It’s hard not to desire the latest and greatest technology. The hunger is evident by the thousands that patiently stand in line to snag the newest iPhones or make a deposit for a Tesla before they’ve seen it. New is cool – it’s exciting, invigorating and carries the promise of improvement. But, when it comes to industry advancements, jumping on the hottest buzzword can be a slippery slope that distracts from real organizational issues and seldom delivers on the miraculous expectations set.
If data plays an integral role in your career, there’s no doubt you’ve been both curious and excited about the potential of predictive analytics. Leveraging historical data to forecast and anticipate the future can deliver significant savings in time, energy and resources. In fact some of these predictive elements you’re probably relying on already.
For example, if your company has calculated the lifetime value of a customer (CLTV), you’re leaning on historical data to determine how much a customer will purchase from your company over a given period. The concept itself is nothing new but the application in today’s world of data saturation adds a new element of possibility.
Companies are raking in mountains of customer data including technographic, firmographic, behavioral, demographic, transactional, product engagement and beyond! Predictive analytics offer a way to capitalize on the troves of information collected, accelerating organizational knowledge and insight, while ultimately delivering better customer experiences.
“Predictive analytics offer a way to capitalize on the troves of information collected, accelerating organizational knowledge and insight, while ultimately delivering better customer experiences”
But, as my mom would say, “don’t put the cart before the horse!” To take advantage of the potential, there are fundamental issues to address first. Here are the big ones and what you can do to prepare for embracing the power of predictive analytics in your organization.
Do you have data lakes or data buckets?
I grew up on a small farm in Southern Illinois, and the summers would bring heat waves that made being outside unbearable unless you were soaking in cold, crisp water. Unfortunately, there wasn’t anything close to resembling a pool within a reasonable radius from our home, so we had to get creative. My mother would fill extra-large buckets with water, while my siblings and I adorned ourselves with inflatables, holding as many water-resistant toys as we could carry. What followed was hours of play, standing in our individual buckets of cool water, only slightly restricted by the unnecessary inner tubes wrapped around our little bodies.
Why share this somewhat embarrassing childhood memory in a piece on predictive analytics? When I hear companies refer to having data lakes or data pools, it’s as though all their data is blending harmoniously in one place. In reality, the data breakdown of many organizations more closely resembles buckets.
“The data breakdown of many organizations more closely resembles buckets” – Click to Tweet
For example, the marketing team has a bucket with email, social media, advertising and website data. Customer Success likely fills their bucket with support, chat and ticket data. While, the product team has a bucket overflowing with in-app engagement, server, feature usage data, etc. The list goes on across departments, and often there are multiple, segregated buckets, even within individual departments.
The problem with data buckets, like the buckets my siblings and I stood in, is that they’re siloed and separated from one another. I can stand in my bucket and be entertained for a while but wouldn’t it be more fun if we were all in the same pool together? In the same light, you can learn and harness more potential from your data when departmental barriers are broken down.
Buckets breed organizational divide. A division of knowledge, information, insight and opportunity. With data compartmentalized, employees are forced to make crucial company decisions based on limited or siloed information.
“With data compartmentalized, employees are forced to make crucial company decisions based on limited or siloed information” – Click to Tweet
The power of predictive analytics rests in the ability to blend different sets of data together seamlessly. For example, at Woopra, we’ve built a predictive lead scoring model that combines demographic, technographic and behavioral data with product engagement data. The combination indicates a leads propensity to purchase with much greater certainty than traditional lead scoring models. But, if we didn’t have these data points unified within a single platform, if instead, they lived across Salesforce, Hubspot, our application and website, the model would be difficult to replicate, especially in real-time.
Consolidating your disparate data sources within a single platform will not only illuminate every touchpoint in the customer journey, allowing employees to analyze and optimize with greater certainty but will prepare you for embracing the possibilities of predictive analytics.
Aggregation and accuracy lay the foundation for assumptions
The core technique in predictive analytics is regression analysis. In its simplest form, regression analysis is a way of sorting through the potential variables that could prove or disprove an assumption or outcome.
For example, imagine you’re a marketing leader trying to determine how many qualified leads a new Twitter campaign will deliver next month. There are dozens of factors that can positively or negatively impact lead generation. Is it a summer month when prospects might be on vacation? Are you targeting a particular persona? Why type of content will be in the campaign (e.g. video, CTA, image).
If you estimate that this campaign will deliver 500 new leads next month, regression analysis allows you to sort through the correlated variables that could impact this assumption, the result of which is a predictive model.
Assumptions are the underlying factor of any predictive model. Allison Snow, Senior Analyst at Forrester, explains, “it’s key to recognize that (predictive) analytics is about probabilities, not absolutes. Unlike traditional analytics, when applying predictive analytics, one doesn’t know in advance what data is important. Predictive analytics determines what data is predictive of the outcome you wish to predict.”
“It’s key to recognize that (predictive) analytics is about probabilities, not absolutes” – Click to Tweet
The lack of accurate and comprehensive data is one of the most pervasive challenges to employing predictive analytics. In the above example, to predict how many new leads a Twitter campaign will deliver, you could look at data from previous Twitter or social media campaigns, campaigns with similar budgets, target audiences, content types, demographics, etc. The more complete the data is at your disposal, the greater your chance of producing accurate predictions.
“The lack of accurate and comprehensive data is one of the most pervasive challenges to employing predictive analytics” – Click to Tweet
So, to better prepare for the assumptions that will form the underlying pillars of predictive, put aside any bias when deciding what information to collect and what to ignore. You never know what combination of data points will create the future recipe for predictive success.
Prepping for Predictive
We’ve barely begun to tap the true potential of predictive analytics. But, tackling preexisting data consolidation, aggregation and accuracy issues will deliver immediate value to your organization and set you up for the seamless implementation of predictive models both today and in the future.
“The best way to predict the future is to create a foundation where prediction is possible.” – Click to Tweet
So, before sipping the “predictive analytics kool-aid,” kick down your data buckets, consolidate data sources and continue fueling your data pool. The best way to predict the future is to create a foundation where prediction is possible. One where, anyone from a marketer to a sales representative has the resources and infrastructure to test assumptions and adopt predictive tools to boost efficiency, productivity and customer knowledge.
This post is part of our contributor series. It is written and published independently of TNW.