Join DMA

Your Diamond in the Rough: Making Marketing Analytics Shine


Post Date: November 4, 2013
By: Stephanie Miller

The problem of why analytics is not more widespread is not about knowledge, willingness or lack of proof of value, it’s because of organizational ignorance of how to use analytics in a meaningful and business changing manner.  The opportunity created by that void was the basis of a recent DMA webinar, “Diamond in the Rough:  Making Analytics Shine in Your Marketing Organization,“ the inaugural event in the 2013-14 DMA Analytics Council presents series.

“Operating from your gut is useful and helpful but not the way to run your business today,” says webinar co-presenter Leo Kluger, Program Director, Business Analytics Transformation, IBM.  “We always start with discovery – identifying the key strategic objectives. This may sound obvious, but it takes some clear thinking and these conversations take longer than you think.

“We ask, ‘What sort of problems do you have?’‘How can we trace that back to the strategic objectives?’‘What is at the core of what is really going on?’,” he said.  “Unless you answer a business objective that is real, then the analytics work doesn’t get used.”

Marketing analytics can transform your marketing effectiveness.  Citing an MIT study, Leo reported that a majority of respondents say that analytics and information (including big data) creates a competitive advantage within their market or industry.  That is a 70% increase since 2010.  Organizations already active in big data activities were 15% more likely to report a competitive advantage.

ROI from analytics projects can include improved revenue, reduced costs, increased productivity and efficiency,  or improved customer satisfaction, Leo said.  “Get very quantified results – something that  increases the win rate, or identifies the clients who are most likely to renew,” he advised.  When you impose this kind of rigor in the beginning of the process, people who are not committed will drop out, he said.

Be realistic, as you won’t move from neophyte to expert in one project, he said. You will never have perfect data.

“Most companies are drowning in data, but starved for insights,” he said.  Big change requires a powerful spark – which may not be a big project, he said.  “It needs to be a big idea.  You need insights on the causes and changes that will drive business.”

At the end of your assessment work, “Drop a big document on their desk.  Of course only the first five pages matter.  But we have to align the organization to make sure they are ready to ACT once you do the work,” he advised.  If your goal is to transform a business, recognize that organizations are complex.

“These are all organizational issues, not analytics skills issues,” Leo said.  So a commitment to the process and the right insights is within everyone’s grasp.

Analytics work engages both the “artist and scientist” in your organization, agreed co-presenter Devyani Sadh, Ph.D., CEO of Data Square, a marketing analytics consultancy, who spoke about methods to use analytics effectively in modern, complex marketing organizations.  Simple analytics are descriptive (reports) but advanced analytics will include predictive (modeling, simulation, forecasting) and prescriptive (optimization methods).   She started the discussion by introducing the notion of the “Base Analytics Toolkit” which is comprised of Segmentation, Classification Trees and Predictive Modeling. “Segmentation is often overlooked and considered simple, but I would say it’s one of the most important and “complex” analytic techniques because it requires not just statistical but also marketing expertise;  thus, a true marriage between art and science,” she said.  “Employing both sides of the brain is harder to manage than doing advanced work in just one discipline.”

Classification trees, on the other hand are simpler, and the most commonly used data mining technique.  Take a group defined by some a priori attribute (e.g.: transaction) and divide it based on various criteria (e.g.: age, gender, time of transaction).  You can get some really good insights just by doing simple things.”

Modern analytics is built on a process-people-technique-technology continuum, she said.  She outlined a basic three–step approach to all analytics:

  1. Gather and prepare the data using rigorous exploratory data analysis
  2. Select the appropriate analytic technique and perform the analysis
  3. Evaluate results via a validation sample and continually refine accuracy.

Beyond the base kit that nearly every analytics team can use, there are a lot of advanced methods. Some of the techniques Devyani reviewed in the webinar include:

  • Classification and regression trees
  • CHAID & CART
  • Discriminant Analysis
  • Cluster Analysis / Factor Analysis
  • Linear and Logistic Regression
  • Regression: Binary, Multinomial, Logit, Probit, Preference
  • Structural Equation Modeling
  • Non-linear and Non-parametric
  • Experimental Design
  • Survey Design and Analysis
  • Survival Analysis and Time Series
  • Multivariate Adaptive Regression Splines
  • Bayesian Methods / Networks
  • Collaborative Filtering
  • Markov Chains /Monte Carlo Simulations
  • Multidimensional Scaling
  • Perceptual Mapping
  • Conjoint Analysis
  • Optimization
  • Artificial Intelligence
  • Neural Networks, Fuzzy Logic
  • Genetic Algorithms
  • Radial basis functions
  • Support vector machines
  • Text Mining, Sentiment Analysis

Devyani then outlined a predictive analytics roadmap for the average organization broken out into five progressive analytic Tiers. Typically, Tier 1 marketing analytics is about “right person identification,” Devyani explained. “Following that, right message, offer, and timing optimization could be conducted in Tiers 2 and 3 with an integrated analytic contact strategy forming the basis for Tier 4. Finally, new datasets and advanced technology could become the hallmark of Tier 5 which directly addresses big data and real-time analytics.

Multi-step sales processes involve multiple customer behaviors each of which can have an impact on profitability, she noted.  Tier 2 and Tier 3 analytics can help identify customers that rank high based on joint outcomes.  Once you identify the key business drivers and patterns that engage audiences, you will generate impressive results.  For example, Staples and Fidelity report doubling their performance by focusing on offer timing whereas Schwab and Wells Fargo show significant returns by focusing on relevancy driven by trigger-based marketing and lifecycle messaging, she said.

“Tier 4 Analytics can help create relevance by optimizing the offer, frequency, timing and sequence of communications across channels and campaigns for each customer across their lifetime,” she said.  She also discussed how the marketing mix itself can be informed by analytics investment, which ensures optimal spend as well as reach.

Tier 5 analytics – Social data – and all “big data” are adding exciting new elements to marketing analytics, she said.  “With machine learning tools, we can keep up in real time with consumer activity across channels – the amount of data that is being processed is extraordinary, and the information gleaned can be acted upon quickly.”

“Don’t be discouraged if things seem complicated and you don’t find yourself engaging in Tier 5 “big data” analytics right away,” she said. A significant amount of insights and ROI can be realized just from the Base Analytics Tool kit and Tier 1 aalytics. Use this demonstrable ROI to create the basis and business case justification for investing in the higher Tiers.

##  ##  ##

Editor’s Note:  This event is part of a six-webinar program produced by the DMA Analytics Council.  View the entire schedule and register now.   Your membership may include participation in this or other member communities.  Please ask your account manager or contact Member Services.

Share Now: Tweet about this on TwitterShare on LinkedInShare on Facebook

0 Comments

Leave A Reply