Measure Everything

Behavioral analytics is really about going back to the future, over fifty years ago merchants sold their products directly to consumers – tailors, bakers, butchers, shopkeepers all interacted with their customers at a very intimate level – enabling them to provide personalized service and to know and learn their clients tastes and preferences.  Today the Web and wireless replicate this intimate and congenial store front environment in a digital format.   

This new digital store front enables today’s marketers and enterprises to track the proclivity of clusters of consumers via behavioral analytics; it enables the creation of ‘tribes’ of consumers who tend to have similar preferences, price points and behaviors.  It enables the creation of consumer ‘buckets’ that is the clustering of their similarities via the use of machine learning algorithms.  These tribes and buckets use a combination of demographics, psychographics, and observed behaviors to classify consumers and microtarget to them at a new level of detail.  Years ago marketers used to say you are where you live; today the mantra is you are how you behave.   

However the use of such tools does not make the process of behavioral analytics an automatic system of classification. The machines are tools which need to be trained and aimed by human handlers; as such they need a strategy for accomplishing this.  The key strategy for executing and leveraging behavioral analytics for the marketer and enterprise is to plan and design a framework from which consumers’ behaviors can be captured and modeled.  Every behavior should be analyzed to determine what outcome results from them.  

Consumers should be gradually organized into buckets of unique outcomes, for example at the most basic level this can be consumers who buy vs. consumers who don’t buy.  At your website you need to know where visitors come from, what they buy, how much they spend, how long they stay, what ads they click on – measure everything – the objective is to study the patterns of consumer behaviors and consumption, in order to anticipate their appetites and preferences. 

Create an ongoing framework of processes from which buckets of consumer behaviors can be refined down to the unique service and product levels.  The strategy is to create a continuous and systemic method of quantifying consumers’ behaviors and to continuously measure everything – this can include error rates, consumer lift, total sales, conversion rates, false positives, abandoned shopping carts, etc., start small but gradually expand your efforts. 

Behavioral analytics is commonly bundled with the term data mining. But data mining is actually a marketing term coined by HNC a San Diego-based company over a decade ago to pitch their ‘Data Mining Workstation’ over time the company went into the business of detecting credit card fraud for banks via their FALCON service.  HNC eventually morphed into Fair Isaac the keeper of your FICO score and mine.  So what is the point? Data mining is a marketing slogan, what we are really talking about is the prediction of human reactions and actions by quantifying their behaviors.    

For marketers and enterprises the analysis of consumer and customer activity starts with the use of machine-learning algorithms which can generate conditional IF/THEN business rules or scorecards to quantify and monetize consumer behaviors.  These algorithms can be embedded in other operational systems, such as those that concentrate on business intelligence, web analytics, call site operations and CRM systems.  Behavioral analytics is at the core of modeling, profiling and prediction in marketing, medicine, law enforcement, counterterrorism and business intelligence.  Most importantly behavioral analytics is the future of advertising.   

Behavioral analytics can be accomplished at any consumer touch point; however enterprises must carefully plan on how they will capture and create important consumer data streams.  The problem for enterprises is that these consumer data streams from websites, call sites, processing centers, email servers and an array of other transactional legacy systems are scattered in different locations, each in unique formats performing diverse operations in a totally disjointed manner.  

There are several Internet mechanisms such as log files, cookies, beacons, forms, JavaScript, databases, email servers, call site and transactional software which can be enlisted to create consumer data streams for an enterprise.  There are also real-time demographics products – they offer lifestyle clusters segments and profiles – which can be used to enhance the quality of internal consumer data streams, in order to gain a more customer-centric view and precise ensemble of segmentation models from behavioral analytics. 

There are also ad and demographic networks and recommendation engines which an enterprise can subscribe to – in order to enhance its online presence and sales – in this instance behavioral analytics is outsourced to the network provider.  In this new behavioral analytics paradigm: data collection, integration and analysis is a seamless stream of activity taking place over a ‘reactive network’ and not a static warehouse – allowing an enterprise to customize targeted offers, incentives and advertisements – based on what it knows about each unique consumer.  

Most of today’s consumer research and find the products they need and want on the Web.  Depending on the price they either buy it online or offline.  However, purchasing a pair of shoes is different from buying a car which requires taking out for a spin and haggling about the price.  Understanding how the targeted consumers behave through these multiple channels for product or service is an important factor in developing a successful behavioral analytics strategy for marketers and enterprises.