It’s been our great pleasure to speak with hundreds of business leaders in every industry about predictive analytics in 2017. And in the early part of 2018, we’ve spoken to dozens more.

What Did We Find?

They want to improve revenue and profitability, reduce cost and risk, and ultimately compete better in the digital world. But they’re using predictive analytics in very different ways to achieve those goals. Everything from reducing supply chain delays, equipment downtime, and customer churn, to improving campaign conversion, customer value, and beyond.

We also found that the problems they run into when trying to leverage predictive analytics are surprisingly similar. In fact, 7 real-world problems keep coming up in conversation after conversation.

In order of importance, here are the top struggles that business leaders are having with predictive analytics in 2018 and what they want to see in the coming year.

#1: The Remedy Problem

When you are given a prediction — for instance, a high-value customer is going to churn next week — it’s not always clear what to do to fix the predicted problem before it occurs. And if you can’t fix a problem, knowing about it ahead of time is hardly helpful.

Business leaders want every prediction to come with a recommended action or “remedy” so their workforce can take steps to effectively prevent the prediction from coming true. Learn more about the remedy problem.

#2: The Dirty Data Problem

Everyone’s data is imperfect with problems like missing values, redundant data, and more. Cleansing or scrubbing dirty, messy, imperfect data so it can be analyzed grinds the whole prediction process to a halt.

What business leaders want is some tolerance for imperfection. Software that accepts their data as-is, so they can minimize the data cleansing needed to simply get their prediction project off the ground. Read more about the dirty data problem.

#3: The Data Skills Problem

It’s difficult and expensive to attract and retain the data talent you need to analyze data, and build and update predictive models.

Businesses want to rely less on in-demand data experts. They want to automate analysis and prediction so they can move much faster and operate much more cost-effectively.

#4: The Explainability Problem

Predictions, and the remedies for them, are black boxes. There’s no rationale explaining why a prediction was made or a remedy was prescribed, and it’s hard to have confidence in something that’s so opaque. What’s more, predictions and remedies are often formulated and presented in a way that’s hard for humans to understand.

Business leaders want predictions and remedies — and all insight from data —  to be transparent and understandable so they can confidently take action.

#5: The Wasted Data Problem

Businesses have a lot of useful data stored in data warehouses, databases, enterprise applications, and more. But the diversity of the data (diverse types, formats, locations) makes it hard for predictive analytics tools to consume. As a result, most predictive models are built on a much smaller, easier to consume subset of all available data. And that limits the quality of the predictions generated and their impact to the business.

Because businesses invest a lot in the collection and storage of data, they want to tap into its full potential, squeezing out every possible drop of value. That means having predictive analytics technology that can analyze and predict on very diverse data sets.

#6: The Integration Problem

Many analytics tools are standalone apps…self-contained apps that don’t integrate with the everyday business applications enterprises are already using as part of their normal workflow. Because they don’t integrate, prediction lives in a silo separate from the work its meant to influence. And any attempts to join prediction to the work at-hand are clunky and inefficient at best.

Business leaders want analytics to be like an add-on feature, embedded in their existing applications like CRM, supply chain planning, marketing automation, and field service management.

#7: The Speed & Scale Problem

Most companies have a good amount of data, but they aren’t truly dealing with the “big data” challenges some companies face. Some larger enterprises have terabytes of new data streaming in at blazing fast speeds. Traditional data analysis processes and tools can’t keep up with this volume or velocity.

Businesses want a predictive analytics solution that can handle their data – no matter the speed or volume. So companies that have big data now, or that grow into big data in the future, don’t have to worry their prediction technology will break down when it’s needed most.

So What?

Business leaders have spoken. And we hear them loud and clear. There’s clearly a big gap between what business want from predictive analytics and what’s available to them. And Emcien is prepared to step in and give it to them in 2018, through enhancements to our own software tailor-made to address the biggest pain points in predictive analytics today. Stay tuned for future product announcements!

 

Emily Gay

Marketing Director

Emily helps companies understand how new data technologies can solve their biggest challenges. In-house and agency-side, she's spent nearly a decade helping brands use data to make smarter decisions and optimize KPIs.

How is automation transforming analysis? Download Redefining Analytics to find out.