Financial Services

Home Loan Collections scorecard to predict delinquency and minimize write offs

A retail bank used logistic regression based scorecard to predict repayment propensity.

The retail banking client had a significant home loan exposure. The bank wanted a scorecard to predict the likelihood of an existing home loan customer making a repayment. This scorecard would use customer’s own behaviour and historical behaviour of other customers for scoring them. They wanted to determine the node points of the collection process where predictive scorecard would serve best and then, develop these scorecards and devise their implementation strategies.

Reliable. Scalable. Actionable

The key objective was to build a scorecard that is reliable. They understood that  the right scorecard would allow them to help customers at risk and minimize their own losses.

  • Write of losses reduced by 4.1% within 6 months of scorecard deployment
  • A net revenue impact of close to USD 5 Mn.
  • Insights on repayment behaviour improved loan approval process

The bank used the scorecard results to build customer personas and their risk profiles. This helped them optimize loan sizes based on the risk profile of the new home loan applicants.

Long, cumbersome & manual decision making

Banks are under increased pressure from fintech and other disruptive financial solutions to improve operational efficiencies. With the ramped up data infrastructure and investment in technology driven solutions, our client is now better placed to use advanced analytics based solutions.

One of their key focus areas was to develop a predictive scorecard for existing home loan customers – and identify customers who posed a higher risk of default. The bank wanted to use the solution as a springboard for integrating more advanced analytics and technology driven solutions into their operations. They wanted to test if a faster home loan approval process could be developed – giving them the opportunity to leverage the existing customer relationships and taking the fight to the challengers – with a taste of their own medicine.

Scaling revenue growth with propensity modeling

The home loan scorecard solved the immediate, short-term objective of the bank in reducing write off losses. It also showed the bank develop an understanding of how advanced analytics, integrated with technology will help them fend of challenges from fintech and other disruptive competitors, including:

a. Leveraging the scorecard output to develop risk personas to improve loan approval process

b. Optimize loan sizes based on risk profiles using an advanced analytics solution

c. Reduce human involvement by eliminating manual, rule based assessments, making the process faster

The bank now has an automated decision process that is unbiased, fast, reliable and scalable.


Portfolio optimization for a hair care brand to drive higher marketing spend effectiveness

A hair care product client achieved marketing spend effectiveness using market mix modeling (MMM).

The hair care brand had a variety of products within their portfolio. They were having a difficult time optimizing spend across multiple products under the brand. Each of the product had a different target group, seasonality of demand and market positioning. They were looking for a marketing spend plan that could optimize the spend mix to maximize sales across the portfolio, boosted the branding and generated results that were insightful and actionable.

Great marketing mix for greater branding

It is more important to use the right marketing mix, rather than to spend money. The study provided the client with a clear view of return on marketing spend and better investments.

  • Spend reallocation across product portfolio to boost marketing sales
  • Better understanding of halo effect and cannibalisation across the portfolio
  • Improved returns using marginal return on investment estimates (MRoI)

The portfolio market mix study allowed the client to look at marketing as an integrated effort across different products within the brand portfolio.

Multiple products. Incoherent spend mix.

The hair care brand had multiple products catering to Multiple target groups, having different seasonality and Multiple price points. They had some idea of the halo and cannibalisation effect, but had not really looked at their marketing efforts at the portfolio level. The client was looking for an integrated market mix study on their brand portfolio and a co-ordinated spend mix for all products in the brand portfolio.

There were some challenges from a portfolio mix study. The portfolio was dominated by 3 products, few were generating some sales and about 40% of the products in the brand portfolio contributed less than 15% of total sales. Most of these products had low marketing visibility – due to limited spend budget. Competition was scattered across markets and included private labels and established brands with multiple brands and single product brands.

Each marketing spend $ serves a purpose

Portfolio optimization helps the client define budgets and maximize returns on all the products. The solution helped them develop a co-ordinated spend mix across target groups, seasonal trends and price points. Some key takeaways from the study are:

a. Leveraging the market mix results to maximize marginal return on investments

b. Co-ordinated messaging across product portfolio to maximize short term sales as well long term branding

c. Uncovering strength of each product, across geographies and target groups

Leveraging the results from portfolio optimization using MMM, the client got a comprehensive view of their marketing activities and the impact on sales.


Moving from reactive to proactive maintenance – using predictive analytics

A utility services company uses Predictive Maintenance to minimize downtime.

The utility provider has a network of about ~4200 distribution transformers located across the service areas. Maintaining these number of DTs is an operational challenge for them due to breakdowns, high scheduled maintenance cost and public inconvenience. They wanted to build a more robust, scientific and predictive solution – to forecast distribution transformer failure with high degree of confidence – therefore also improving customer satisfaction with their services.

Risk of failure

The utility provider was look at a machine learning driven solution that could predict the chances of failure with a high degree of reliability.

  • A failure prediction accuracy of 75% over validation dataset
  • Pre-emptive scheduling of maintenance, thereby reducing downtime
  • Better planning & scheduling of maintenance

A web based interface with automated reporting was provided to the utility services provider at every 24 hour intervals.

Random breakdowns. Difficult to predict.

The utility services provider had spread across teams.  The operations team held some data, the maintenance team had few other data sets and so on. They did not have a centralized database to dig into. The distribution transformer (DT) to meter mapping had a lot of data quality and consistency issues. And since DTs have a life of more than 10 years on an average, the data was not available for the lifetime for most of the DTs.

In addition to the above, scheduled maintenance data was not properly maintained, with a high percentage of missing values. The biggest challenge however was the low percentage of failure cases – creating a highly imbalanced class with ~.025% failures. Because of this failure rate, standard Classification algorithms would not work, and we proposed a boosting approach to create more balanced classes.

Automated, accurate prediction of failure risk

The predictive maintenance (PdM) solution deployed for the client had an accuracy of 75% on the hold-out sample.  

With the web based interface having integrated reporting, the client now gets reports at desired intervals. This enables the client to:

a. Identify DTs that are at higher risk of failure and proactively address them before a breakdown

b. Improved and planned maintenance scheduling, reducing inconvenience and revenue losses

c. Improved customer satisfaction levels with better, uninterrupted supply of power

Predictive Maintenance (PdM) solution enabled the client to serve customers with the desired efficiency.


Beyond cross-selling – Recommendation engine For a hyperlocal retailer

A hyperlocal retailer implemented recommendation engine algo to cross sell & improve item visibility.

The hyperlocal retailer started their business with milk and egg delivery. Once they had onboarded significant number of customers, they expanded the items being sold to ~ 10,000 SKUs. Their biggest issue was that about 70% of their revenue was still being generated through milk and breakfast product offerings. They wanted to implement a solution that would help them in SKU visibility – that would eventually lead to a more diversified revenue stream.

Not just in-app recommendations

We suggested taking a leaf out of Amazon’s practice of not only in-app recommendations, but also through personalized communication with the customers – through emails and SMS

  • Improved unique SKU visibility per user by 14%
  • Repeat sale of non-milk SKUs increased from 3.2% to 5.8%
  • Average basket size improved from 1.53 to 1.61

Historical purchase data and few other data points associated with purchase history were used to develop a collaborative filtering based recommendation engine.

Multiple attributes. Lack of clarity

The hyperlocal retailer was inundated with purchase and browsing data.  They were use association rules and market basket analysis to analyse the purchase basket and using it to reach out to the customers. This solution was useful in the shorter term when they were offerings limited number of SKUs. With addition of more than 10,000 SKUs, the biggest challenge for them was to make customer aware that they were selling more than milk and egg products.

Another unique element of their business model was the repeatability of purchase. They observed a clear trend of customers buying a few products at regular intervals e.g. a customer would buy 1 litre pack of a specific brand of olive oil every 3 weeks. They wanted the recommendation solution to incorporate product purchase cycle into the algorithm. We developed a constraint based recommendation engine solution for our client.

A wise business turns chance into good fortune!

Any business spends a lot on customer acquisition. And the cost of acquiring new customers is about 8x more than retaining an existing customer. Cross sell and Upsell are proven retention strategies that are used by offline businesses.

The issue with implementing a cross sell solution (similar to offline businesses) is the volume of products on display and the limited real estate for app based purchase platforms. Recommendation engine solves these problems for digital businesses by:

a. Improving SKU visibility

b. Suggesting products based on propensity of purchase [Cross Sell & Upsell]

c. Personalized SMS and email campaigns

These benefits make recommendation engine an indispensable solution for digital stores.


Beneficiary Management System to measure true reach of a non-profit programs

A pan India non-profit used data governance, data management principles to identify on-ground beneficiaries .

The pan India non-profit had more than 2 million members participating in different programs related to skill development, livelihood, poverty alleviation and vocational training. The programs were designed based on local needs, and often overlapped in scope and coverage. The biggest issue with the non-profit was that they were not aware of how many different families they were reaching, and if the same person was getting enrolled across different centres.

Mapping beneficiaries to social programs

The non-profit wanted a better understanding of whom they were serving and how many unique households were being served through the social programs

  • The unique id mapped beneficiaries with 99.8% correctness (pilot of 10,000)
  • Family trees could be generated using the unique ids
  • A web based interface with digital solution to register new beneficiaries

The solution was deployed through a cloud based server and could be accessed through the web interface from anywhere. Even at locations with no internet connectivity.

Multiple centres. High duplication

The non-profit had multiple centres across localities where people could walk-in, enquire about the programs and register themselves or their family members. Each centre was supposed to enter the details, including family details, home address and phone numbers. Each of the beneficiaries was provided a registration number, that had a high potential of getting misplaced or being used by someone else in the absence of verifiable records.

The biggest problem was the lack of understanding of why the beneficiary data was important and the need to maintain them. They were looking at the solution merely as a tool to measure the true reach of their programs and the number of households impacted. We presented how the beneficiary management system was a prelude to number of improvements that could be enabled – from program implementation to return on investments.    

Not just a data entry tool

Beneficiary Management System solved the primary requirement of our client to remove duplicates, identify the unique number of beneficiaries their programs were supporting. It also allowed mapping a family tree to measure the number of households they were serving. This helped them communicate better with the funders and improve program outcomes.  

Apart from above, the non-profit is now able to measure the impact of their program in a better way, through social return of investment (SRoI) measure. It has also allowed the non-profit to have a database of all individuals they have worked with, allowing them to create a vibrant community of beneficiaries. They have used the information to reach out to corporate partners for skill based job fulfilments. And lastly, the non-profit has now defined KPIs to measure their operational efficiencies & performance each month.

Text Analytics

A conversational bot that generates marketing spend plans using performance data

A media client generates marketing plans in minutes using text analysis (NLP).

The client is one of the Top 5 media agencies responsible for media buying and planning for their clients. Developing the media plan process was slow, manual and repetitive. Every time a new request came in, they would aggregate data from multiple internal sources, build the plan and share it with the client liaison. The whole process was notoriously slow. They were looking for a solution that could minimize their dependence on the data team and generate plans instantly.

Marketing spend plans delivered in minutes

The client wanted to serve their clients better, taking all the requests and necessary details through a conversational interface and minimize the time to serve

  • Average session time of 7 minutes on the bot
  • Average response time of 38 seconds to generate a marketing plan
  • Turn around Time (TAT) for generating marketing plans reduced by >80%

The conversational bot made the marketing planning process objective, automated and fast. The team can now have a greater focus on the strategy and campaign performance

Manual, slow and custom plans

The media agency had a defined process to generate media plans. Every time a new request from a client’s planning team came in, the entire process of registering the request, getting the right data and plan customization was triggered.

Not only being slow, it created issues with manpower management, multiple back and forth with the clients and less than optimal turn around time for the customers.  Their biggest issue with creating an automated platform was the multiple versions of data stored across systems. Serving multiple clients through same systems was slowing down the data pull requests and impacting database performance. All these was leading to a manual, inefficient and utterly complex planning process. They were looking for a solution that was usable by non-technical users, allowing them to focus on strategy and performance.

Machine learning based marketing spend planner

The conversational chatbot automated the queries that were being sent to multiple databases – reducing variability with better optimization and faster results. The media agency was able to differentiate itself, with faster turn around time with natural language processing (NLP) based custom marketing plans.

The media agency is now able to:

a. Generate marketing plans within minutes leading to higher focus on strategy and execution

b. Standardize and create efficient process for the marketing planning process

We built a solution that the experience of interacting with a human media planner, blending in the benefits of a NLP driven conversational chatbot.

Digital Analytics

Customer analytics game plan for Telcos

Data is bread and butter for the telecom industry. An industry that generates huge volume of data has arguably had a difficult and limited success with analytics. The performance and data processing requirements for telcos is huge – handling millions of calls per second, billion of events per day and ability to manage millions of devices at a time – and to ensure high standards of service & network quality – with user expectations of zero downtime. Managing these infrastructure and customer expectations is a gargantuan task – and pose unique advance analytics and machine learning challenges.

The success of analytics in telecom hinges on the basic challenge of keeping customers happy. Analytics and insights have a key role in keeping customers satisfied through two key KPIs:

a. Increase Average revenue per customer (ARPU)

b. Improve customer satisfaction and reduce Churn

Any analytics strategy for telcos should ultimately lead to achieving the above two KPIs. And some of the solutions to achieve these are:

1. Customer Churn Analysis: Churn is in indicator of customer satisfaction, continued patronage and the growth prospects. Real-time data is most useful in churn prediction, as it is based on the most recent customer data rather than end of month churn reports. Deploying a churn prediction model based on real-time data, with right offers would allow retention to be faster and more effective.

2. Customer Satisfaction Scorecard: Customer satisfaction is not a static metric. It changes with each interaction a subscriber has with the telecom network. The customer satisfaction should therefore be generated in real-time – allowing telcos to identify subscribers whose satisfaction score is consistently going doing, and those below the acceptable threshold – to make the right intervention and reduce the risk of churn

3. Call Centre optimization: For customer care team to be successful, the support team should have access to subscriber data, as well as all data and resources required to address subscriber queries and complaints. Machine Learning based algorithms can be used to provide real-time feedback on an ongoing interaction, and generate recommendations and answers for the customer care agent to resolve issues faster and more effectively

4. Competitor Analysis: Telcos should not look at subscriber acquisition, retention and customer experience in isolation. Competitor activities, subscription plans and marketing activities need to be monitored, and strategy to counter competitors should be dynamic. Machine Learning based algorithms allow processing of unstructured data – campaign monitoring, subscription plan offers etc. to develop an effective competitor monitoring and insights strategy

5. Social Media Insights: Subscribers now leave a lot of footprints and crumbs of insightful data on social platforms – while interaction with friends or common interest based communities. Social Media should not be looked with a myopic view of managing brand reputation by responding to tagged subscriber complaints. The social platforms have a lot to offer – and telcos should leverage these platforms as a source of rich and valuable data

6. Personalized Targeting: A lot has been said about personalization. While being intuitive, it is not an easy task to achieve. Telcos need to develop a recommendation system by combining subscriber and subscription plan data, along with behavioral and demographic info to identify subscriber needs – and then reach out to them with targeted and personalized communication to show value, care and ability to serve them with a personal touch

None of these should come as a surprise. A customer is relaying her life story 24*7. Just imagine the richness of data and the endless possibilities it brings. All that remains to be done by telcos is to make use of such incredibly rich data.

Customer Analytics is the key to higher ARPU and reduced churn

Customer data and insights generated through these data are of no use if the customer is dis-satisfied or had already unsubscribed. Timing is the key, and this can only be achieved if you have proper data infrastructure and analytical solutions to generate insights at a quick pace. Telcos should not wait for the month end reports to act on customers – with advanced analytics and machine learning based real time decision systems – telcos can leverage the full potential of their data

Reach out to us and make use of all that advanced analytics, machine learning and AI have to offer – to create a family of happy and loyal customers

Xtage Labs is an advanced analytics and machine learning based decision insights company. We work with businesses to derive insights from data, and improve the decision making processes.

Get in touch with us to find out what we can do for you.

Marketing Analytics

Paid, Owned & Earned Media

If you are into marketing, you must have caught the buzz around paid, owned and earned media. If you are not, and wondering what the fuss is about, stick with us and you will know everything about them. Think of them as three pillars of digital marketing – that contribute to the complete digital marketing strategy. It is important to note that the three are not completely disjoint, and there are some overlaps between the three. It is therefore useful to understand the difference and how to measure your efforts in each one and come to understand how they affect each other.

Paid Media

Paid media is as the name suggests – any marketing that has been paid for. For offline marketing, this would include TV advertisements, Print Ads, Radio Spots and OOH among multiple others.

For digital marketing, there are a few variations:

a. Pay Per Click (PPC) Ads

b. Search engine Ads

c. Promoted Content on Social Media

Owned Media

Owned media, as the name suggests is the content created by a brand and published through a channel owned by the business. This includes the website, blogs, ebooks and white papers. Podcasts are a newer medium of owned media. Brand owned social pages also come under owned media.

Owned media is focused on content driven marketing, and SEO is the primary focus of owned media.

Earned Media

Earned media is the content and conversations generated by third party around your brand and products. These content is generally published on third party platforms e.g. a news channel talking about how your brand donated $s to the government emergency fund for Covid. Generally, earned media is driven by your paid and owned media marketing efforts, and the activity is picked up by some news website, blogger or a journalist. Product reviews, social media mentions and press coverage are included under earned media.

Of all the three, earned media is most impactful, as it is generated by people who talk about you – and have their own sphere of influence and trust. It is therefore better placed to generate organic reach for a brand. Just imagine a potential buyer of a smart phone looking at product review websites, and reading positive feedback of your product OR someone asking his friend for a suggestion on best phone to buy, and he/she recommends your brand.

When planning a marketing strategy (primarily Digital), it is important to understand the role of paid media, owned media and earned media in the marketing mix. And as with any investment, your marketing mix effectiveness should be able to measure the return on investment (RoI) from each – to optimize the future media strategy.

Write to us if you want to measure the effectiveness of your marketing efforts in terms of paid media, owned media and earned media.

Xtage Labs is an advanced analytics and machine learning based decision insights company. We work with businesses to derive insights from data, and improve the decision making processes. Get in touch with us to find out what we can do for you.

Data Science

From 0 to 1: Implementing Data Science solutions

Before we get into ‘How to implement data science solutions’, let us formalize the definition of data science:

Data Science is all about using structured and unstructured data to identify trends, validate hypotheses, and predict future outcomes. Communication of results through charts and graphs is a part of data visualizations.

It has been well established now that data driven insights can provide businesses with the tools and methods to predict future trends, events, and behaviours. For example, banks could use data science to predict the customers who are more likely to require a home loan or the customers who are more likely to miss a credit card payment. Similarly, CPG brands can use their past advertisement data to measure the sales generated by their marketing spends, and optimize the marketing budget to achieve maximum return on investments (RoI). You can check a few more case studies here [ Resource Center ]

With all the buzz around data science & machine learning these days, it is very easy to go overboard – without you and your organization being ready to unlock the true potential of data science. Here is a checklist of items before embarking upon your data science journey.

1. Define the problem: Identify the problems you want to solve using data science. You should know that data science is not a magic wand that will make all your problems disappear. Data can answer almost all questions – but for that to happen, you need to ask the right questions

2. Data science cannot replace choice: One of the reasons data science can never be 100% machine-driven is that decision making is always a selection between choices – that can at best be rank-ordered, but there would always be an element of philosophy, reason, experience, future plans in making a selection of choice today.

3. Estimate return on investment (RoI): Estimate the amount of benefits that you expect to draw from data science. It is not always easy to express the benefits of data science in terms of dollars. You need to identify the right metrics to measure the benefits.

4. Build a roadmap: Once you understand the benefits of data science, develop a clear roadmap of how you want to achieve those goals – whether you want to build an in-house team, or you want your data science tasks to be completely outsourced. You could also choose a combination of the above two.

5. Siloed or Integrated: You also need to define whether your organization is going to pursue a siloed approach to implementing data science or whether you want to view it as an organization-wide effort. Both the approaches have their pros and cons, and a rational decision, which works best for your organization, needs to be taken.

6. Data Governance: Define clear data governance rules, including data ownership, architecture, policies, data quality, rules for resolving data related issues and policies for data management. Your business might have access to private data, which needs to be protected. Proper data security, confidentiality and access rules need to be defined.

7. Graphical Analysis: A lot can be inferred by looking at simple one-dimensional graphs and cross-tabulations. Simple graphs can help you spot anomalies or identify trends. In the age of big data, do not ignore the power of simpler methods.

8. Identify experts: Not all people are apt in handling and interpreting data. Identify people within your organization who are more comfortable than others in handling data. Data science is always contextual, and the more experienced a person is within your organization, the better he/she can contextualize data science.

9. Engage Data science Consultants: While it is a nice idea to use in-house expertise, asking for external help can reduce your turn around time. Even if you have sufficient expertise, sometimes, having an external perspective helps you see things differently.

10. Test, Learn & Modify: Any successful strategy needs to evolve with time. The same is true for data science. You should not expect to use the same techniques or the same model over and over again. The problems need to be revisited over time for getting the best return out of your data science exercise.

Today, most businesses view data science as an essential capability. The reduced cost of data storage and drastically improved computation infrastructure is changing business models. The businesses that will be able to unlock the key to data driven insights are going to win.

Xtage Labs is an advanced analytics and machine learning based decision insights company. We work with businesses to derive insights from data, and improve the decision making processes.

Get in touch with us to find out what we can do for you.

IoT Analytics

Industry 4.0 – Quō vādis?

Industry 4.0 is the newest phase of Industrial Revolution being brought through faster wireless communications, real-time data streams, machine learning and AI driven automated decision systems and inter-connected devices and machines. It is also referred to as Industrial IoT or smart manufacturing. Industry 4.0 is revolutionizing decision making through real-time insights across production, demand forecasting, supply chain management and inventory management.

Evolution of Industry from 1.0 to 4.0

Manufacturing has evolved over the three centuries, and there have been four distinct phases of industrial transformation:

1. Industry 1.0: The invention of steam engines, and evolution of manufacturing from manual labour to optimized machines driven by steam powered engines. Happened in late 1700s and early 1800s

2. Industry 2.0: Brought about by the introduction of steel and use of electricity in manufacturing. Mass production concepts like assembly line formulated. Happened in early 1900s

3. Industry 3.0: Began with the introduction of computers in the manufacturing process. Focus shifted from the use of analog to digital technologies. Started in late 1950s

4. Industry 4.0: Started with the use of inter-connected machines (called Internet of Things or IoT). Has allowed connecting physical and digital machines. Started in 2010s

What is changing OR going to change in Industry 4.0?

Industry 4.0 is changing the manufacturing processes. We present a few use cases here:

1. Supply Chain Optimization: Manufacturers have the ability to integrate their manufacturing process with the supply chain platform. This provides them real-time insights, market trends and demands to deliver products at a faster and cheaper rate

2. Asset Monitoring: With enhanced tracking and monitoring capabilities using wireless connectivity, video based intelligent feeds and location based real-time data, manufacturers can identify supply chain issues and asset quality risks to optimize asset transfer, disposal and adjustments in real-time

3. Predictive Maintenance: Every manufacturer relies on machines, and the machines require regular maintenance to work efficiently. The different maintenance strategies are:

a. Reactive Maintenance: Repairs and maintenance done on machine failure

b. Preventive Maintenance: Failures are costly, and maintenance done at regular intervals to prevent failures

c. Predictive Maintenance: Condition based monitoring of operational parameters to predict failures, and carrying out the maintenance just before a failure is predicted to happen

A close up of a logo

Description automatically generated

Benefits of Industry 4.0

We list some (not all) benefits of adopting Industry 4.0 capabilities to your business:

1. Gives you the competitive edge

2. Increases collaboration and promotes shared decision making

3. Predictive analytics using real-time data reduces inefficient preventive maintenance costs and downtime

There are many other benefits, most of which do not fall under our skillset.

To outperform in today’s business environment, manufacturers need to embrace Industry 4.0 or risk falling behind.

Xtage Labs is an advanced analytics and machine learning based decision insights company. We work with businesses to derive insights from data, and improve the decision making processes.

Get in touch with us to find out what we can do for you.