Are biases still holding back marketing analytics?
Photo: Getty Images/Olivier Le Moal

Are biases still holding back marketing analytics?

Gartner predicts that by 2023, 60 percent of CMOs will slash the size of their marketing analytics department in half because of “failed promised improvements.” The “root” cause of the shortfall was found to be cognitive biases and an old-school culture.

A Gartner survey of 377 users of marketing analytics in decision making found one-third cherry-pick data to try to tell a story that aligns with their preconceived decision or opinion.

In addition, roughly a quarter said decision makers do not review the information provided by the marketing analytics team (26 percent), reject their recommendations (24 percent) or rely on gut instincts to ultimately make their choice (24 percent).

Overall, marketing analytics was found to be only responsible for influencing 53 percent of marketing decisions.

Longstanding data management challenges – including “data are inconsistent across sources” and “data are difficult to access” – were listed as the top reason analytics are not used when making decisions.

However, Gartner noted that marketing organizations regularly respond to these challenges by integrating more data or acquiring different technology seen as a “fix-all approach” to data integration — yet diminishing marginal returns are often found when pursuing a 360-degree view of the customer.

“CMOs often believe that achieving marketing data integration goals will lead to greater influence and increased value of marketing analytics,” said Joseph Enever, Sr. Director Analyst in the Gartner Marketing practice, in a statement. “The reality is that better data won’t increase marketing analytics’ decision influence alone. CMOs must address the real challenges — cognitive biases and the need for a data-informed culture.”

Organizational silos, data quality and subpar training are also often cited as the reasons for data-driven marketing underperformance.

NewVantage Partners‘ tenth annual “Data and AI Leadership Executive Survey” also found continued slow progress being made in organizations becoming data-driven. Only 26.5 percent of respondents have a data-driven organization and 19.3 percent, a data culture.

Suggestions to better establish a data culture include:

  • Hiring data and AI executives with experience in organizational change. 
  • Engaging change management experts to overcome barriers to data-driven cultures.
  • Celebrating data successes across the organization.

BrainTrust

"Truly data-driven decision-making is possible, but not without some extreme changes at the C-level."

Jeff Weidauer

President, SSR Retail LLC


"...as good as AI is at recognizing patterns in data, it falls short in answering the question of “why.” And nailing “why” is at the heart of great marketing."

Heidi Sax

Director, Growth Marketing for Wizard


"We like to pretend that data are objective, but they are really just the expression of a series of choices made by whoever develops a survey or an algorithm."

Ryan Mathews

Founder, CEO, Black Monk Consulting


Discussion Questions

DISCUSSION QUESTIONS: Do you agree that “cognitive biases and the need for a data-informed culture” are still holding back the potential of data-driven marketing? Do you agree that marketing analytics investments will or should be scaled back due to continued shortfalls?

Poll

13 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Cathy Hotka
Trusted Member
1 year ago

If personal biases are the problem, why cut back on analytics, which are the solution?

Gary Sankary
Noble Member
1 year ago

One of the benefits of Machine Learning and Artificial Intelligence is that these tools can find discrete correlations in data that would take years and luck to find using legacy methodologies – building hypotheses and proving or disproving. New data and analytics allow these tools to give insight into “what else we should be looking at.”
That said, the bias for gut feeling, especially in retail, is tough to overcome. We’ve had assortment optimization and automated planograms for years, and yet the main barrier to adoption is that buyers don’t trust the results. I firmly believe that this is changing. It has to. Retailers are looking for more ways to be competitive, and the low-hanging fruit in calculus is mainly gone. Finding new insights to attract retail customers will require more and more analytics and new and innovative types of data.

Zel Bianco
Zel Bianco
Active Member
1 year ago

It’s not one or the other. If that were the case, all advertising would be slashed from our screens as the analytics should always be the main consideration, but beliefs from experience should always be part of decision making.

Too many technology decisions are made at the top CTO levels without enough consideration given to end users and their specific use cases. We also see too many solutions that are brought in for digital transformation but are difficult to learn and use and therefore, are not adopted widely throughout the organization.

DeAnn Campbell
Active Member
1 year ago

The biggest challenge to making data informed decisions is not having visibility of where the gaps are in the data used to build a customer picture. Shopper behavior is influenced by factors beyond the shelf, but without being able to connect that customer’s movement across all of their shopping channels then the picture built is skewed by assumptions and biases. Enriching customer data with technology that connects shopper engagement across online channels and integrates that information with robust in-store analytics will help to remove assumptions and biases from the equation.

Jeff Weidauer
Jeff Weidauer
Member
1 year ago

Most organizations will tell you they make decisions based on data. But the data tells us that this isn’t true – most decisions are still made using old-fashioned gut instincts and experience. Truly data-driven decision-making is possible, but not without some extreme changes at the C-level.

Dion Kenney
1 year ago

The data seems to support the idea that biases are preventing marketing analytics from affecting decisions, ironically. Who can argue with the data? The same was once true of Wall Street traders, until the quants could demonstrate outsized returns based on sophisticated tools and algorithms. Looks like an opportunity for someone to transform the industry (and launch a career making reputation!) with some sophisticated A-B testing and a couple of big bets.

Gene Detroyer
Noble Member
1 year ago

The short answer is yes. Marketing/advertising people have been inserting their biases against data/research for as long as I can remember (50 years).

Even in the past few years, when I consulted, it was not uncommon for the company to engage researchers and ignore or even contradict the results.

My conclusion is that the research will not be accepted any time research contradicts that gut feeling that THE MAN has. “I know better.” Pure hubris.

But the good thing is that they are cutting the budgets for analytics. It sure does not make sense to spend funds on something you will not believe.

Nicola Kinsella
Active Member
1 year ago

To become a data driven organization you need to evolve your internal processes, KPIs and evaluation metrics to match. Both evaluation metrics for people performance, but also data performance. It takes time to design and roll out a program like that. While cognitive bias training should be part of the overall change management approach, at the end of the day you need to completely rethink the way you measure outcomes and the inputs to those outcomes.

Ryan Mathews
Trusted Member
1 year ago

Of course. We like to pretend that data are objective, but they are really just the expression of a series of choices made by whoever develops a survey or an algorithm. As everyone from Heisenberg to Hawthorne has found in a variety of testing from quantum physics to workplace observation, the act of measuring itself changes what is being measured. Add to that all the research on the limitations and cultural biases of standardized testing and you begin to see that data are pretty subjective — at best. But in marketing hope springs eternal, so I don’t expect to see a dramatic cutback in budgets or much of an improvement in results.

Evan Snively
Member
1 year ago

In my experience, confirmation bias is the most problematic from a “data insights” standpoint. Having a hypothesis (something to confirm) is an important part of what makes an analyst good, but being able to admit when that hypothesis is inaccurate and being able to articulate that story to a larger group is what makes an analyst great.
People want data today as a CYA, I don’t see that changing anytime soon. So when you are selling an idea to a human, appealing to their biases can be a necessary evil in order to get a green light for a project that needs a little above and beyond buy-in.

Heidi Sax
Member
1 year ago

Good marketing is a combination of having strong gut convictions and minding the data. But there’s a reason “How to Lie With Statistics” by Derrell Huff has been a go-to marketing text since the 1950s. And as good as AI is at recognizing patterns in data, it falls short in answering the question of “why.” And nailing “why” is at the heart of great marketing. For this reason, marketing analytics investments shouldn’t be the be-all, end-all and I’m comfortable scaling them back.

Craig Sundstrom
Craig Sundstrom
Noble Member
1 year ago

I don’t doubt the premise — nor do I necessarily endorse it — but without any examples, it’s hard to get beyond the “So…?”

Clay Boatright
Clay Boatright
1 year ago

“Bias.” Ooooo, scary word, let’s freak out. How about “Additional considerations based on previous experience”? I guess that’s not as salacious.

A great business philosopher once said “The greatest illusion of the information age is the belief that once a problem is identified, it is solved.” That’s because many people would rather admire a problem forever than actually solve it, because trying to fix it takes risk. Cowardice runs rampant while hidden behind the smokescreen of “we need more data.”

I’ve been an insights ad data hound for over thirty years and have learned a couple of things. 1) There’s a big difference between objective analysis and fact-based-selling. Though many C-Suiters say they want the former, but what they really want is the latter. 2) You only know what you know and the rest is filled in with intuition based on experience. Obviously, the more data and insights a decision maker has the better, but at some point there’s a diminishing point of return on the data investment and you have to make a call.

I submit virtually all decisions are based on some facts, thus making them “fact-based decisions.” However, decision makers may be exercising their privilege (another scary word) and do something the analyst disagrees with. Okay, professional disagreements happen all the time. It doesn’t mean the data is wrong or not needed, it means human beings are involved (gasp).