Hello everybody – we do hope this article finds you well.
Advanced analytics solutions can significantly improve your bottom line but you really are making a difference when these are integrated into a holistic strategy.
Let’s have a peek over into our pipeline for analytics applications in the financial industry so that we better represent what we mean by holistic strategy. Below is a generic diagram that is not intended to be exhaustive and is drawn from our experience especially with companies with retail credit risk exposure
Holistic analytics flow:
Now let’s deep dive into each stage:
Do you remember us preaching how you should always start with the business-case (business pain-point)? Well, nothing has actually changed. Looking through your overall business strategy and implementing a holistic advanced analytics approach works only if you face your data: in the beginning, in the middle, at the end… To word it differently – data is throughout the whole process and should be considered at each and every stage.
A couple of key points to take away while thinking about data:
- Data availability, storage, ease of access, self-service;
- Data quality and its dynamics over time;
- Potential misalignment between seemingly equivalent data calculated by different business units;
- Data quantity – are we in the possession of sufficient data for a certain exercise. If your analytical potential estimation tells “not” – we suggest that you play the long game to collect further fields, information, incorporate data from the web, etc. These efforts will pay-off for your business.
2) Customer life-cycle
Our customers go through many stages during their life before, within and after being with us. Having a full-cycle, connected attitude towards our most valuable asset is crucial and is what makes us stand out.
3) Advanced analytics tools
Incorporate an advanced analytics tool along with the knowledge gathered while building it for each stage of the customer life-cycles. Thus, boost overall performance and hence bottom line.
A few examples:
- Develop predictive Leads scoring – before investing into marketing efforts (e.g. brochures in the mail box) utilize Machine Learning to increase conversion rate and hence ROI;
- Analyze how changes in the price of your product affect sales through promotional effectiveness (see our article on the topic). Furthermore, get a structured grasp of which are your sales volume drivers and attend to those respectively;
- Once that a customer is through with the application process (already accepted by your decision system) – manage their risk and realize cross and up-sale opportunities with an Application Probability of Default scoring model;
- For the already defaulted (bad) customers – apply collections scoring to optimize collections efforts, channels and timing;
- You do not only have an application Fraud – transactional Fraud can also be an object of investigation.
4) Complimentary metrics and calculations
You can even further boost your advanced analytics tools, their interpretability and how they fall within the overall strategy. This you achieve by calculating or recalculating some measures that help you strategize:
- Having more reliable Customer Lifetime Value (CLV) calculation based on Machine Learning (ML) approaches can help you decide where to spend more money on retention where the latter you base on your Customer Churn solution (see our article on the topic). Paying attention how everything we do is linked together? J
- Coupling Application PD score (see our article here) with knowledge about prospects financial status (salary, expenses, credit bureau information) can further clear the clouds around your Maximum Offer decision on account or group level;
The environment that our business operates in is not static and we need to address this in a timely manner:
Your customers change along with the economy dynamics, the presence of competitors, innovations, etc.;
- Your business changes based on strategic decisions which aim to meet market demand and boost sales;
- The economy (macro) itself changes – sometimes unpredictably (black swan).
Even though having ML within your decisioning system can prove to be extremely beneficial it is important that we monitor the performance of our models and keep them up-to-date both with most recent data and business goals. This will helps us stay on the crest of the wave and bear the most fruit of investing into data.
Can you pin down one fundamental tool that we did not include in the picture above? Well, one such thing we do believe is Customer Segmentation – utilizing unsupervised learning to group customers alike and dedicate your efforts based on targeted profile. This analytical tool we did not draw since it actually is feasible within each and every stage of the customer life cycle. Example: couple customer profiling with the Behavior PD and you can further increase the results of your marketing and potentially collections efforts.
Attitude is everything
Having a holistic attitude towards your customer do pays-off.
We would like to remind ourselves, the business and our fellow analytical colleagues that having a model with the highest possible accuracy is not an end in itself. We should always remember the following:
- What our goal from a business perspective is?
- How does our model integrate with the rest of our strategy?
Let us know in the comments your experience in bringing up your company to advanced analytics level and if you would like us to cover a certain topic in our next articles.
With this series our aim is to increase data science coverage and to make data-driven decisions an integral part of more companies around the Globe.