The whole aspect of customer segmentation is to know the customer at a segment level number one which means the organization first needs to go through segmenting customer segmenting as a process right so what do I mean by customer segmenting could be customers who are your online customers your brick-and-mortar customers that could be one form of segmenting the other form of segmenting could be customers who are loyal to you customers who are big spenders customers who are who don't spend many customers who are likely to you know to churn outright.

This kind of segmentation is also possible the challenges while many organizations have some form of segmentation though in the olden days you know this segmentation was always business rules-driven right it was not data rules-driven it was business rules-driven so as a result you know organizations were not able to effectively continually segmenting different types of customers in a better manner okay so customer segmentation was there but was very very very costly earlier okay so so what was needed was a better way of segmenting customers and that's where machine learning and AI came in place right the moment ml and AI started a platform started you know making the bus technology platforms around ml.

AI also started really picking up speed, as a result, a complex process like customer segmentation was effectively done time and again not just one time it is not a one-time investment, but it was more like a daily investment or a weekly investment right if an organization wants to do customer segmentation every week now the technology infrastructure and the platform becomes much easier to do it okay so how was this how was it done earlier or probably even today in many organizations it is all based on past business rules, and you know the segmentation focuses mainly on what the customer has spent not on how many times the customer has come or what kind of engagement that the customer has had and more importantly you know it was driven by this management principle rate eighty percent of your revenue comes from 20% of your customers, so it was predominantly based on this concept where you know it is all the revenue driven not engagement driven.

But with ml with predictive analytics coming in play, there was an opportunity to really understand the engagement level of customers so how do you understand the engagement level of customers how many times does a customer visit your online website or your shop right, and how recent the customer has been right and more importantly what kind of questions that the customer is asking right have there been complaints that customer is raising continually all this kind of structured unstructured and semi-structured data has now seen is now effectively being processed because of the advent of machine learning and obviously advent, of all the satellite technology solutions associated with machine learning right like Big Data solutions so with all of this what has been really phenomenal with ml, is I don't just get to run and understand my customer segments once, but I get to understand as a retail company owner I get to understand it.

Whenever I likerright, andd I based on based on what my segmentation are it allows me to strategize targeted marketing campaigns what do I mean is for for me if I know there is a particular customer segment who are big spenders, and they are not they are not engaged well-meaning the customer comes in once in a year spends a lot of money, and then he is he is absconding for the next one-year right so this is a this is a segment of customers I would like to have a different marketing campaign encourage them to engage well with the organization differently so this kind of targeted marketing campaign is something that was that has become much more possible you know with the help of machine learning right and what is more important is the way machine learning works is it looks at all the patterns you are your supervised learning algorithm your classification algorithms is able to easily classify different segments of customers number one, and it keeps on classifying meaning the algorithm learns.

How many other times you run it if it learns and there are you know there are cases today where organizations are running segmentation in real-time right what do I mean by real-time it is not a batch process in the background it is some it's more like a real-time process so that is the difference between an online ml process and an offline ml process so um so that is becoming very very possible with machine learning and predictive analytics why so well that said what is also possible with good customer segmentation is the organization has now a goldmine of information about existing customers that it can apply to improve multiple things number one as I said targeted marketing number two cross-selling and upselling of products right any customer that has probably been buying one set of products how do you cross-sell and upsell a different kind of products to the customer a recommendation of customers at the segment level becomes much easier with the help of customer segmentation.

So there are ever so many use cases that are possible more importantly you know domains in like banking where you know b2b domains like forget about b2c domains but DB to be domains where you know there is the business to business customers' ability to understand the existing customer segmentation and apply that to your leads and prospects that pattern to your leads and prospects has really improved the rate of new customer acquisition with companies right so there's a lot more use cases of ml and machine learning and predictive analytics I've just taken customer segmentation as one use case okay there are many other use cases that are possible a customer recommendation product Association caught analytics wishlist analytics search analytics you know click-through prediction there are so many use cases that retail organizations are now applying machine learning across the board ok.

So now what we will do is more around customer segmentation right try to look at a use case around customer segmentation and customer segmentation can be done in many ways not just through business rules but also through a certain analysis type called RFM analysis recency frequency and monetary analysis which many of you might have already known but more importantly, whereas segmentation is helpful is you know how to use this data which is something as sim as simple as recency frequency and monetary and how to use this data, and you know and really make and really understand your customers better right so that is where the biggest benefit of true customer segmentation so what we are going to see is a good use case of how to use recency frequency and monetary and how to segment the customers better using that and how to run machine learning using RFM right so this is the use case that we'll be seeing today okay.

Let me go back to my code right, and I'll cover this quickly so think about this as for every machine learning project or a data science project there are a few phases or few steps that are needed right number one is what is called Gaeta ingestion where you download the data that you are going to analyze right so that is number one number two would be data preprocessing where you whatever data that you have whatever raw data that you have what kind of process you need to apply on that raw data right the process could be in our case you know first of all removing bad data right that would be the case the second one is understanding.

Post a Comment

Previous Post Next Post