Once upon a time, getting to grips with data and applying analytics was a killer business differentiator. Not now.
Analytics routinely supports operations, drives processes, and helps make decisions.
The reports generated enable organisations to track sales performance, churn volumes, and so on. It’s an essential part of assessing and correcting tactics and strategy to keep a business on its planned course.
The trouble is, this is ‘rear-view mirror analytics’. This model tells you where you’ve been, using historical performance reports as a decision-making platform for future planning and investment.
This is fine in a stable market, but in our current world of disruptive unpredictability, differentiation (survival even) depends on looking forward.
Trying to predict what’s going to happen and enabling decisions that can secure a sustainable future; preparing for projected events, and realigning if necessary.
Take the blinkers off
Future predictions are only as good as the data provided.
A model based on traditional transactional sources can only make limited forecasts. What about all of the data beyond these transactions?
Take churn for example. We can tell the model about a customer’s spend over time – whether it’s declining or fluctuating, and how many products the customer holds.
What we miss are the interactions and behaviours of the customer. The time they called into the call centre and were negative, or when, browsing new tariffs online, the customer was forced across three channels before they could complete an action.
These kinds of interaction paint striking pictures, which are a lot more informative than many of our traditional customer metrics.
Putting analytics into context
So the question is, how do we leverage these interactions?
The data is in a myriad of different formats, data types, and complexity. Employing one analytic technique alone no longer makes the cut.
To make use of call centre logs we need text analytics and sentiment extraction.
To look at customer journeys in digital we need path analysis, and we can further-analyse movement between channels by making use of attribution strategies.
Graph analytics, also, can help us understand whether a customer is influenced by other people in their network of family and friends.
Running all these analytical techniques provides a new set of behavioural and interaction variables that feed into our predictive model.
Now, the model knows how customers are being affected; how they feel, behave, and experience the brand. We’ve moved from a descriptive customer view, to a contextual view.
The analytical game-changer
Traditional methods for improving our analytics and the modelling we carry out, is making small incremental changes to the variables we input.
This results in slow progress and improvement. We want to change the game by injecting new knowledge and analytics.
>See also: What’s the block with predictive analytics?
Multi-genre analytics allow us to extract value from new and old data, enhancing our understanding of customers, products, interactions, or a-hundred-and-one other critical events.
In turn, we improve our analytical maturity, providing a higher level of confidence and more robust answers based on broader-ranging evidence.
Sourced by Yasmeen Ahmad, data scientist, Teradata