<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=306561&amp;fmt=gif">

Moving from reactive analytics to proactive analytics

Written by Corinium

Moving from reactive analytics to proactive analytics

Written by Corinium on May 20, 2019 1:00:22 AM

CDAO Sydney CDAO Data and Analytics

Ahead of the third Chief Data & Analytics Officer Singapore conference, we caught up with Murari Mohan, Assistant Vice President, Partnership Analytics, Business and Data Science,, NTUC Link to talk about moving from reactive analytics to proactive analytics, the cultural hurdles to be addressed in order to drive intelligent data strategies as well as the most significant steps to be taken to move from strategy to execution.

Check Out CDAO Singapore

Moving from reactive analytics to proactive analytics – where should an organisation start in order to ensure they are using the best available data to inform decision-making?

This depends from business to business, if you are tracking monthly behaviour or weekly behaviour. For example if you were buying groceries for instance, that would be a weekly behaviour, thus we would need to track the weekly behaviour of the customers and then decide which method of tracking fits best, whether its reactive, proactive or a mixture of them both.

Based on my prior experiences and with NTUC Link, we have a number of collaborative partners who have been utilizing reactive analytics to better understand the customer behaviour in relation to their products. This idea of reactive analytics means that the business identifies who their customers are and will then assign specific offers based on their transaction history. For example if a customer came in during January to complete a certain number of transactions, stops coming in during February and March. From the analytics we can see that they have become inactive, so we should target them with certain offers of benefits to try and make them reactive again.

We have also developed a proactive approach as well, which is based on past activities of the customer. Again for example if we know that someone is active in January but our data shows that from the previous year they were also only active for 2 months, January being one of those two, with the use of our predictive model, it will be able to predict out of our current active customers, who are most likely going to be inactive over the next couple of months. As this is predictive, it’s not certain that they will not come back, but based on our proactive analytics approach we can see out of one hundred thousand customers, the top twenty thousand who are most likely to not come back.

Now we have an idea about who these customers are, we can provide them with offers to get them to return. But how can we measure this? Before the offer is released, we will notice a decline in sales volume with these customers, however after the offer the sales volume is going up. That way we can see, whether or not the proactive approach is actually helping the business.


What are some of the major cultural hurdles that need to be addressed in order for an organisation to drive intelligent strategies that align with business priorities and information needs?

This will depend on which line of business or which industry the business is working in, I am currently working in the loyalty section of the business and here you have retail customers broken up into online and offline customers, who tend to have very little appreciation for the use of analytics in targeting. Some think that by investing $10,000 in analytics to build a strategy they can spend a certain amount in the marketing channel and thus create immediate results, and that is a specific cultural hurdle that I have been experiencing. One of the biggest cultural hurdles I face in my role as analytics head is more to convince the stakeholders to invest more money and time into an analytics driven or data driven strategy, which would support the business and would give us a bigger boost by the end of the year. This hurdle come about because the results are not immediate, however, with analytics strategies, you can’t see the immediate impacts, for most part you will need to wait a full cycle of the strategy to actually see the results.


How do you test and prove ideas to gain robust and reliable, ultimately more intelligent insights?

Analytics is all about test and proof, we’re always proposing different strategies, in my case, my team proposes strategy B, however the business is already using strategy A, so now you have a champion strategy, why would the business want to change that? To make the change in the business, we use the process called the champion vs. challenger approach. The champion strategy is the one currently being utilized by the business. As an analytical person, I come up with a new strategy, we call it the challenger, and I challenge the business that my strategy is better than the champion one in use. My challenger strategy will not come into full effect until I get the proof that it works better. To get the proof, the champion strategy will run on 80% of the data and the challenger will run on 20% of the data based on a random selection. If by the next month I can prove that the 20% of my strategy data worked better than the champion then for the next campaign, we will run the challenger strategy on 30% of the data and the champion on 70%. And again if I prove that the challenger is working better than the champion, we have thus proven that the challenger is consistently out-performing the champion we will change the strategy accordingly. This is how, in analytics, we test and prove our ideas to gain robust and reliable and intelligent insights.


What are the most significant steps to be taken to move from strategy to execution?

In my current role, I am not only part of strategy building, my team also does development and implementation. The first part of the process is to talk to the business to discuss the idea, the vison and a new strategy. The next part is the implementation, now, when implementing a new strategy, you have to work with all pillars of the organisation – sales, marketing, operations, data teams, and if you are working on a product, also with the product teams. And when communicating the idea with these different business segments, it’s important to be very technical, and communicate the idea differently across the segments in a language that they understand. The next step is to have a thorough document flow which highlights the specific responsibilities of the different teams working on the implementation. In order for this to succeed, it’s important to have a person who oversees all of these tasks and who has a thorough understanding of the strategy. The only issue with this element of the implementation is that it does slow the process down. However when you convince your merchant, you need to ensure that everyone understands the expectation levels in terms of the time it takes to execute. It’s important to note that if the implementation is not done properly, regardless of how good the strategy is, your result is going to be sub-optimal. Because when you’re still learning the strategy, you are changing the existing strategy of the business to a new one and if you don’t see a significant impact, you will lose all the future relations.



Murari M 

Murari Mohan

Assistant Vice President, Partnership Analytics, Business and Data Science , NTUC Link

Related posts