The death of prediction
One of the things the 2016 US Presidential election taught us was that a gap remains between data predictions and real-world performance. Columnist Joshua Reynolds believes this is a particularly important gap for marketers to understand -- and fix -- in 2017.
Prediction has always been a tricky business. Weather forecasts, airline schedules and stock tips are often the butts of jokes for the very simple reason that predicting the future is not only difficult, but most valued when the future is hardest to see.
Reliable, trustworthy predictions depend on having the right data inputs. The real trick is knowing if and when you have a blind spot in that data — which is tough, because not knowing about it is exactly what makes it a blind spot.
The same holds true for predictive analytics and their use in marketing. Our ability to predict something — like a revenue outcome or the performance of a sponsorship — improves exponentially when we’re already familiar with all of the factors surrounding it. But in most of those cases, gut instinct and personal experience will already lead us to the right answer.
The value of prediction — and forward-looking analytics, in general — increases exponentially when we’re operating in unfamiliar territory, without the benefit of an intuition informed by experience.
Human assumptions are some of the biggest impediments when working with marketing analytics — we think we know what matters, even when we don’t. What we need is a way for the data to talk to us when we’re operating in new situations, to let the data itself tell us what variables we ought to be considering.
Ultimately, this is what helps illuminate our unknown unknowns. More importantly, this is the necessary first step toward discovering game-changing strategies. Because in the end, analytics isn’t about predicting outcomes. Analytics is about informing the human decisions that change outcomes.
This is particularly important as marketers are expected to know how much money they’re generating, with whom, and why. In February of 2016, 4,683 people on LinkedIn used “Chief Revenue Officer” in their job titles. By November, a mere nine months later, that number has swelled 41 percent to 6,585 as revenue responsibilities become a daily reality for marketers. The CMO, who has a strategic seat at the table, isn’t just a “chief marketing officer” — they’re a “chief mystery officer” who can tell their CEO, CFO and board why revenue is moving the way it is, and what to do about it.
Enter explanatory analytics (a data science methodology embraced by many data scientists, including my employer, Quantifind), which explains what just happened, why, and what you can do about it.
Already we’re seeing signs that 2017 will be the year we dispense with the forecasts and embrace platforms that give marketers the real reasons behind real results in real time. And progressive marketers will look for analytics platforms that do three things:
1. Alert me when it matters
The first function of any analytics platform should be to reduce data fatigue. The data itself should trigger human curiosity at the right moment and indicate when something is worth examining.
Let’s say a national hamburger chain rolls out a limited-time offer (LTO) that is a free upgrade to garlic fries. Explanatory analytics can tell you, in real time, whether that LTO is likely to have an impact on your own brand and if this is something you should reorganize your day to address.
So the most useful analytics platforms will lead with a dashboard that tells marketers what’s happening with revenue, what’s not, and when it’s time to leap into action. After all, a strategy is not a strategy until it tells you what you’re not going to do.
2. Let me explore “why” on my own
The second function of any analytics platform should be to help you figure out why things are having an impact on your business and what to do about. Data can tell us many things, but it can never tell us, definitively, why revenues are moving up or down and how to respond.
But what an analytics platform can do is filter out, at incredible scale and speed, what definitely does not have any bearing on revenue performance. And it can serve up the most likely suspects of causation for a human to explore.
Going back to our fast food LTO analogy, while predictive analytics might tell us how much money that burger chain stands to make from its free garlic fry upgrade, explanatory analytics can tell you what’s happening and why, and inspire ideas around what to do about it.
Maybe the LTO is overperforming with teens in the Pacific Northwest because teens there love garlic and hate spending money. Being able to intuitively explore the platform to find immediately actionable insights is essential if marketers want to impact, not just track, revenue performance.
And remember, not all insights are actionable. Sometimes the data reveals operational or product-related issues that are beyond marketing’s control. It’s good to know what’s going on, but in the end, marketers also need to know what they can do to drive business in real time, not just explain it after the fact. Which leads us to…
3. Quantify marketing’s impact in real time and real dollars
2015 was the year we learned that “vanity metrics” like clicks and likes have little to no bearing on revenue performance. But 2016 was the year we discovered how hard it is to solve this problem. It’s not enough to simply measure the impact of marketing after the fact with some complicated consulting project, dig into brilliant explanations, and deliver a post-mortem on a campaign. That’s like putting lipstick on the wrong end of the pig.
Instead, what marketers need is the ability to make in-flight, revenue-driven course corrections in the middle of a campaign or sponsorship. If the analytics are presenting real-time visualizations of what people are doing — not just what they are saying or feeling — then they can be used to improve performance, not just rationalize it.
What’s more, one of the few places where predictions alone are actually still quite useful is in predicting likely outcomes if nothing changes. For marketers, when securing budget, calculating a hard ROI in advance is often challenging. But one thing marketing analytics platforms can do more effectively than ever is to help the company calculate the cost of doing nothing. This baseline is often motivation enough to pull the trigger on the next brilliant marketing maneuver.
So as 2016 draws to a close, many marketers are looking ahead to making 2017 more profitable, predictable and personally fulfilling.
Changing marketing’s overall relationship with data and asking our analytics platforms to trigger our curiosity and our intuition — and most importantly, our understanding of revenues — will be key to a happy and prosperous new year.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.