Big Data, Analytics, Data Mining, Modeling… are just some of the terms that seem to be in favor as the “hype” words for past few years, and since these refuse to go away, we need to understand how this can be used to add effectiveness to our work as marketers.
In simple words, various sources of data are available to a typical company and the idea is rather than merely collect this data and store it in servers -we need to find ways to use it better. For example it may be possible to make the following significantly more efficient – deployment of marketing, sales efforts, supply chain, pricing decisions. In fact not using the data to improve these is like a detective ignoring DNA samples, fingerprint data etc even though he/she has easy access to it. Data is changing almost every decision making process around us and a good working knowledge of how we can use this opportunity is a good idea.
Identifying the right sources of data
From the narrow confines of the marketing and sales function in a consumer facing company, there are many data bases available. The internal sales figures going back a few years ( this could be again from primary or secondary sales), external sales reports that report competition numbers as well, marketing input data ( advertising related, BTL, consumer promotions), pricing, distribution, trade schemes, online interactions data, production data. The idea is that if we bring all of these data into one single space and study, and manipulate (or model) them we get new insights that help our business.
If all of this appears like a boring IT job, you are right. While the IT boys are capable of pulling all of this into one place, they cannot do so without the insight from the domain team aka you the marketing guy or gal. For example, they may try and see the effect of advertising on primary and secondary sales when they should look at retail sales or an estimate of it. They need an underlying understanding of how all these parts interact. Given that data will become more and more important for decision making ( guess there is no escaping this!), it is the data focused marketer who will win both in the internal stakes for the corner office as well as be the architect of the successful brands.
So how does one move forward?
Start looking at the numbers and find ways of getting reports in as real time at as granular levels as possible. Do you know the impact of your primary scheme at a region level? City Level? Shop level? This will mean some smart report formats and an IT system that pulls the data and reports on the fly. In other words, Business Intelligence Software. Start understanding what it takes to get a handle on this in consultation with your IT partner.
The above is still in the area of what happened and not why it happened. For the latter, one needs a causality analysis – which means statistical modeling. The model essentially figures out input-output relationships and assigns causality. The sales in a particular period were driven by changes in distribution, consumer schemes, or advertising etc. You will need a statistician that can assist you in building these insights (Welcome to RainMan!!)
Another step in same process is to build predictive models even if one does not understand the actual causes but by identifying variables that seem to influence outcomes. For example, hot weather could mean increase is soft drink sales but very hot weather may mean a drop. The model does not help you understand why, but helps you predict sales with temperature as one of the variables as this is empirically right.
Published with permission from RainMan Consulting
