Masterclass Marketing and Retail Analytics | Great Learning



Asia yeah hi guys so we will be starting with the master class right now a small introduction about me my name is shreya Chakravarthy I look after the academic operations and part of the academic operations team and with us today we have mr. Himanshu Monroe so a little bit in robot Himanshu before he takes the floor mr. münch is currently and will estas watson a leading global risk adviser insurance and reinsurance broker he's parodying the market and climb insights global function in Mumbai his efforts to increase sales marketing client management and service operations effectiveness have been supporting businesses through a robust business research engine his industry expertise sales intelligence customer relationship management Net Promoter Score revenue synergies and a lot and a knowledge management initiatives for a holistic view around actionable climb insights so yes so now mr. Himanshu I request you to take the floor and start with the webinar the flow is all yours thank you very much Ron that was really helpful thanks for that wonderful introduction hello everyone good morning and hope you are having a great start to the weekend so let's get on with the the master class for marketing and retail analytics and before we get into this a brief about the structure of the session yeah so the way we will go we have about one hour for today's session the way this would be structured would be neatly into two halves so we are at about 11 8 11 10 right now so let's count one our from this time onwards we'll break it today's session into two halves so while the first one would be about running through or glancing through various marketing and retail terminologies terms techniques frameworks the various models that you would encounter in this journey right and will break open for few questions which you can type on chat box and we let's see you know how many of these questions can we pick up within this session itself which so we'll break open for about five minutes of questions and answers at the midpoint and then the next thirty minutes would be about the real crux of it as we say in analytics right so let's talk about a few models and techniques and their real-world applications how is this really useful how can we really apply some of these learnings right so that would be the next thirty minutes and of course towards the end will once again resolve sometime about five minutes for questions and let's see how many of them can be take right so that's quickly the agenda the structure of today's session and before we get in a quick introspective question to all of you right so why why are we doing today's session and why marketing and retail analytics per se the reason being if you look at analytics or marketing related analytics its retail which tends to have a lots of data around various customer touch points retail seems to be one of the sector's which is practicing the actual applications of analytics to the maximum and we will see some of these applications which we may or may not have realized while we get in touch with these interactions right we don't maybe understand that there is analytics at the back of it so we check a few of these interactions of course some other questions that we could really ask and today's session is our analytic applications Pete retail or marketing are the sector agnostic I belong to a b2b sector I belong to a b2c sector or the kind of businesses b2b or b2c does it really makes a difference of course you know so it's it's about a bird's eye view of the techniques and models a top view and then the bottoms up approach as well so some real-life cases and how do we actually solve them in our businesses yeah so that's the structure of today's session let's go and as I mentioned right the first thing itself before getting into anything that the why question what's the ultimate goal of analytics why are we doing this there could be lots of answers right solving a business problem increasing profitability your top lines your bottom lines making a decision lots of things that we could come up with in terms of our answers why do we want to do analytics but if you look at it a bit holistically the crux is what this gentleman once mentioned famous for his quotes Steve Jobs and apples right ultimately the real answer is your customer experience or CX as we say it nowadays right you can come up with a lot of models and techniques and frameworks and technologies right but the real question to be asked is why are we doing this are we doing this technology for the sake of implementing an analytic model or a technique or is a customer reason at the end of it right so you need to start off with the customer experience beat at an end-to-end level or look at it through various touch points are we making a difference and that's the way you need to then come up with analytic techniques and technology that should aid your customer experience journey yeah and why do we speak about analytics applications in retail some real issues real business issues in the real retail sector the questions that come up for marketeers for business leaders is you know how do you place the merchandising aspect various products and categories in your story it doesn't matter you know what's the size of the store but placement plays a big role and the customer psyche do you have to walk to the length of the store to pick up milk and you would have observed this right milk and meat products are placed at the end of the aisle at the end of the store and then there are Quick Pick products like your chewing gums your blades your toothbrushes or various other quick picks or tow fees or mints at the counter itself there is a psychology to this there is analytics driving these decisions which products would sell most at what times how do you replenish your store what's the inventory you want to keep is their analytics defining your supply chain management are there products that really go together right and does it make sense to keep them together that's you know that's that's what we call as the clustering or market basket analysis and retail and we will touch on this topic right in today's session so does it make a difference and keeping various products that tend to go together together in the store itself does the location of your store make a difference does the ethnic mix where it is situated the kind of population that makes a difference to your product mix to the way you market and to the way you place your products and our customers in your locality or in your region vfm customers so we FM value-for-money are the customers more of premium products buyers are they more into health segments are they value for many customers so it's more of need based segmentation of your customers and their their needs and requirements to drive your store related strategy so at a micro level your store related strategy and at a macro level what could be the supply chain management mix driving the retail the large retail chains and stores yeah you can easily categorize all these requirements in a three by three grid or matrix yeah you can look at all your requirements and strategies through the lens of are they operational operational improvements are they tactical are they in the short term or are these decisions more strategic and analytics can play play a critical role in all these viewpoints in terms of business strategies yeah so that's the first lens operational tactical strategic or it could also be the next part of the lens the other part of it is what happened in the past what's happening currently how is it happening so which is the lag effect or leading effect of your decisions are you analyzing data which happened which occurred and then taking decisions for the future or are you also playing the predictive game looking at the various patterns and trends and forecasting the future yeah so the what will happen part of it so you will see a lot of operational decisions your inventory your demand your shelf availability the measurement of your KPIs or success business metrics your order status your promotions that are happening right now are all operational kind of methodologies that can be driven by analytic based decisions you have your more tactical approaches and more strategic approaches as you will see towards the right of this slide your customer behavior analysis your scenario planning you know if X happens if this event happens if it's it's a rainy day what would my stroll look like right it's it's a particular festive season how should I stop my products and which products based on the kind of festival that's approaching you can look at causations what is causing what kind of uptick from customers and and correlations yeah so a good difference between causation and correlation so things that can be highly correlated need not be causing each other yeah and vice-versa you can evaluate your promotions at a tactical level you can never evaluate the effectiveness of your larger advertising campaigns at a more strategically level yeah you can look at your promotion costs your marketing costs and of course you can you should be looking at your store level profitability as well so it could be a large brand like say a big bazaar but you would also have your store level analysis around what's the profitability of a store litter situated in a suburb or a store situated in say South Mumbai so as we discussed retail allows us to look at a 360-degree view of your customer the various touch points through which we are getting reams and reams of customer data yeah and there is the way you can look at your data or describe data as you will see in this particular slide a script of data which tells you about the features and characteristics of your customers your their demographics their age their gender their income patterns where they are situated what's their life stage are they single are they are the married are they double income no kids that kind of life stage so that's all about your descriptive data a factual data it's it is what it is you can't change it right you then have your behavioral data which is based on customers buying patterns orders transactions the way they pay the way they shop the way they explode their entire usage history which is mood behavior driven yeah and it's not necessary that customers with the particular demographic will behave the same way as well and that's the reason the very basic fundamental of analytics why do we do analytics is because your behavior may not be caused by your descriptive or demographic parameters you have your interaction data where customers are providing you with some footprints they are leaving you with some transactional level interaction level data so it could be through social media through your chat transcripts emails call center nodes the way people have been looking at your websites or the way people have been looking at your stores so it it can be both digital as well as physical experiences of your customers that's the interaction data and finally you have attitudinal data so it's not so there is a fine difference between behavioral data is what people have done which has happened in the past attitudinal data is more about what people can do what your customers are disposed towards which can be collected more through your service or the way people have been reacting on perhaps social media this is what we could do right so those are signs of what could happen in the future opinions preferences survey results as you will see social network data right ultimately the the goal is about increasing predictability you you do all the interactions that you look at or or analysis that you do through this data it could be descriptive diagnostic but it could also be simulative and predictive to understand who what how why to understand the deeper questions about your customers and your needs and their needs yeah so this is the entire retail analytics canvas we won't of course read to all the things that have been mentioned here but in a nutshell as you would see as this is where analytics can really play a key role in store operations we spoke about Market Basket analysis we will be covering it today how do you men maintain your merchandise is your marketing analysis the marketing mix the way you promote advertise through various mediums the way you manage your categories within a retail store and of course the the bottom part of it right which is the loss prevention which could happen through various events now let's talk about customer loyalty so when we mention customer loyalty in retail or for that matter in any segment or any industries that you guys would be belonging to I'm sure you know that's a big big metric that we try to gauge that we try to measure monitor and improve which is your customer stickiness your customer satisfaction customer loyalty various terms that we can use for this and that's what you try to do in retail as well you tend to look at your customers through various lens and understand the levels of loyalty the levels of their attraction to your brand their stickiness to your brand are they visiting your stores or brands because they have no other options because you happen to be within their vicinity the closest what if they've also competitor coming up very soon what are their chances of attrition of going to the production brand so how fickle your customers are are they really no matter what you know absolutely stuck to your brand or products or do they tend to waver a bit experiment a bit or are they actually looking out big time based on their purchase patterns based on their opinions and that's what you can we can do through various data mining techniques the product lust ring trying to identify various segments through various perspectives and eventually identifying your best customers and why do we need to understand our best customers is because it's a typical parrot or 80/20 rule you don't want to have equivalent promotions or time spent on all your customers you would rather devote your energies your promotions your spins or even your TV ads the way use reach out to your customers to your customers who are most likely to purchase from you and at the same time most likely to give you higher yields or returns in terms of each rupee or dollar spend on your marketing's or promotions yeah again through a customer journey in retail and I am sure you know I would just want you to try to understand it from your own sectors and industries as well what is a customer transaction profile where all are we receiving customer data and what all can be the source of our analytics it could be purchase data we mentioned that earlier custom behavior it could be loyalty data the number of times you have made a purchase one very important aspect of customer loyalty which we will again talk about in this session is the is the recency frequency monetary model the RFM model in retail yeah so which is about how recent have customers shopped in your store how frequent they have been so they could have recently purchased say yesterday but are they purchasing just once a month or are they coming on all months in a year or are they coming on particularly Paige so that's the frequency so you know recently how recent they are how frequent they are and the biggest part of it the monetary aspect so they could be visiting you regularly they could be visiting you off late as well but are they really giving you big-ticket purchases or are they just purchasing maybe breads and eggs out of your store so it's really important to know what our customers purchasing so that's the RFM model that tells you about various segments of your customers which are your really high-value customers and which are not so high valued customers then of course we have the demographic data factual which it is which you can get through various means you can get it through surveys you can get it through customer transactions and of course you know marketers do try to keep this data as life as possible that's more about the KYC right age income light stage we mentioned gender occupation education lifestyle interest so there could be particular customers who are more interested in more sports part of the equation so so purchasing more sporty products or health products or organic foods it's really essential for you to know are you situated in an area which has customer preference preferences leaning towards these kind of segments and then of course the opinion or attitudinal data which can be gathered through service it could be more micro or tactical in like you know how likely are you to product purchase this product which is about to be launched or what would be your reactions to this price differences in a customer at what price point are you most likely to purchase this product and you know your likelihood to shop in some other soul which we are planning to open up in a new location we spoke about Market Basket analysis and this is what a typical Market Basket analysis would tend to answer as you know so there are there is this interesting Walmart study in us where it's known that beer is kept next to baby diapers in various stores and there is it's not based on Vincent fancies of stores managed store managers but because there is an and detailed analytical exercise at the at the end of it which determines that diapers and beer tend to sell very well together there is a high correlation and and there is an understanding as well so could be young fathers who have just dropped by to purchase diapers and then they happen to take along beer cans as well and there is drawn analytical and behavioural data to prove it and that's why it makes sense to keep beers and diapers together in your store so that you have higher level of customer uptake so that's what Market Basket is all about you know so within your customer basket what are the products that are most likely to go together sell well together and in that case why not keep them close together in your store yeah various things that can come out of this bananas go very well with conflicts this is again coming from a Walmart sturdy flashlights go very well with Halloween costumes think about it make sense yeah bug spray goes well with hunting gear so you know products that go well together sell together retail analytics so we spoke about customer segments so your various models or as we call them as propensity models disposition models would allow you to define your customer segments and not just on the basis of customers income their demographics basic demographic based segmentation I would say is a thing of the past when you are just looking at your customers as high revenue low revenue they're incompatible with low level income could be buying really premium products say premium Mobile's premium technology products so you need to define your customer segmentations based on their actual purchase history and based on their preferences opinions and attitudes and of course their life stage where they are at yeah so you could have new earning members just join the workforce starting to on and they are highly oriented towards high tech products and that's the way marketers would tend to define their customer segments fresh foodies natural organic product oriented customers you have high fliers they have less time but more disposable money they would come to your shop less often but their ticket price would be much higher so every time they come over they tend to purchase a lot or stock for the month and then of course you have your typical budget driven customers which are more around value so even a one rupee increase in a particular product say bread or eggs or milk could take them away from not purchasing so that could be how sensitive they are to various price points so what would you do to get them through the door keep the the basic some of the prices of basic products at reasonable levels at accurate levels and then try to increase the customer profitability by selling them other products based on their needs but being very very sensitive about typical products where they they can be whimsical about the price points that's the way you would define your customer segments that's the way a typical analytical model would look this is a cross segment financial stats we won't get once again you know to the the depth of this but that's the way you would look you know if it was 100 percent of your customers how many are the percentage of trips net dollars so these are various segments that you are seeing at that the columns wick fixers urban seekers marketeers have their way of giving various names to their segments and you can look at the differences through this basic cross tabulation of data that's the way households in this particular examples are profiled on their shopping behavior and you can then look at the various differences in shopping patterns another part we spoke about is product clusters that's also defined by Market Basket analysis so you know you can look at these various clusters which may not be the same product category but they happen to sell well Gowell together you have your trip segmentation yeah so which is so your purchases may not just be defined by all the parameters we discussed earlier you could be having different reasons and different needs at different types of times of the day that drives your purchase pattern yeah your routine replenishment which could happen say once a week could be about various grocery products you could make quick trips over the weekend or perhaps during during evening times early morning times for stocking up on beers and cigarettes and then of course you you would have your typical breakfast kind of trips where you would definitely not want to go into a large retail store a large supermarket or a mall but you would want to go into your local keanu store around for making your breakfast trip your cereals your your coffee and that's the way you know your your way strips are determind store segmentation carrying on from the last slide so you would want to go to various stores for various reasons and I'm sure all of us are aware of the specialty store called de Catalan right the sports specialties – how far are you willing to travel to the Catalan stores I think there are just two or three of them in Mumbai so I'm sure customers would be willing to travel say 20 odd kilometres because they want to go to a specific store for a specific reason but would you be willing to go that far for your your big bazaars or D much of the world perhaps not yeah so that's interesting to know and key to know your store profile the size of your stores address office or the vicinity or locality you are based in the customer data that you have been collecting around your area there attitudinal and survey data the competition around your your locality right so you could be a Demark but you also happen to have have hyper City or big bazaar or lots of competition around your buzzing locality that makes a difference as well yeah cus your retailers are very very interested to know about your catchment area analysis which is as we mentioned how far can your customers travel for this particular store so would it be a radius of say one-mile three-mile 5-mile or are people really willing to travel to this speciality store for about say 20 miles and that determines where should you be present or opening up your new store based on these kind of analysis and market understanding we have promotion analysis which is if you come up with a particular direct mailer yeah Derek Miller's are when you send out specific pamphlets or brochures do your customers it could be physical it could be through emails right and each one of them is is a marketing spend right in terms of actual marketing spend the postage spend the printing spend the the spends that your marketing team is there are spending on that campaign in putting up that campaign together so that's why it needs to be it can't be everything for everybody you need to know who is the customer segment that I am targeting and what are the incremental gains that I am getting out of this particular campaign or direct mailer so that's another typical analysis around your Direct Mail costs and as you know it's the law of diminishing returns so tell a point you send it to these many percentage of your customers your margin would increase but then at a particular point it would start plateauing so you need to know what's that plateau point well beyond this point it's a law of diminishing return so this is the maximum number of customers that I need to send my particular direct mailer – that's your promotion analysis by a-1 pet what point can you get expect getting incremental gains personalization another great facet of analytics and artificial intelligence of course you would see this in a lot of your ecommerce brands a thing that really comes up is you know customers who bought this product also brought this kind of products that's driven by a really high level of analytics and artificial intelligence in the background which tells you about customers actual behavior their shopping behavior their their patterns and which emerges as a particular manual or model for other possible customers potential customers why don't you buy these things as well so you know how do you want to make references to your customers why you have bought this how about checking on these kind of products as well now that's the customer loyalty part which is your high-value customers your medium medium buy to customers and your low value customers this is based on customers Spence who are most profitable who are least profitable and that's the way you would see the various shopping cards that are offered to you can you become a platinum member if you spend this much or gold member that's all driven by estimating the lifetime value of your customers through these kind of models we spoke about the RFM model so on the extreme left you will see the frequency how often have you been buying then this next table is the monetary value how what's the average ticket price of each of those visits and then you have the next four columns which are about recency have you been shopping in the last two months three to four months five to six months over six months if you try to break your customers through this basic with the recency frequency monetary kind of model this is what you can come up with which is you know you break up through recency lens your customers into two segments customers who have been visiting you for at least once in the last three to four months or they have not visited you more than the last five months so that makes them more recent and not so recent or perhaps lapse customers then you break them through the second criteria which is of frequency and the third one is monitoring so and that's what will enable you to define this six segments customers who have been very recent what but purchasing low value and low frequencies your first time customers they might have just happened to pass through your stores customers who have been buying low and with low frequency so we look at the other part of it are your low value customers that you should definitely ignore and not spend any of your marketing spends on the real important segments are the ones at the bottom so you're early repeat customers they have been coming often they have been making decent purchases their frequency has been fine you need to really look at ways to upsell to these customers who are your early repeat customers the other part of it the corollary is your defectors they had been buying quite a bit they were visiting your stores quite often but they have not visited you for the last five to six months which are your laughs customers they have just gone out of the window you need to look at various analysis to see what was the reason which drew them away from your brand and at the bottom of it are your high-value customers they are decent they are buying a lot they are coming more and more often your biggest marketing attention should be to this your high-value customers and then of course you have your core defectors they had been buying a lot they were really frequent but wow you know they have not come to you for a for a long period so your defectors and core defectors is which what needs to be analyzed and your early repeat customers and high-value customers are what needs to be reached out much more through upsell and cross-sell yeah that's the way you would look at your RFM analysis an action that you can take to your RFM is incentivize and preserve the ones who are devoted the ones who are buying a lot and visiting a lot potential people who can buy much more you know expand and nurture reach out put them more often through various quick fixes or discounts or promotions and then you have a borderline or lapse cases you need to require them or at times you need to ignore them as well why – so your customer lifestyle attributes also leads to various kind of segmentation this is this particular part is self-explanatory so various products some are low price some are the positioning is fresh for some the positioning is you know it it caters to a particular taste but like you would see in the extreme right every sambar masala its branded at South Indian its quick to make ready-to-eat kind of products so what is your positioning that's the way your stores would define various customer positioning and then try to eat them together so is it you know the kind of few scenes is it the kind of brands is it exotic so these are your customer positioning elements that play a large role in defining whom do you want to reach out to yeah you also make your lifestyle or customer persona based on these kind of positioning elements the attributes in your basket the kind of products that you are buying tells a lot about you as a customer segment and some other useful models and in retail we come to the first part of it you know as I as I mentioned we take a break here we have other useful models in this in retail your segmentation based CRM based and consumer insight piece so these are some models for you to think about reflect and introspect that you could do in retail for sure but you know can you do it in your industries are some of these models sector agnostic yeah why not so as it goes you note for the end of this the interim period in our presentation is don't find customers for your products find products for your customer so let's take a break for questions so Sri on to you do we have any questions so far right so far we still getting questions so we do not have any questions as of now so as I said the Ashmont you said the floor is open for questions so you are please do ask away right so the there is one question so the question is if there are a group of medium value customers and a small group of high value customers contributing equally to the top line and contributing equally to the scales of a particular company then what we analytic are good for addressing loyalty yeah so good question so I think this is what we are looking at right so you're early repeat customers and high value customers medium value and high value customers and what will analytic tell you right analytic will tell us focus on the monetary part yeah so you should devote a large chunk even if they are right now giving you top lines which are equal think about it right you are devoting lesser time to people who are giving you big-ticket returns the high value customers you know they're the in terms of the value right it could be equal to your top line but imagine these are the customers who are giving you more revenue per customer so you would really like to focus more on these customers reach out to them through more promotions more coupons at the same time you of course won't want to ignore the other segment as well you would try to upsell to them through more personalization based analytics as how about buying these products which could go well within your clusters so it's managing both those segments but of course having a preference towards the one which are giving you more monetary returns even if from a top line perspective they are equal right right thank you thank you much for that answer one more question do the interesting question so how what if the product is a one-time buy like we all know that there are some premium products in the market or some product which is just a one-time buy so how can analytics have been identifying the customer base of who to target yeah exactly so that's where you would look at customers life stage so I understand you know your biggest your white white house or white products like your refrigerators you could have a bycicle which are your television which are your one-time purchase that's where what's really important is to understand the life stage of a customer so which life stage is he he or she had what are they potentially looking out to buy think about it you know somebody who has bought a big-ticket item like say a house or changed a property you would get this data from various sources now people who have purchase a property would definitely be looking at purchasing your white goods and that's what analytics can help you understand you know if you are around these real estate stores or locations now customers will be going for white goods and I need to start reaching out to them with much more targeted campaigns yeah right a one more question before we move on to the next part of the session so here the question is how analytics will be used to identify like for if let's say the new products introduced in my is like you know in a in a particular market in a particular supermarket let's say so how do we not which customers to target like how analytics will be used or like how analytics is used in this scenario that's that's a very good primary research question and that's why you know you see there are products launched in the the market on the floor right but they don't come up or they do they are not air dropped there is a lot of survey and attitudinal data and opinion based primary research that has gone into the background before products are launched on on Vega Street in various retail stores that you see and that's where primary research comes handy so I'm sure you must have seen a lot of these exit surveys or product samples menu visits various stores is it's just to understand your preference just to understand your propensity there are a lot of product based analysis or it's search which happens here is the product which we are planning to launch how likely are you to buy it and what is the price that you are expecting for this particular product and that's where the price elasticity is understood as well at this point the product would be optimally priced at this point it would be too premium and too expensive I won't even look at it and the other part of it at this point this would be so cheaply priced that I would start doubting its quality so these are all the pre research or surveys that happen before you see new products being launched right right right so I think that's the number of questions like I think we should move on to the next part or next segment of this session and all the one just small note to all the participants please do ask questions keep on posting questions over here on the box we will again get back to it at the end of the session please do leave but please do ask so that we have make more questions over here over to you module Thank You Ralph so as we mentioned we look at some success models some success stories that we have seen various retail stores or retailers beat in India or beat examples from abroad or outside India is these are some stories that we see based on the models and techniques that we have spoken about the first one being understanding your shoppers to increase their share of wallet now that's a phrase which you will keep hearing a lot when you are deep into analytics is what's my customers share of wallet yeah so share of wallet is could be your customers disposable income you need to know for a customer how much is he his overall spends are looking like and within those overall spends what is it that I'm gaining a share of his wallet yup so that's a situation where a premium retailer wanted to understand my shoppers how much are they buying from me and how much are they buying from my competitors to understand the competitive differentiation so this is what you will see if you look at the chart on the right so this is a typical spider chart and at the bottom this chart represents your four peers peer one peer to to peer four and the retailer which is you yourself so how are you differentiated with your competitors in on these four parameters a total spends be International spends total number of customers and online spend so you can guess you know based on this this is a typical travel portal something like say I make my trip which has done this analysis now this analysis helps you understand look at Pier one the online spends the number of customers I mean it's really really significantly higher than anybody else on that entire group right and this then gives you that trigger point to ask other questions why does this fear of mine has so many customers so much online spend so many number of transactions that's a share of wallet analysis yeah you can look at similar kind of projects now in closer to retail environment as well if you look at retailer one retailer two retailer three now these are various parameters which are important for retailers inside store and outside store as well you know which is your brand brand imagery as well so you would look at some of those parameters your corporate values your customer complaints so this is more towards the servicing part or the image of your brand and then towards the right of the spyder chart you will see more in-store related parameters navigation store environment is it's talking of unique innovative products quality of products price and value so this data you can again get through primary research or you can look at it through transactional level data as well where you are actually looking at the bills of customers and understanding this amongst various retailers so there are primary research agencies who could do these kind of service for you and of course they would keep the names of some of the competitors more like a syndicated research they would keep it confidential but there is a possibility to gain this kind of data which then tells you quite pointedly how is my total shopping experience looking like and which are the parameters that I need to focus on yeah how would you increase customer stickiness loyalty and profitability by measuring customer satisfaction so here's let's look at this situation you know a leading specialty apparel retailer is under a lot of economic pressure the sales are down it now wants to understand what's my customer value and customer values what we explained earlier right your high value customers of your your early adapters your defectors and you want to look at their shopping behavior to retain the best and increase stability in spends of others so if you look at the sample output towards your right these are your various way that you have defined your customers high value high frequency Konami spenders the the ways that you have defined these segments and then you are looking at their interactions yeah-ah at so how would a hundred percent Danny up within these customers on a particular period and if you look at the unprofitable net migration part of it so these are your high value low frequency customers which have been shifting over so it is looking at the shifting of these segments yeah so how is the customers so your customers are not static we need to understand that a customer who is high value ones could not remain your high value customer for forever so how are they actually shifting so that tells you a lot about the pressures that your customers are facing and what could you do to to salvage the situation now we would want to focus the next few minutes on two of these models so one is your market mix modeling right so this is an example of your marketing Spence and now the kind of techniques that you would use to determine this would be your typical regression models we won't of course go through the the jargons or the techniques or the statistical heavy loaded part of it but this is what it is right so if your dependent variable if the problem is my sales volumes I need to look at my sales volumes and the impact of a recently launched campaign or an advertisement or a promotion on my sales volume right now in today's times of integrated marketing campaigns right you would run your campaigns on various mass media be it on TV print you would look at your in-store promotions your various promotions that are running inside the store to talk about that campaign you would look at sampling related spins where you are offering a few free samples to your customers and then there are other factors right which determines your sales which are more external in nature so your distribution your competition your supply chain and that's what would define you know a dependent variable is saints what is the contribution of all these other independent variables which is in totality called as my marketing mix and how much is the load or importance that that each of this factor is giving to my sales so that's regression techniques they allow you to come up with these kind of models they would give you the load point of each of those attributes and then let's look at it this way right you need to understand as a retailer as as as a marketeer is your base what are your base factors and what are your intramental factors now incremental factors are factors which are you know you you tweak a bit of those factors and you can see your sales rising so imagine you know if you were to raise your TV spends your TV advertisement spends by a particular amount that should lead to a rise in sales yeah so those are your incremental factors and your base factors are something like say your competition you have no say in determining your competition you have to live with it so those are your base factors we have which you have to accept as a businessperson as a retailer as a marketeer and your incremental factors are which you can tweak increase decrease status quo so that it makes a difference to your sales volumes yep so that's the basic differentiation now let's take this quick sighs price what is price you know is it peace or incremental it's base yeah so your price of the product is your price of the product your TV radio radio these are the factors which you can tweak we mentioned right so these are your incremental marketing elements competition base you can't change that category volume sales again base you can't influence a category if a category is selling you know if SEO is the size of a sports utility vehicle in India is this much then that's the category you have to play with in that category your print again that's incremental that's external your trade percentage volumes based on feature display incremental you can go to the trade stores and increase your point of displays your point of sale display in retail stores you have the possibility circulations discounts coupons redemptions incremental seasonality index you have to live with it it's base right your your retail stores tend to do well on festivals on weekends that's the nature of the game internet operations clicks again incremental you can influence that coupons it's the same thing which is incremental shelf talk in-store promotions in incremental so it's essential to look at all your marketing elements are they base or are they incremental can you make a change consumer price index that's the way the economy is behaving you have to play within the economy you can't determine that right an interesting model which is around your brand loyalty we have been speaking about brand loyalty imagine you know if a new Cola wants to come up and it needs to understand the customers fidelity or how likely are they to switch over from their existing brands so you imagine you know if this was a Pepsi brand a and if this was a Coca Cola brand B so now this chart tells you about how loyal are my brand a customers and brand be customers so what you will see is percentage of buyers on your y-axis and their exclusivity levels as a changes from left to right from brand a to brand B right so if there were 100 percent of customers as you will see in this break up about less than 38 30 percent so something about say let's say 28 percent of them are exclusive to brand a so in their shopping analysis over a period of time they have only purchased so about 20% of your customers have stuck to grand a no matter what similarly on the extreme right you will see a much lesser let's say about 10% of the are your brand B exclusive loyal customers so this chart tells us pretty fairly that brand a customers are much more loyal or are they much more stuck to their brand so that's good news for brand e but that's for another potential Cola brand coming in they have to fight a pick a battle with brand a as compared to bad brand P in terms of loyalty and that's the way you would see the loyalty dipping from left to right so of course you would see much bigger bars bars towards the left because those are your customers now let's look at the second bar so they have bought brand a much lesser number of time yeah now this is your Market Basket analysis now that's a very interesting thing of course we are trying to give you a very top level view of how this analysis would look like but imagine if you have zillions of data points right and in retail how would this analysis look it's real big data yeah but here is it in a very very easy to understand manner the Market Basket analysis imagine you have your five customers walking through your your store and this is what their basket looks like in terms of products that they have bought and this is what we will try to lies in a market basket analysis when you want to look at which products to keep close together and which products can be paired with each other so when you see a lot of these schemes right so you buy orange juice you will get soda free or vice versa that is not based on some random marketing decision it is driven by analytics now a better way to understand look at these four customers their baskets look like customer one shoes socks tie belt customer two shoe socks tie belt short hat now these are your four customers and the four transactions that they have made in your store now what is market basket analysis right for this four for this particular example let's consider socks and time how often would you see socks and tie going together so if you look at all these four transactions in within this form how often do you see socks and tied together in a basket it's in transaction one and transaction 2 so that's two out of four baskets have seen socks and tie going together now this is called as your support in in terms of statistical way that you will look at this data is support a support means the frequency within the total number of transaction where you would see this particular correlation happening so two out of four times you would see socks and tie going together now the second part of it which is the real market basket bit is how often when you have purchased a tie are you likely to purchase socks as well a pair of socks right you will see you have purchased I thrice transaction one two and three now within this three transaction you have purchased a pair of socks just twice which is in transaction one and transaction 2 now that is your confidence level so your association rules are inherently two-tiered one is support the next is confidence my a confidence level of socks and tie going together is two out of three which is 66.67% now that tells you write a high percentage 50% of my customer baskets have both these products and when people buy Thai at least 2/3 of those times they do buy a pair of socks as well so it does make sense there is a good amount of statistical significance attached to this that you keep your socks and tie together in your store now if you look at this it might look particular a bit complicated at the start of it but there is a very simple way of looking at this chart so let's start from extreme left so the earlier example that we considered was about 4 transactions but obviously your stores have much larger number of transactions so you have imagine in your top left database D which is about thousand transactions in your store if you look at tid which means number of transactions that's transaction ID so hundred plus 200 plus 300 plus 400 so these are thousand transactions now we are looking at a data of thousand transactions items are your products within your basket so so that's the way the baskets are 100 of baskets or transactions have items 1 3 4 200 235 and likewise now that's your existing data set you would scan this data and you would come up with the support the support thing that we spoke about earlier the number of time you would see various transactions right so inherently you will see there are about 5 different products 1 2 3 4 5 in these baskets how often have you seen product one you would see product one in the tid items mentioned one hundred and three hundred that's the time you have seen product one so that's where the support is – now let's look at product – how often is it mentioned it's mentioned in 200 300 and 400 so it's mentioned twice likewise you will do it for all the five products and then the next part which is l1 you will eliminate the one which is off minimum support or one which has minimum presence so you can fairly understand product for which is present in just the tid hundred those items which means it has minimal presences in your shopping basket you will take it off this is you know doing a market basket analysis through the law of elimination that's what will happen to your analytical model in the background we are just trying to open up that black box and see how does this model work now the second part is if you look at the remaining products one two three and five how often are they mentioned two three three three right when they are paired and when we say paired is when is so it's one is mentioned now in these products twice twice rice rice you look at item set c2 so now within these products one two three and five what are the pair's that we can think of so there are six pairs 1 2 1 3 1 5 2 3 2 5 & 3 5 how often are these spray products pair scanned B we come to 1 2 is present let's look at that data right database D and compare it 1 2 is present just one in PID three hundreds one freeze present twice in tid hundred and tid three hundred so these are the strength of those pairs once again you will eliminate the pairs which are east of Lee strength so you take off one and two you take off one and five and you are remaining with these four pairs 1 3 2 3 2 5 3 5 of them the strongest is 2 5 which is practically mentioned products 2 & 5 are are present in baskets 200 300 and 400 now finally you will look at triplets fight and the triplet is 2 3 & 5 which is present in baskets title tid 200 and title 300 now that's your basket basket analysis products 2 3 & 5 tend to go together twice you will see they are 200 300 they are present in 500 out of thousand baskets now that's huge probability right 50% of your baskets do can contain 2 3 5 all these three products it makes sense to keep them together or market them together right so we come to the end of today's session we have maintained the timeline within an are some phrases that you should be using and that you will come across or I would say you know you should start getting attune to as you pursue this analytics journey is it's okay to not know everything right there are lots of techniques and frameworks and industries and sectors in an analytical journey you couple that with things like artificial intelligence machine learning deep learning there is lots going around in this area so it's ok to not know everything pick up your strengths pick up your areas you want to target and pursue them we mentioned this at the start of the session so you know does it make a difference at I if I belong to a b2b industry or b2c kind of industry no because in the at the end of it everything is edge to edge which is human-to-human analytics is a kind of psychological it's about psychology it's about statistics it's a science right which determines customer behavior so at the end of it wherever there is customer behavior there is human at the end of it so it's sector and industry agnostic as long as there are humans taking decisions they will always be analytical models that can predict those decisions much better next part not an all analysis is analytics right there is some basic slicing and dicing of data across tabulation so don't let people get away with this myth you know we are doing sub deep dive analytics a lot of analytics could also be simple analysis and understanding of your data and it need not involve any statistical models or machines or techniques don't miss words for the trees focus on the data the consumer insights they will always be a lot of noise the data that you see in your companies in your systems don't get distracted and focus on what is the ultimate goal that you are trying to achieve through that small exercise you are undertaking with that data yeah another part is inch-wide mile deep or mile wide inch inch deep that's that's a phrase that you would hear right should you boil the ocean and look at a lot of data to come up with a model rather boil a smaller ocean look at maybe 20% of data or 10% of your sample data which is much more smaller much easy to analyze much more reliable and come up with models based on that rather than firefighting the data battle and trying to get everything solved in one poke in one go there is a difference between causation and correlation that's number six two events could be going together every time I have a flat tire India wins the match in World Cup that's highly correlated but is it is my flat tire causing India to win no so we need to understand similar relations within our data two things could be going together but they are not causing each other so let's not get fooled by it segment of one so there are many companies and industries nowadays who are proclaiming we can look at our customer segments as small as one which means each customer is a segment in itself really no it's not worth it your segmentation exercises should be large enough for you to be focusing your marketing and your business spends on right companies can claim to say analytics can claim to say that we know what are the preferences of each single of rker of our customers but still it makes sense only to look at customers through a sizable segment before making any critical decisions and finally if you try to pursue analytics within your organization's you also have to keep the alternative business case ready you would often face this question which is about you know what's the return on investment what's the cost of doing this particular model why should I do it what if I don't do it so you should be able to convince to your management your organization's fair enough but what's the cost of not doing it if we well not to understand our customer segments there is a much higher loss in losing some of our existing customers to a competitor so you know keep your alternative theories ready if you are trying to convince your organization's about picking up analytical models or a few smaller chunks in the analytical journey yeah if you could spend ten seconds of all the things that we have discussed today in trying to caption this slide so this is what we spoke earlier I think just a minute ago is don't miss the woods for the trees yeah so stay focused as you try to pick up smaller chunks of analytical analytics based projects within your organization's quite open to questions sure thank you thank you so much for the session we do have a lot of questions over here yeah yeah so the questions say hey how business analytics in digital marketing can compliment each other yeah actually I think it's a it's a it's a subset they go perfectly well together we haven't covered it in this session but but web analytics is a completely different thriving and striving area as well web analytics right the number of clicks search engine optimization search engine marketing how our customers behaving it on your websites on your digital forum so think about it guys you know it's just a different distribution channel of doing business digital right some of these ecommerce sites they're in tile model is digital so that's your retail store instead of a physical retail store so actually all the analytics right that we spoke about can be very well implemented on your digital marketing forums as well so like we mentioned merchandising imagine you know the way you place your products on your websites is another way of placing your products that are clustered together so actually digital marketing is the the marketing part of it is the outside the customer interfacing part of it how you reach out to your customers but what goes in the back end is all driven by analytics in the digital world as well right right so this is one more question and I think this is really important so can a person someone is asked can a person with 15 plus years of experience let's say non analytics background transitioning like what is the way that to off scale like if yes then what is the missing link can a person with 15 plus years of non analytics background transition into the field of analytics brilliant question and I'm sure you know a lot of us have these questions you know is it too late for me to catch that analytics bus so the good news is no right and as I mentioned is pick up elements within your organization if there are areas or opportunities smaller datasets smaller problems that you can solve reach out to the people who make a difference and propose them you know as you learn various tricks of the and techniques within this course is pick up smaller chunks small projects take baby steps convince people create business cases that this is what I want to solve a smaller problem through analytics now if that is not possible within your organization if you see a ceiling it doesn't you know a hamper if you keep looking out you know maybe in smaller analytical driven organizations startups there are lot of lots of entrepreneurs out there in the analytic space you can you know but you need to excel that's the point you are always fighting with the youngblood you are fighting with the young ones right so you have to excel in one particular area pick up a small project a technique that you really excel in and go out and display the only difference is that you will have to have hands-on knowledge you can't just speak analytics in analytics you have to get your hands dirty proof projects have some sample case studies some success stories and start reaching out to people in a nutshell a single word answer it's possible even after your 15 years of non analytics experience I think I think that's a perfect way to put it thank you so much and I think we will close with one question the last questions yes so one more question so people are here a little bit concerned that hey if I'm not from a technical background can I still post my analytics yes yes okay you know that's that's a detail subject in itself but again I'll try to keep it short yes so you know your success in analytics is not determined by if you are not from a technical background so there are two lens to analytics one is the IT part of it the data visualization creating datasets data warehousing that's also a subset of analytics and next is the statistics part of it which is more about techniques and models that we have been discussing all along but even if you don't belong to either of these segments you know take heart that still you can have a career in analytics because analytics also needs his business translators so people with good domain expertise you may not be a statistician or an IT personnel but you may still be rich in your domain understanding about say telecom or media or manufacturing or retail if you understand those domain you can play a big role understanding these analytical frameworks and you know being that business translator that domain expert so you will have teams working for you creating the models but your success in analytics is about how do you sell these models the impact of these models to your management and leaders so you need not be a technical expert but you can still succeed in analytics and it's never too late to you know make a start and understanding these concepts right right I think that sort of brings it to a close thank you Himanshu for taking the session and like giving us insight into how analytics works in to marketing and retail and how we can always transition there's no barrier and is the one stopping transitioning carrier that antics so thank you thank you so much for the session sorry and thank you thanks guys have a wonderful weekend bye bye

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *