Deciphering Analytics
When we see , hear or even speak the word Analytics, it conjures up, Graphs, Charts, Numbers running all over a computer screen. It is often associated with a ‘Data Scientist’ virtuously spewing jargon on Random Forest, K means, Rsquare or some such.
Is that Analytcs? Partly , Yes. But there is more to it.
Having worn different hats at different points in my career as a Sales Manager, Product Head, Co-Founder running a firm of Consumer Behaviorists and now as a CMO and Analytics Head , I have a more useful definition of what Analytics really is
‘Analytics is seeing what others see, hearning what others hear, read what others read, but be able to come up with thoughts and actionables that others DO NOT.’
It hence goes beyond the models and the charts, for they are merely tools, as much as a jackplane is in the hands of a skilled carpenter. It is his skills, mindset and discipline which gives us a flawless table, not the tools.
So what skills are needed to be good at Analytics?
Clearly dexterity with numbers is a primary skillset and so is knowledge of mathematical models, data transformation. But this is commodity.
Increasingly teams in Analytics will need the follwing skillsets and tool to thrive and make a business impact.
1. Mind-set – A near childlike curiosity which is not jaded by repetitive tasks involved in the analytics process (Data Extraction, Preparation, Loading, and Reporting). This means that the analyst has to always be open minded to newer possibilities and different ways of looking at the same data.
2. Patience – An underrate virtue. Very often predictive analytics, projections take time to play out in the market place. There are always contingencies, factors outside the control of even teams who are implementing the solutions. Since output is delayed/ not in line with expectation there is always a clamour to relook the input. This is where patience and staying with a plan gets critical.
3. Feedback Loop – No analytics project can be complete without a feedback loop. This is the Secret Sauce of this trade. Importantly feedback
loops may often be beyond the control of the Analysts e.g. – Frontline Sales Teams keying in all details of consumer interaction into the CRM. Hence the design of any process regarding Forecasting, Predictive Modelling needs to have the feedback loop pre designed or based on KPIs which have fewer contingencies.
4. Culture – This is where the Analytics team needs the support of the corner office. The right Culture at an organizational level means that the ground level feedback on consumer facing interactions flows back as Feedback. Culture means that the organization has the patience to wait for results and not be trigger happy on making changes to algorithms, models or logic.
Analytics is seeing what others see, hearning what others hear, read what others read, but be able to come up with thoughts and actionables that others don’t
5. Causality over Correlation – In any data set there will always be interesting variables which show high correlation to the result but questionable to nil causality. In my own experience I have seen a situation where we saw that consumers who buy Whirlpool fridges tend to take Personal Loans. We used this data point and actively tried to convert every consumer who wanted to buy any other brand of fridges to Whirlpool in the hope that they will eventually take a personal loan! The plan backfired!!
One of the best examples of correlation WITH causality is perhaps how continued education (effect) of the girl child in rural areas (especially the most backward of areas economically) is a direct sign of improved quality (correlation) of life due to easier access to micro finance (cause). It is important for the Analytics team to identify Cause- it helps identify which on which variables to nudge consumer behaviour.
6. Awareness of ‘Small Data’ – Access to Hadoop Clusters, AWS Cloud computing etc. means we can crunch unstructured data will never before. This is what we call ‘Big Data’. My experience as a Consumer Behaviourist taught me the importance of ‘Small Data’.
Small Data is basically going back to consumers who exhibit a particular behaviour and simply asking an important question – Why?
This is an exercise which Analytics Managers / Heads in fact all P&L managers must routinely do- step out in the field and simply send time with consumers to understand why they behave so. This makes the analytics far more potent.
As the CMO of Fino Payments Bank I often wondered why a Micro ATM offering merchant in the middle of Jaunpur district did extra ordinary business. No data point could support the outcome. We simply spent time with him to figure his secret – He opened his shop earlier than the market and shut 2 hours after the market closed (was hence perceived to be always available) and greeted every customer who walked into his shop with a smile and asked for their well-being. This simple act endeared him to folks who wanted to withdraw money and they repeatedly visited him! No data point in an excel sheet would have ever captured that.
7. Simply. Simplify and Simplify. – To most things mathematical are an anathema and hence threatening. To take the business teams along and get their support an Analyst should have the skills to simplify his/ her craft. Once business understands the basis for predictive modelling, use cases for AI etc., they appreciate the solution, impact. Else it becomes a case of ‘I told you so’ when projects fail and many will fail.
All the points above are the softer but critical aspect of Analytics which often get ignored in the quest for more compute power, more variables to be added. But all the above are irrelevant if the basic quantitative skills are absent and if the company does not invest in the right infrastructure.