Archive for June, 2014

Jun 25 2014

Marketing Mix Modelling from Multiplicative Models

E-Commerce or Digital Marketing has emerged as a separate and rapidly growing business domain with many businesses thriving solely on them. With internet acting as a strong channel to perform business, there also arises a simultaneous need to use this channel effectively to add value to your business.


2014 – The year of Digital Marketing Analytics

According to a popular study on the Forbes magazine, 2014 is the year of Digital Marketing Analytics.

DMA Businesses that use digital marketing analytics for improving customer acquisitions, increasing brand loyalty, increasing ROI from their marketing channels, etc have a better competitive advantage over other businesses who have not yet ventured out into this space. The internet search giant company Google Inc, has also created a massive impact in the field of E commerce analytics through their Google Analytics platform which provides a variety of KPI’s and statistics to help digital businesses track and measure their online sales and marketing. AdWords, E commerce reporting, real time analytics are some of the popular services provided in this platform.


Digital Media Analytics

While these services are good enough to track high level trends and provide basic directional metrics, there is often a need to dwell the past data much deeper with sound understanding and usage of advanced analytical techniques to get better insights. One such area in the field of digital market analytics which relies heavily on the use of econometric models for decision making is Market Mix Modelling. The Gartner IT glossary defines Market Mix Modelling as “analytical solutions that help marketers to understand and simulate the effect of advertising (volume decomposition), and to optimize tactics and the delivery medium”.

In simple terms Market Mix Modelling refers to the estimation of statistical models to measure and analyze how various marketing channels are effectively contributing to your sales. If is often used to find the optimal mix of the market channels and forecast future mix which would help one to maximise the ROI from various marketing channels and sales. In particular it helps the business user address key questions such as:

  1. How do we decompose our sales to key drivers to understand the combination of these channels which would contribute to sales, market share or profits?
  2. How are these key drivers impacting sales over time?
  3. How do we calculate the efficiency or ROI from these market channels?
  4. What happens when we change the budget by channel


Market Mix Modelling – Methodology and Illustration

Market mix modelling relies on three popular model forms to understand the impact of various market mix variables on sales. Users use either of the three based on their requirement from the analysis.


Table 1. Summary of Market Mix Models functional forms


Additive models tells us how much of a change is generated in the response variable in unit terms with one unit increase in the explanatory variable and is only used in scenarios where the impact of each additional unit of the explanatory variable is identical. For example this model is not suitable in determining sales decomposition of seasonal brands such as yogurt or ice creams. However the estimation of additive models are easy and sales decomposition can be directly implemented to understand the variable contributions. Semi logarithmic models on the other hand show how much the absolute level of the response increases/decreases proportionately with the independent variable. An even better logarithmic model gives us the variable elasticities which exactly measures the responsiveness of the independent variable to the dependent variable. This model comes very handy to understand how price and promotion discount elasticity impact sales and hence the most preferred model in industry standards. However modelling and decomposing logarithmic or semi logarithmic models which are in the multiplicative form are indirect and quite challenging. A good understanding of the data and techniques is very essential to model market mix variables in the multiplicative form and to decompose it to arrive at something similar to the below figure.


Figure – Decomposing sales to key drivers

Market Mix Modelling from Multiplicative Models

In the above figure sales of a popular ice cream brand has been decomposed to understand how the key drivers such as its own price, the competitors spend, and its market channel spend from Press and TV impact the sales volume of the product. Sales volume has been decomposed here to its base effect (shown in blue) which usually can be interpreted as the sales derived from the brand value of the product and the incremental effect derived from the sales drivers. Impact of the market mix variable TV can be seen clearly in the months of Jun/July where the TV spend (shown in yellow) is made. A multiplicative model has been implemented here to capture the right effects of the variables and to understand the seasonality of the product sales as the product has a very seasonal demand.


A link is available to experience the true value from Market Mix Modelling here.


Right information at the right time through the right means

Adoption of modelling techniques in business and the insights gained from it are vital to the survival of any organisation. Here in this article I have showcased how the concept of market mix modelling developed in the right fashion could add much value to your business and help you foresee things which you could not have before. Right information at the right time through the right means is really worth your investment of time and money!







Jun 20 2014

How can you maximize your ROI through Targeted Marketing Campaigns

Every consumer company that exist in today’s world have vast amount of transactions data, containing information about every product that is sold and about every individual who has purchased it. Propensity and Uplift modeling would help in answering questions like:

  • Which marketing campaign will allow you to reach more customers?
  • Which is the customer segment on which the marketing campaign should be targeted?


aaum logoPropensity’ which literally stands for ‘change in behavior’ helps in finding which of the marketing campaign has the most positive impact on your ROI. This model works on the basis of score matching technique and then comparing the customer group who were exposed to the campaign (treatment group) and the customer group who were not exposed to the marketing campaign (control group). Comparison is made with respect to the underlying characters of every transaction between the treatment and the control group.


Uplift modeling helps you to identify and target the customers who want to hear – isolating the ones who don’t. Many marketers overlook the pricey consequences that can result from sending mass, untargeted emails or marketing communication messages to their entire audience.

Uplift modeling has helpe dUS Bank better target its customers, and as a result, it has reduced its mailing volumes by up to 40% and has achieved a five-fold increase in campaign return on investment compared with existing programs”

There is a fundamental segmentation that separates customers into the following groups:

The Persuadables : customers who only respond to the marketing action because they were targeted

The Sure Things  : customers who would have responded whether they were targeted or not

The Lost Causes  : customers who will not respond irrespective of whether or not they are targeted

The Do Not Disturbs or Sleeping Dogs : customers who are less likely to respond because they were targeted

LoyalSIGHTS helps you to identify the persuadables, the only segment that provides true incremental response.

Jun 10 2014

Crouching “NoSQL” Cautious “SQL”

“According to analysis by Wikibon’s David Floyer (and highlighted in the Wall Street Journal), the NoSQL database market is expected to grow at a compound annual growth rate of nearly 60% between 2011 and 2017. The SQL slice of the Big Data market, in contrast, will grow at just a 26% CAGR during that same time period.” NoSQL adoption will go big bang but will it impact all the organizations? Will SQL lose this battle? Does everybody need to care about NoSQL? First of all, there is no battle to begin with. NoSQL doesnt mean NO SQL. It simply means “Not only SQL”. Wikipedia defines NoSQL database as

                  “A NoSQL or Not Only SQL database provides a mechanism for storage and retrieval of data that is modeled in means other than the tabular relations used in relational databases. Motivations for this approach include simplicity of design, horizontal scaling and finer control over availability. The data structure (e.g., key-value, graph, or document) differs from the RDBMS, and therefore some operations are faster in NoSQL and some in RDBMS. There are differences though and the particular suitability of a given NoSQL DB depends on the problem to be solved.”


Let us understand SQL and NoSQL a bit more better!



Clearly each one has its own merits and shortcomings. There is no standard solution that fits to all business requirements. Sometimes it is more shocking to see how people react/make decisions on this front. See for yourself one such fight.



Platforms, technologies should be chosen to fit business and not vice-versa.  While there are problems SQL isn’t suitable for, it still got its strengths. Lots of data models are simply best represented as a collection of tables which reference each other. Therefore SQL atleast for now should be good enough for most of the organizations. Right now, it’s fair to say NoSQL is only relevant to a very minor proportion of businesses. But that proportion is very very significant! The changing data arena would sooner see a technical innovation driving business innovation very similar to what we have seen with smart phones?

There wasn’t a need for smart phones but the technical innovation provided business opportunities. Likewise, big data promises a huge opportunity for NoSQL.