Today is my fisrt day at the DMA, The Direct Marketing Association that holds its anual conference in Boston. I will give some feedback on the first conference I follow on Predictive Analytics, a subject that we follow very closaly at Getplus. This workshop approaches the modernization of marketing-centric technology that addresses online/offline data management, predictive and prescriptive analytics, interaction management, and orchestrating data-driven personalization to maximize marketing relevance.
Every customer journey is unique. And every touch point is an opportunity to leverage valuable data to nurture customer relationships and deepen customer intimacy.
Real-time digital content personalization is growing up analytically, and the rise of marketing solutions that enable marketers to finely personalize digital content for visitors based on array of data attributes is maturing. Phased approaches in which marketing organizations capture clickstream data to be merged with other sources of customer data for data-driven personalization; Marketing analytics teams can use this merged data to build approachable predictive models that can be deployed into real-time digital-personalization applications; Demonstrate how marketing analytics, customer/visitor experience orchestration, and data-driven personalization synchronize
Click stream data is huge
Clickstream is a data stream of hits (i.e. clicks), pages, sessions (i.e. visits), transactions, and more created by online visitors to your organization’s web properties. To deliver comprehensive customer insights, firms seek to merge clickstream data with offline channels. Web & customer analytics teams are attempting to work together, but their projects struggle to get off the ground due to a clash of approaches & culture.
The raw data you put in the system is much different from the data you get in your reports. Plus data scientists speaks its own langage (regression , neural network, machine learning) and the Web Analyst (trafic, marketing behaviour)
Normalizing data is very important
You need to have one single view of the data, therefore building a data model that can be vizualized in a Tableau or other dataviz solution. But the truth is taht you spend 90% of yout time preparing the data, and only 10% analysing the data.
90 % of your trafic is anonymous
But even with anonymous, you can make analysis. But it is much better if you can identify the customer using solution like :
- google analytics premium
- adobe analytics premium
- equanty analytics premium
Now lets have a look at data driven Personalization
True digital personalization at the individual level remains elusive for most enterprises who face challenges in data management, analytics, measurement, and execution. But you can easily do it at the Company level. Therefore AB Testing and Optimization is a core business goal of digital marketers. This insight drives strategy !
But still there is a lot of new things to reinvent with the creation of data predictive analytics. Even DMP does not have predictive analytics.
All these terms are saying the same thing : database, statistics, predictive, pattern recognition, machine learning, They all talk about Data Scientist (paid 200 000 € a year).
“A person who creates models that use predictive or prescriptive analytics, but whose primary job function is outside of the field of statistics and advanced analytics. They are "power users" who will be able to perform simple and moderately sophisticated analytic applications that would previously have required more expertise. They often reside in the lines of business and have deep domain expertise” Gartner, Inc.
But the truth is that you have to find a Unicorne, somme body who understands as well statistics as Marketing context.
Analytical Maturity Curve
The solution that will win teh market are the one that will mix Marketing value with data Viz to represent CEO, CMO Key Vizualization Indicators at the easiest level. exemple below is a dececision tree on who to convert the best trafic.
How to start ?
- Collection Engine Collects raw clickstream data, for every session and every user accessing any of the web properties of the enterprise
- Normalization Engine Transforms raw clickstream data into a normalized data model
- Analytics Engine Consists of all tools and processes used by enterprise analytics team to analyze the normalized data and build required models
- Decision Engine Uses the output models of the Analytical tools and processes to perform decision orchestration in real-time when Personalization Engine requests treatments
- Personalization Engine Presents content on the digital channel using treatments received from Decision Engine
