Log In


When Predictive Models Break: A Case for Real-Time Data

Posted September 8, 2020
RealTime Analytics Post

Any time human behavior is altered by significant events, we can’t be sure exactly how much people will change their spending and which habits will win out. When three or more major events happen simultaneously, we lose the ability to predict behavior with any amount of confidence.


Relying on static data is dangerous in today’s world.



Looking back on the last few decades, we can say we enjoyed a stable market. There was the dot-com bubble in 2001 and the global financial crisis in 2008. Otherwise, innovation trends were steady, and buying cycles were predictable. 


Sales forecasting worked. Marketing saw a steady migration to digital prospecting. We knew consumers favored apparel, cosmetics, and luxury items. Investors knew what to expect in terms of growth, and business owners could rely on historical experience to make decisions.


Then 2020 happened.



Marketing departments scrambled to recover investments in major trade shows, shifted in-person events to webinars, and got creative with virtual happy hours in a desperate attempt to generate leads to backfill canceled field events. 


We’ve seen businesses slash marketing budgets and go all-in on digital, but how do we know what’s effective and what’s a waste of time as we experiment with new ways to draw prospects?


2020 Broke Predictive Models

Life has changed as we know it, and, not surprisingly, so have consumer buying habits. Looking back to prior pandemics, we could predict essentials would fly off the shelves, and that some people in the United States would balk at wearing masks. Predictively, consumer confidence dropped with the recession, and people are spending less.


What we couldn’t predict is what people would spend their money on outside of “essentials.” People are investing in home repair projects, decor, luxury home entertainment items, and swimming pool installations. Retail reported moving to online sales with an increase in business casual work shirts and sweatpants as people adjusted to working from home on webcams.



Any time human behavior is altered by significant events, we can’t be sure exactly how much people will change their spending and which habits will win out. When three or more major events happen simultaneously, we lose the ability to predict behavior with any amount of confidence.


The predictive models we relied on for budgeting, forecasts, and making major business decisions are no longer useful. Today looks nothing like yesterday, and there isn’t a seasonal trend or steady pattern we can use to make assumptions. It’s a mistake to turn to traditional forecasting to predict sales and marketing output right now.


Marketers Need Quick, Iterative Reports

Departments across organizations should no longer rely on a quarterly reporting cycle. Things are changing day-to-day. The key to success in an unpredictable market is the ability to adjust to the environment quickly. This isn’t possible without quick, iterative reporting cycles.


Historically, businesses adopted quarterly reporting cycles, and analysts developed reports to support these cycles. We had conventional, high-level metrics that inferred marketing’s impact on sales and consumer satisfaction.


While these metrics still provide value, people can no longer wait three months to make a decision on a trend that’s forming today. Monitoring data and elevating insights every few days will help business leaders guide their teams to adopt new strategies. Trending early indicators will help determine whether a new tactic is working.



Real-time data requires an investment in the right infrastructure and resources. If data isn’t normalized and aggregated, you will wear out your analyst while they jump from system to system. If data is isolated in separate systems, you can’t see the big picture. To prove the value of each channel, each tactic, and each person, marketers must invest in a data strategy.


Relying on static data is dangerous in today’s world.


Experimentation Is a Good Thing

When done the right way, experimentation can lead to greater success. Your team must define what data is necessary to develop a standard of success. This will avoid blowing out your budget on tactics that don’t work. Results must be measured throughout a campaign (not at the end of the quarter) to make the right adjustments.


Data must be properly leveraged to get the most out of your investments. A talented analyst will be your greatest asset in an unstable market. Need ideas on how to measure and evaluate your experimental campaigns? Check out our interview with Adam Smith, Digital Marketing Manager at IMPLAN.


Leadership Needs to be More Involved with Analysis

Developing reports in a vacuum is never a good thing. We see large companies moving away from leveraging IT or Business Intelligence groups because they lack the business knowledge necessary to translate findings and dig in further.


Marketing leadership has the experience needed to figure out which numbers can’t be impacted due to factors out of their control, what can be immediately changed, and what needs to be researched further. Contextual input is essential when nailing down a theory. 



Analysts don’t have time to triple-check their data sources during iterative reporting cycles. You may run into data issues and errant theories. Errors should be allowed as long as they are detected and corrected. Fail forward fast is a common development mantra, and it applies in today’s analyst world.


Marketing leadership and their analyst teams need to move away from establishing an iron-clad business case before making a change. Shift to quick, small adjustments. You’ll be able to spot patterns faster, run more profitable campaigns, and prepare for the new normal. Your business will thank you.


B2B Marketers Use Real-Time Data

Marketing leadership has been pushed–for years–to adopt reporting cadences by financially driven executive teams. The C-Suite has been the driving force behind establishing leading indicators, tying goals to business objectives, and data accountability. There’s been a disconnect between what marketers are being asked for and what they are providing, but this can’t continue.


Many marketers are moving to real-time analytics. Some do so proactively. CaliberMind customers like MobileTech Rx adopted our customer data platform because they “needed the right insights to make adjustments to their marketing campaigns not mid-quarter, not mid-half, but in some cases mid-month.” This requires a quick, iterative approach to campaign improvement and – you guessed it – real-time analytics.


The marketers who remain resistant to investing in data infrastructure and real-time analytics are going to be forced to change their ways by executive teams who know digital marketing tactics produce data that can and should be analyzed. Blindly investing in “revenue-generating” activities is no longer tolerated without demonstrable ROI. If marketers want their departments to grow, they have to prove their value.


Real-time analysis is marketing’s new normal.

View Our Other Thought Leadership