If you are starting your journey as a marketer using big data and wondering about the specific applications of predictive analytics and how it could impact your marketing decision making process here are a few ways to think about it.
 
The key imperative of embarking on the journey into big data and predictive analytics is “actionable insights that help achieve marketing goals”.  Predictive analytics using machine learning will help you extrapolate past behaviors, reveal patterns and help predict future behaviors.
 
A. Simple Applications  
1. Lead Conversion : The life-blood of sales is lead conversions and its common sense to conclude that not all leads have equal potential to convert. Predictive analytics can help you identify from past conversions the ideal demographic/ behavioral characteristics of a strong lead vs a weak lead. Prioritising leads in this way will help sales teams create a plan for engagement based on immediacy of conversion.
Note : Weak leads might be dropping off in a consistent pattern that might require a different set of actions to be performed to help convert them into strong leads.

2. Churn Prevention : Retention is always more profitable Vs acquisition. Customer Lifetime Value is the cornerstone of consumer centric marketing. Therefore, reducing churn by determining the factors that lead to churn of specific segments of consumers and creating plans to proactively prevent churn is one of the most useful applications of predictive analytics. Likelihood to churn can be based on visible factors like complaints lodged, frequency of interaction with the product, time of year, weather, demographic shifts or by a series of complex factors and invisible factors like competitive pricing, in-market innovation in features etc.. Based on the reasons for churn we can create the right ‘nurture’ programs to address the segments. 

B. Strategic Applications

1. Identify High Value Segments : Identifying high value customer segments is important in two ways – helps you grow the profitability per customer metric improving your ROIs and it helps you understand success metrics that can be scaled across segments. Predictive analytics will help you understand the likely behaviors/ spend capacity/ spend propensity per product within the high value segments. It will also help you identify the tipping point for high value segments allowing for scaled conversions.
2. Up-sell / Cross-sell : Data of consumers who have recently traded-up and across offer insights of who/ when/ how they have done so. Consumers choosing a time of year (sale or festive season), usage of loyalty benefits, increased usage frequency etc will point to certain trigger points that lead to their change in behavior. This could come from a change in lifestyle, lifestage, season, income shifts etc.. Matching internal consumer data with external 3rd party data might be necessary to provide a clear picture of when/ how they have traded-up/ across. 

 
There are more than 4 applications for predictive analytics (Continuous Multi-Variate Content Testing, Post Trade Programmatic Audit etc) but the list will help start the right practices to help ensure early wins within the orgainsation.

The real utility of big data comes to life when we are using it to analyse hidden patterns, analyse correlation and predict the possibility of future occurrences. This has direct implications to how marketing can use big data as a powerful tool to enhance the consumer experience, drive efficiency in spends and improve their effectiveness.
 
The promise of Machine Learning is to help duplicate the human experience of learning by extrapolating from past behaviors to predict future behaviors. The difference will be the scale/ speed / accuracy and consistency – all of which will improve exponentially. Machine learning, because it works towards determining patterns (Vs pre-set algorithms that are set to solve a specific problem logically) is especially useful in ‘real-time’ scenarios (the better way to talk ‘real-time’ is actually ‘right time’ given that business cycles vary depending on category) where the time available to discover the pattern and attempt predictions is extremely limited. It works best when you want to understand what happened in the last week and what is likely to occur in the next week.
 
Before talking any further, I’d like to step back and underline the key imperative of big data – actionable insights. Insights that are actionable within the business cycle of the predictions. Insights that are deterministic, help generate clear conclusions or hypotheses and will be accepted by the broader organisation as action-worthy.
 
Here are some pointers for marketers towards helping big data predictions :

1. Start small : Building a robust big data model and making it stable (wrt data feeds/ predictions) is part art, part science. It will take a few experts within the organisation (Data Science/ Analytics/ Category Experts/ Sales) to provide insights to help stabilise the model at accuracy. Learning is messy and the initial parts of gaining momentum can be confusing. You don’t want large teams to be consumed by the learning curve.
2. Point North : Are the predictions aimed at helping achieve the top business goals ? Identify the prediction results to the goals, have a clear template for analysis (combining the prediction with human expertise to creating solutions), the process (frequency, time to market, responsibility, budgets).
3. Right Time Vs Real Time : Every business cycle varies and has a different rhythm of going to market. Organisations have unique challenges in decision-making, approvals processes and executing investments/ making changes.
4. Right Data Vs Big Data : In finding the signal you need to help eliminate “noise”.  Big data does not mean all data. This is where the internal expertise within the company can truly add value in helping teams make decisions to ignore certain datasets for specific predictions.

 
Predictive analytics for marketing decisions is an emerging art, enabled by science. While its engine is driven by the past the true potential lies in taking the leap from predictions to imagine new possibilities.

90% of the Big Data that is being generated everyday is not in the form of numbers but words (unstructured data). What people write-click-share-view-upload everyday can reveal real insights about their motivations that can help marketers build relevant and conversion oriented experiences. Specifically emails, reviews, requests, complaints, FAQs from consumers can reveal insights @ scale that can be used by marketers to observe patterns, capture trends and identify high risk events.
 
The process of applying machine learning and natural language processing techniques to text data is complex. Not surprisingly, given that language is about concepts-meaning-inferences-nuance and specificity the process is part science and part art.
 
Lets look at two steps within the entire modeling process that has direct linkage to marketing  (Choosing a modeling technique to run your data, Data Transformation, Test-Training Data etc. are crucial parts of the process) :
 
Step 1 : Feature Extraction
Extraction of meaningful phrases within text, entity extraction (person, places, products and organisations). There are a wide range commercial text analytics tools that help you do this. Some of them are : IBM Watson Alchemy API, Lexalytics, Microsoft Text Azure Analytics API.
 
Step 2 : Codification
After the extraction of key phrases and entities is done the next step is to create tokens (phrases/ entity that will signal a key event has occurred) is basic first step. This step requires an in-depth understanding of category and consumer to help capture the signal Vs the noise. Whats meaningful and actionable for one category may be less so for another. Red flags for a skin care brand could be words like (allergy) while (call drop offs) would be more worrisome for a telco provider.
 
Example A : For an Insurance provider focussed on building a reputation for simple and easy approvals, analysis of consumer reviews-FAQs-call center transcripts might indicate that consumers are dropping-outs because of the confusing application form that they need to fill in. In this case positive/ negative phrases related to application-form filling etc become meaningful.
 
Token Example : Creating a token with negative phrases linked to application become the trigger for a company to respond immediately and appropriately to rectify the situation.
 
Detecting the signals Vs the noise for your brand in ways that are actionable will allow you to effectively improve your brand goals in immediate and measurable ways. Lastly, continuous human intervention is needed as patterns change over time, new vocabulary and contexts emerge.

“Big data is about finding correlation, while small data is about finding causation” – Martin Lindstorm.
 
Marketing is obsessed with harnessing the power of big data – rightly so. Big data is extremely valuable in understanding patterns at scale, helping brands take advantage of trends and for marketers to drive better ROI. But this laudable effort should be done in conjunction with the reverance for ‘small data’.
 
Small data and the role of insights in marketing is different from that of big data and provides valuable inputs into marketing that big data, by definition, cannot. 

1. “Why” is the single biggest question that big data cannot answer and requires small data. For example "Churn Prevention" is one of the most common applications of big data. Predicting patterns of behavior of customers who have churned helps create 'flags' that signal when the behaviours occur for other customers. But this does not solve for why churn may be occurring in the first place. To solve for this we need to understand the context of the churn behaviour. If the behaviour pre-churn is identified at scale is a complaint registered then the nature of the complaints might point to a solution to prevent churn. If the pre churn behaviour is isolated to certain segments/ life-stages then it might make sense to determine if the organisation is primed to serve customers when they move life-stages.   
2. Big Ideas come from Small Data : Observing people using your product, talking to them about your product and how it intersects with their lives can help you discover new ideas for your brand that when scaled can become an enduring asset. 
3. Small Data Feeds Big Data : Sometimes micro observations are important to understand tipping points. A high abandonment rate on a website can be captured, analysed ad infinitum by big data. But a survey conducted specific to understanding the reasons and nuances of abandonment can be more revealing and effective. Built over a series of such events small data improvements/ hypotheses can be scaled into big data observations and rules.

 
Small data and big data are both valuable but different. Using both intelligently Vs choosing one should be the way to go. Small data should feed valuable hypotheses/ ideas and inspire while big data should discern patterns @ scale and inform / prioritise.​