest that they do. For all future A-B-C combinations, you can systematically suggest they also take D.
The machine learning program might well identify other patterns and relationships as well. According to the percentages of customers who buy different combinations of products, a predictive analytics program could then pick out all the “hot combinations” for suggesting to customers that they also buy specific additional products. Give your sales system a few more instructions, and it could start making these suggestions in real time as customers are placing their orders, for instance, on your online sales website.
Predictive analytics and the artificial intelligence techniques that drive it can be applied to almost any business domain you care to think of, including repairing factory machines before they break, detecting fraud attempts, and keeping IT environments secure. Below, we’ve chosen three areas to look at more closely – sales, human resources, and supply chain analysis.
Engines for Big-Ticket and Brick-and-Mortar Sales
Although online sales companies may have pioneered the use of predictive sales analytics, other sales organizations are now getting in on the act. For instance, key account managers traveling to see customers on their premises for big-ticket items like aircraft or warehouse automation only have a small fraction of their total work time available for real selling. Predictive analytics can show them how best to allocate that selling time, using another kind of recommendation engine that says, “This other customer might also like what you already successfully sold elsewhere.”
Even brick-and-mortar retail outlets are using predictive analytics to boost sales. Coffee-house chain leader Starbucks has already announced it will deploy an AI-based recommendation system in its establishments worldwide.
A quick question for you – by how much do you think you could improve your business by using recommendation engines or similar predictive sales analytics? Research firm Aberdeen found that companies homing in on customer needs and wants through predictive analytics increased their organic revenue by 21% year-on-year, compared to an industry average of 12%.
Improving Supply Chain Performance and Customer Satisfaction
A large part of supply chain operations is forecasting. To keep customers satisfied while staying profitable, supply chains must ensure the right products are made in the right quantities and then stored in the right places, such as retailers’ shelves, distribution centers, and manufacturers’ depots.
Retail predictive analytics can help your supply chain perform better by showing where demand is likely to be strongest, and which retailers and distribution centers will need which kind of replenishment. It can also suggest which transport conditions (weather, congestion, even industrial action) will affect the optimal choice of transport modes and routes. With the right interface to your analytics, you could also get up-to-the-minute forecasts of production and shipment requirements simply by asking for them out loud to your system.
Pharmaceutics companies, for example, can use predictive analytics to refine drug sales and supply forecasts in the light of events like hospital fires or other disasters. Supermarket chains can factor in local weather patterns into analytics to predict location by location which product shipments to make. For example, more sun lotion and barbecue sets for a sunny, summer weekend in one place, more cans of soup for rainy weather in another!
Helping HR Keep Valued Employees and Improve Management
Human resource departments are aware of the cost of employee churn to their enterprise. High turnover leads to high costs to recruit replacement workers and lead time to train new employees before they can become productive.
In companies where this is a significant problem, predictive analytics can come to your aid, perhaps even coming up with some surprising insights. Using data from different sources, you may find that previous tactics such as higher pay, promotions, and higher performance ratings did not reduce churn – instead, in various combinations, such as receiving promotion without a corresponding pay raise, they increased it.
Other companies use predictive HR analytics to predict which managers will do well and which could use additional management training to head off personnel catastrophes before they happen. As a cautionary note, however, companies also make a point of using these analytics to help their HR teams make decisions, but not to take over the decision-making process. Predictive analytics in HR, as in other areas, often have the most to offer when used together with human experience and judgment.
Get Started with Predictive Analytics
Your predictive analytics will only be as good as the data you feed into them and the ability of your analytics platform to cope with the necessary volume and diversity of data.
Further statistics from Aberdeen show that companies achieving the best results with predictive analytics tend to use multiple data sources, including unstructured, text-based, machine or Internet of Things (IoT) data. The more data you can analyze and the faster you can analyze it also means the more likely you are to get viable, actionable insights in a timely way.
It’s important to look for a platform that allows you to easily bring in, clean, and mash up all kinds of large datasets for results like those described here. After all, you didn’t think we were going to tell you how great predictive analytics could be for your business without giving you a good solution for getting them, did you?
Jack Cieslak is a 10-year veteran of the tech world. He’s written for Amazon, CB Insights, and others, on topics ranging from ecommerce and VC investments to crazy product launches and top-secret startup projects.