POST WRITTEN BYJon VeinCo-Founder & CEO of MarketShare, which helps large companies measure, predict and dramatically improve marketing’s impact on revenue 

It’s happening more and more. Big companies are using advanced analytics technology to improve marketing effectiveness and generate more revenue per ad dollar spent. As a result, they are delivering tens, and sometimes even hundreds of millions of “found” dollars to their bottom lines. Short of creating some killer new product or service, there are few if any ways a big company CEO can move the needle quite so dramatically.

But it’s not easy. Analytical approaches and effectiveness vary widely as big brands struggle to cram Big Data benefits into a revenue bottle. Scale is often an issue. Organizational adoption at companies not yet analytically oriented can also be a struggle.  Concepts such as attribution remain misunderstood. And the ability to translate hard-won insights into better decision making varies widely.

Yet the rewards of advanced analytics are undeniable. And companies getting it right are often reluctant to talk about their success because it generates a competitive advantage they want to protect. As a result, the successful CEOs are not necessarily letting other CEOs in on the secret.

There are, however, some common traits and helpful tips emerging from companies successful with advanced marketing analytics. This “cheat sheet” offers six tips that every CEO should be urging his or her marketing organization to heed:

1.  First Do No Harm: In many ways, advanced, “Big Data” marketing analytics is like proverbial “rocket science.” MarketShare, for example, has an exceptionally high ratio of PhDs on staff. But the science is not isolated. It is meant to serve the reality of today’s increasingly complex marketing ecosystem. Good math, without context and business judgment, can lead a CEO down the wrong path.  Those practicing marketing analytics should subscribe to the same oath as the medical field, “first do no harm.” Done incorrectly, fancy analytics can actually yield results that are worse than those driven by gut and intuition.  In other words, if analytical approaches or results don’t comport with your gut as a CEO, challenge them – do not assume that merely because those delivering analytical recommendations are smart or experienced, that those recommendations are correct.

For example, if you’ve launched online ad campaign, but the competition cuts prices by 40%, your actual results will not comport with the modeled predicted results unless your models account for those sorts of competitive realities.  While rich analytics can, and often do, results in “a ha” moments for CEO’s, more often than not they reinforce gut and intuition, but with far more accurate results than gut, intuition and traditional approaches alone.

2.  A forward-looking bias is essential. The sophisticated models being built today to allocate and optimize marketing resources bear little resemblance to marketing mix modeling of yesteryear. Today’s better models don’t merely analyze historical information; they uncover relationships between market factors. They account for all paid, owned and earned influences, and analyze customer behavior across a full range of both online and offline activities that ultimately lead to purchase.

Most importantly, they then transform those insights into simulations and optimizations that answer questions such as “if we do this, what happens next?”. This is only possible when the analytic models create a seamless view from past to future. And delivering these insights through software, rather than mere PowerPoint presentations, is critical.

3.  Your own data isn’t enough. You need broader industry knowledge. No company possesses all of the data it needs to make good decisions. True insights also require outside data and benchmarking knowledge derived from global experience across a broad array of industries, and the aggregated observations of other companies’ successes and failures. At MarketShare, for example, we’ve learned that the only way to get that experience is to earn it – one model at a time – across hundreds of assignments, dozens of industries, and many countries. But even that isn’t sufficient.

To build truly insightful models you need to complement your own perspective with rich data streams from the world’s leading data providers (Google GOOGL +0.7%, Twitter TWTR -1.8% and Facebook to name just a few) at scale to increase accuracy and generate better predictions, faster.

4.  You must track the entire purchase journey.  Your analytic models must recognize and account for a multitude of indirect pathways that can influence the consumer journey. For example, what causes prospective customers to visit your website? What truly influences them to buy, and to tell others about you? Today’s sophisticated models are actually ingenious systems of equations which combine dozens-to-hundreds of smaller models into a larger “intelligence system”. This takes skill and creativity, but it also takes a substantial investment in technologies to help the modeler explore the thousands of possible relationships and efficiently identify the few that make the biggest difference.

5.  Great granularity is good: Increasingly granular models and approaches can peer ever more deeply into what’s happening at product, market, channel, segment, and even individual levels. This helps you scale analytics across many decision-points, improve predictive precision and apply insights more quickly to marketing decisions. By starting at the top and drilling down, marketers can see big trends AND examine individual pathways-to-purchase. In other words, you need both “top-down” (aggregated, market level) and “bottom-up” (individual pathway) insights to achieve the best results. This is only possible, however, when you have the platforms to build and synch all those models, and the technology to quickly simulate a full spectrum of options to find the best solutions.

6.  Cross-pollinating skills help.  And just as the models matter, so do the people who build them. A-list data and modeling expertise is essential. A mix of influences works best. For example, statisticians have a way of building models that result in high degrees of confidence in the accuracy of historical observations. Econometricians build models that focus on an accurate depiction of how forces move together over time and interact with one another. Operations Research experts view modeling through the lens of forecasting and prediction. And those who have sat in leadership positions in business, finance, marketing and the like bring real-world insights that cannot be ignored.

In short, this kind of mix gives you a “right-brain/left-brain” view that more closely approximates what happens in the real world. Only when you have all of these perspectives do you get the accurate, credible, practical approach necessary to explain today’s complex ecosystem – and peer into the future.

These two article may also be of interest:  Where Sophistication Meets Scale andBeyond Marketing Mix Modeling

© 2014 MarketShare LLC. All rights reserved.