marketing analytics case study This is a topic that many people are looking for. star-trek-voyager.net is a channel providing useful information about learning, life, digital marketing and online courses …. it will help you have an overview and solid multi-faceted knowledge . Today, star-trek-voyager.net would like to introduce to you Marketing Analytics: Predictive Analytics in Marketing . Following along are instructions in the video below:
“This video. I am going to talk about how one can use statistical modeling techniques techniques to solve problems in marketing now if we re working with marketing departments then come across a number of problems number of choices to deal with number of decisions. That is two that have to be made so how do we solve such problems in marketing using statistical model effort mobile you see one of the problems that you normally face in marketing. It s important through the iteration rate also important know the potential customers.
It s important to know how we markets that exist and you know that so the company should go with so that its optimum how to allocate a marketing budget. How to you know have a good advertising strategy how to have how to assess the impact of marketing campaigns. Whether it s effective or not how to know who are the loyal customers. And what are the key drivers for loyalty how to know how to have a really good direct marketing strategy.
Key drivers of cells choosing between different marketing products and strategy. So these are some of the problems that you you know come across in marketing and how do you solve them. So there are several a very good industry practices. That that s used in solving marketing problems.
And we ll talk about few of them. Okay. The first problem that we talked about in the previous slide was knowing the churn knowing what the customer going to leave the company that they are not normal to be with the company they have been you know buying products and services. But now they will be going to someone else so snow natural modeling also known as attrition model.
It so it s a predictive model and it s about predicting the chances of customers leaving so it s a model that uses probability models to find out the probability that a customer will be leaving okay so it s basically a classification or potential turn customers from rest of the group. Okay. So if you know beforehand that how any customers. Okay.
Let s say you know that five percent of the customer and who are the five percent of the customer. We work you know going to leave the organization. Then you will be better prepared right. And you will have a better strategy for them just to retain them.
So retention is something that is the most important thing to know in turn prediction also important to have you know the numbers. Very correctly beforehand so as to have you know an effective strategy for this small section of customer when likely to churn so that different matrix used even in daily operations. Where you know you have the model in place to know the customers potential turns and then you use that you know day to day business operation to ensure that that can be minimized. Many number of models are used for this mostly classification model such as decision 3 logistic regression ensemble models like random forest as well as you know more sophisticated liver network model as well depending on the type of data or the quantity of data you have with you customer lifetime value model so it s very well used in many industries and traditional model have been in used for many decades now so it s about knowing the lifetime value of a customer for a group of customers.
And it s basically in money terms of course value cannot be measured. There are non monetary value also but that cannot be quantified easily so monetary value is is something that is quantified or quantifiable and that s what is the motivation bi8 behind the customer lifetime value also known as clb model. So it s about knowing the length of engagement with the customers. How long the customer is going to be chained so that s you you find out using survival models artificial models then the probability of partitions when is going to what is the probability of the customer leaving so that s a classification model that we saw in the last slide the revenue model so how much to make from customers with you know that kind of engagement for a given engagement.
Okay. So you know the period of engagement. You know what the products are and then you calculate the revenue. So that can be done for an individual and also for a segment.
So that s the motivation of clb modern market mix models again a traditional model have been in use for many decades now so it s a models used for decision making purpose..
And it s not a predictive model rather. It s a model used for knowing inference. Or drawing inference just to get insight from your data so most data science analytics models have used for two reasons right to get insight about about the data and to you know do automation of prediction rather right so. This is more like a problem that deals with insight.
So it has to do with inference instead of predict so that s what i mean so many decisions are taken in marketing. Because you always optimize want to optimize the marketing spin that you are making so that the revenue is maximum that means for a given amount of marketing budget. We should you know have a strategy that optimize the revenue or sells what a mitt whatever matrix that you are following so targeting right audience is important targeting of finding a right time to target customers. What is the right frequency or targeting or having marketing campaigns right channel.
What is the perfect way whether it s social media or its internet offline online. So many you know choices. One has and then price determining the price setting the price then so it s basically used to take decision on marketing spending activities. Okay for instance you might have to choose between whether you will go for tv ads or google advertisement online desmond s and how do you divide your budget between these two okay so that could be one problem that can be considered as a marketing mix model.
The techniques that you would be using would be a number of but some of the popular ones. A linear regression panel. Data analysis mixed models. And so on sales.
Forecasting. Model. Its model use to find out other predict the potential sells in short term or in long term. So that could be long term focused and short term focused and the modeling techniques that you will be using for short term and long term focus would differ.
And that s important. Why you know categorize then separately okay. So the model study you use be using could be time series univariate time series models multivariate so a multiple linear regression model. Even multivariate time series model so such models you use to forecast some time panel data model does are also used you know to forecast advanced machine learning model especially non linear models are also now heavily popular in you know forecasting so more sophisticated machine learning models are now being used so that s cross sell and upsell model well these models have been in used for long time and people who can create car industry people who work in insurance industry.
They re very familiar with such models and now it s being used in e commerce and you know many other activity and even with the rise of e commerce. These these models have become more important to understand i mean one of those days when people lose to contact customers through other means offline means. But now this could be a very direct communication with the customers to email. Who you know the user ids and through social media.
And so on so this model is all about exploring cross sell opportunity and cross sell option nothing. But the marketing jargons for selling something to your customers. Which is already who is already customers like you are it s an existing customer or maybe a new customer that s not an important. But what is important that if somebody is buying one product.
You recommend him to buy another product okay and for instance just to understand what cough sillies. Somebody is buying a mobile phone you re also asking in to buy the cover of the mobile forms that s upsell and cross sell could be you know selling a totally different product. If somebody s buying a mobile phone. You re also selling him a watch okay.
So that watch is a different product than mobile phone..
Does that s cross sell and upsell would be where the two products are there is related. Okay. You know which may be complementary in some sense or may be related at a broad level. So and that s quite effective especially increasing your cells and thereby increase your revenue.
So many models seduce. Many techniques classification algorithms to know what could be the best option opportunity to be the best cross sell. Opportunity linear programming. How do you optimize it ensure that you know that doesn t have a negative impact on your overall selves market basket models just it s a very simple model again just to count the number of times.
Somebody you know buys a mobile phone with the watch and finding the frequency and recency and so on and then having a model to ensure that you know you re recommending correct product for cross sell and upsell loyalty model. So loyalty in marketing is nothing. But finding out customers who are being with with you or with the brand for a long time. And what the reasons are why people are the same brand.
So you should know what a key drivers of your loyalty okay and it s important to know for my overall company strategy point of view and then loyalty could be different for new customers and old customers that means so loyalty that was there in 80s could be different from 90. And could be different work. What is even now so finding a different trained in royalty is also important in and that s what we tried to known from royalty models so the techniques that use in loyalty modeling could be eco metrics models such as linear vibration hypothesis testing anova structural equation modeling and so on segmentation models heavily used in marketing analytics. Anyone working in marketing.
They know that you know to know how many marketing segments or how many customer segments. One has to cater to you cannot have just m. Number of number of large number of mark segments. That would have a negative impact at the same time it cannot have just one segment.
So what is the optimal number of segments that you one should have for your marketing activities. Whether your sales activities is something that is decided from the segmentation model. Okay. So you need to know the number of segments.
The types of segments product category with respect to different segments. The techniques that you would be using could be linear regression decision. Tree unsupervised. Learning algorithms such as clustering principal component.
Analysis. Factor analysis. So these are the techniques that we use in segmentation modeling. Survey analytics.
So value at x. Is something that has become more popular recently especially especially in the e commerce. A and ecommerce business. Because nowadays.
It s very easy to get customer feedback very easily you know gone at those days..
Where you have to go to the customer each customer to get his feedback offline basis is now not the case and now you have online boards or online companies offering survey analytics. Okay so basically it s marketing richards you know people like companies like nielsen and all that great market getting research to solve a right whether it s fmcg product. This retail products whether it s ecommerce product and so on so when you do a survey. You actually study the customers preference you know the feedback on your product and services this could be online offline of course online is becoming more popular and traditionally it has been offline.
But the tools and techniques that you have been using for offline can will be used for an online marketing service by modifying certain things. Which which are required to be modified. So it uses a mixed bag of classification and regression models to understand not just to understand to also have performed strategy. So you can use statistical model to get a lot of good inside from survey data price elasticity of demand models not very often used probably not in model used in many industries in fact i have up with people who you know this pose.
This question about knowing you know the demand how demand get affected with respect to price. But then do not quite know that with data you can actually solve that problem ok. And that s a problem that almost every industry face but not every industry uses in demand forecasting sorry a priced elasticity of demand model so this is the model to know that if you change your price right oftentimes you have to change your price you have to correct to a price could be and a more sometimes you are increasing your price. And that s not a good thing for your customers.
So. How i should go to impact or demand. Okay. So your idea is to know the change in demand.
The chanting demand with respect to the change in price. So. That s what you were trying to model in in this in this modeling technique. Okay now once you know that right you ve been a bit of position to decide what could be the optimal change because end of the day.
What matters to your revenue right so what price would ensure that your revenue is optimum that you would come to know by having a price last model. And how is that related to marketing analytics. Right some some might think people might think that this is more of a strategy. Not of marketing well price is an important factor in marketing.
When you go to a customer. The first thing customer will ask you is the price and you have to justify it and you know going from back taking and backward approach you all need to you need to know that whether increase in price is going to affect the demand or not whether it s an elastic or an inelastic product is something that you know marketing managers need to know before even you know going to customers and selling or trying to talk about the customers so it helps you in building your story. He be testing heavily who s in product analytics. The famous case study could be the emmy testing use done by google in the early days of google.
Many companies to it on a daily basis or in a regular basis like most large tech companies amazon google and so on so these are decision making models. Which are used to choose one decision over the other or one product strategy of the others now every testing used in many fields not just in marketing also in product analytics. And so on but in marketing also you could decide that what what that will be two sentences in your in email subject. Now you could ask yourself.
Which subject would do better. Okay okay. So that could be one one case. So second thing is whether you should write their long email or a short email.
So that decision can be taken from a b testing. And the popular technique that is used to maybe testing is hypothesis testing. Which is very simple one yet. Very powerful.
So that s what a b testing is all about and it s quite popular in marketing analytics. Especially deciding you know which one to choose. And it s not just to do joyce s it could be several others it could be you know abcd as well so. Although.
It s known as a b testing. You know you can have more than you know two categories. Where campaign analytics. Now people working in marketing analytics.
They re very familiar and people who work in marketing. You know that campaigning is it s a very common activities done in marketing departments or by the marketing departments. So the motivation behind in camping and campaign analytics is to to know how effective is your campaign and which campaign is effective. What is the time that is you know giving you the optimal effect.
Which products are giving you the maximum effect. Which segments are having a good impact and so on so that s the idea behind using campaign data. And doing camping and campaign analytics. The tools and techniques that you would be using could be many economic models like linear regression logit probit models principal component analysis many tree based model can be used optimization technique can be used especially for you know ensuring that the decisions variable so that you know that that is going to come out of this campaign analytics are optimum so you know optimization techniques can be used along with economic models so these are some of the you know marketing analytics models that people use in the industry and it s not the existing list in fact there are no number of moles.
So these are the tail 12 models that we ve talked about here you want to learn more about it getting hands on experience on this through videos code and with data. So you can get a steady steady pack. Which you have you know created specifically. People who are interested in marketing analytics.
And not just in marketing analytics. This thing s can be used in cells. Analytic. Strategy.
Analytics product analytics. So people who are interested in cells. Who are in cells advertisement well pre cells strategy. Okay so they can also find this terrific.
Very useful ok and you can contact us. We have this email id. You can contact us thank. You so ” .
Thank you for watching all the articles on the topic Marketing Analytics: Predictive Analytics in Marketing . All shares of star-trek-voyager.net are very good. We hope you are satisfied with the article. For any questions, please leave a comment below. Hopefully you guys support our website even more.
“In marketing, people face a number of problems.While these problems have been traditionally solved using gut feeling, more recently predictive analytics is being used. The problems normally faced in marketing are as follows:nnKnowing attrition rate, potential churn customersnnKnowing how many market segments existsnnHow to allocate marketing budgetnnImpact of a marketing campaignnnKnowing Loyal customers/key driversnnDirect marketing strategynnKey drivers of salesnnChoosing between different marketing/product strategynnA number of models predictive models are being used to solve the above problems namely : Churn model, Cross sell /Up sell model, Attrition model, Loyalty model, Market mix modelnnANalytics Study Pack : http://analyticuniversity.com/nnAnalytics University on Twitter : https://twitter.com/AnalyticsUnivernnAnalytics University on Facebook : https://www.facebook.com/AnalyticsUniversitynnLogistic Regression in R: https://goo.gl/S7DkRynnLogistic Regression in SAS: https://goo.gl/S7DkRynnLogistic Regression Theory: https://goo.gl/PbGv1hnnTime Series Theory : https://goo.gl/54vaDknnTime ARIMA Model in R : https://goo.gl/UcPNWxnnSurvival Model : https://goo.gl/nz5kgunnData Science Career : https://goo.gl/Ca9z6rnnMachine Learning : https://goo.gl/giqqmxnnnData Science Case Study : https://goo.gl/KzY5IunnBig Data u0026 Hadoop u0026 Spark: https://goo.gl/ZTmHOA”,
marketing analytics, predictive models, marketing, SAS, credit risk models, cross sell model, up sell model, data science, career in data science, marketing …