BPO Industry- New Turf for Analytics (Contd…)

December 12th, 2012

These last couple of weeks have been a little crazy for folks here at Gmid Associates, throwing us off the track on almost everything apart from the project deadlines. I’ll give the same excuse for the delay in this post (and hope no one finds out the real reason!).

We pick up the discussion exactly where we left off in the last post – on how analytics is changing the value proposition in BPO industry. So it will be a good idea to read the last post before you proceed. Here it goes.

Clustering/ Segmentation and other descriptive analytics techniques

BPO companies sometimes work under such dynamic environments that the data they need to work on doesn’t stay with them for long. Although at most times they might get the analytical inputs for their campaign strategies from the clients (e.g. segments/ treatment plan etc.), even then most of the times the segments themselves are so broad based that further fine-tuning of strategies becomes imperative. They can use their own historical data repository to identify most important of the data variables, slice and dice the data further and overlay their operational team expertise on this. This leads to further improvement to their clients’ strategies and tighter operations- reducing costs, increasing revenues and over delivering on client expectations.

Depending on the business vertical of the process involved, various descriptive analytics techniques are leveraged for impriving efficiency- from RFM analysis (recency-frequency-monetary) to clustering schemes around important business attributes- a number of tools are used to add layers of intelligence over the data and processes and run tight segment focused campaigns. Data mining techniques are also used to understand strengths and limitations of their resources for optimal alignments and resourcing.

Predictive Analytics

Predictive analytics solutions are very relevant in all BPO business simply because they help optimize processes by predicting end results (e.g customer response in case of a collection call) accurately and hence enabling optimal strategy implementation. The predictive models are developed on historical client shared data and the transactional data that a BPO provider generates by running these processes. To explain with an example, A BPO firm is asked to run an insurance retention campaign on a portfolio of 1 million customers. By developing a predictive model to this end helps them greatly in identifying various distinct segments of customers- those who just need a gentle reminder message to pay up, those who need intensive treatment on why it makes sense for them to pay up and those who probably would not pay up at all and higher intensity treatment on them would rather work negatively for brand equity. All these segments of customer need different treatment strategies. Based on business needs and relevance, predictive analytics solutions ranging from -regression models to time series forecasts to neural models- from simple to most advanced techniques are being used in BPO business today.

Recent advancements in the field of predictive analytics have considerably improved applicability of high end statistical techniques into real time outsourced processes. Solutions today are leveraging machine learning capabilities to identify changing trends on dynamic datasets under process and enabling actionable strategies in real time. These solutions present efficiency improvement areas previously unknown- all because of smart usage of analytics and computing techniques.

These new generation BPO analytics solutions are set to change the rules of the game in the BPO industry. As businesses continue to look out for more than just “cost reduction” and “operational efficiencies” from their BPO relationships, the BPO industry will have to try and make the best possible use of the most important asset they have- data gathered during years of their client relationships. And analytics will help them do just that and more.

Companies are getting more comfortable with engaging one BPO partner for end to end business processes; this gives providers a deeper access to their clients’ work flows and business. This gives them a perfect platform to leverage the data and this domain knowledge- probably in a better situation than their clients themselves- to generate actionable analytics and deploy it on business processes for everybody’s benefit -from the clients to the providers to the end customers. This is the next logical step for the BPO industry. One that makes the world a smarter place.

- posted by Mudit Chandra


  • Share/Bookmark

Tags: , , , , ,
Posted in Analytics in BPO Industry | 1 Comment »

BPO Industry- New Turf for Analytics

November 29th, 2012

In the last ten-fifteen years, the concept of Business Process Outsourcing has revolutionized the global business scene. The trend started with noncore business processes getting outsourced (primarily call centers)and as the outsourcing ecosystem keeps evolving and companies get more confident in the skills of suppliers- as well as- suppliers get deeper knowledge about their clients’ business functions; companies across business verticals are today feeling compelled to outsource more and more “core” processes. There are many reasons for this trend. Suppliers are getting so efficient in the business processes that they can deliver even better than what their clients could ever manage themselves. The value added services that BPO providers provide give their clients distinct competitive advantages that are completely new to them. And most important of these competitive advantages come through exploiting Business Intelligence and Data Analytics.

It’s difficult to find any company today that doesn’t outsource its mission critical business processes …and for good reasons. In fact, with the fast pace that global BPO marketplace is evolving, advantages that outsourcing firms cited historically- cost savings and increased efficiencies- are mandatory requirements now. As the bar of expectations keeps getting raised, companies are looking for solutions that can enhance the value of their existing BPO relationships.

For BPO firms, keeping ahead of the competition means they have to use new tools and techniques to move up the value chain. Applying analytics on the large pools of profile and transactional data accumulated during their long term client relationships coupled with acquired business knowledge gives them actionable insights to be deployed on the processes- giving their clients both top and bottom line benefits.

Clients are slowing realizing the fact that they are not leveraging the full value of huge amount of data generated in their back end outsourced processes. They know this will help them serve their customers better, run their operations more efficiently and increase shareholder value. They are now expecting these values from their services providers.

Although usage of analytics in BPO operations is increasing every day, there are still some road blocks to this. These are inertial responses to a new paradigm that any industry faces. BPO industry in general works on thinner margins and the added costs of analytics tools and skill sets could be a hindrance. Also a provider may not have the depth of visibility to a client’s business functions that is necessary to transform analytical insights into actionable intelligence. Finally, many BPO firms have not been great at generating and warehousing clean (workable) data simply because it didn’t matter to their business previously.

But as there are more and more examples of companies using analytical tools to their advantage, this initial shyness is slowly going away. There has also been a trend of bigger BPO providers acquiring analytics service provider firms. This gives them twin advantages- one, they can deploy the analytics on their existing business processes and give their clients distinct competitive advantages. Two, they can tap into their BPO clients’ needs of analytics services. With the advantage of already having relationships and proven track records, they have more chances of winning analytics business over other KPO firms.

The point is – BPO industry is slowly warming up to the power of analytics and we thought it would be apt to cover how analytics is shaping the mother of all outsourcing industry- Business Process Outsourcing next in our ongoing series.

We will start with the most basic usage of data mining and analytics techniques in the BPO industry and try to move up the ladder as we go.

To start this conversation, it is imperative to mention that BPO business provides a synergistic environment to add value through analytics because of the very nature of this business- There is already a strong foundation of data around client processes and so is a deep appreciation for efficiency improvements.

Data Cleaning and Mining

A lot of data that BPO firms get from their clients, as well as what they generate during operations contains a lot of errors and garbage values. Their ability to clean that data efficiently and use the analytical insights gained to drive performance can give them distinct pole positions in the market place.

During one of our segmentation engagements with a BPO major, we noticed that the contact data provided by their client companies was unsuited to machine processing and needed much cleaning before we could do any segmentation and campaign design. Using the data as-is (or with some manual cleaning) was giving a serious blow to their business efficiency and profitability. We used text mining techniques to substantiate the errors, and found that over 95% of errors were caused by a small number (~200) of text entries. These bad entries were deleted/ modified or replaced (whichever the case) to eliminate these errors. This number was further reduced to under 0.5% by iteratively repeating this process within a short span of time. A further study of their legacy databases pointed out that most of these errors and the data quality were quite standard in nature and were present in almost all data dumps they got from respective clients, so we embedded these data mining steps into an easy to use automated tool which could be used to treat these data sets every time a new data dump came in. So now when this much cleaner data goes into segmentation schemes, the results are much better, relevant and ready to use. And the resultant segmented data can be readily mounted on their automated dialing machines for treatment. This gives them a lot of “low hanging fruits” for process improvement- Manual cleaning is not needed, so shorter turnaround times and zero resource dependency by virtue of standard process. Clear tab on quality- the tool publishes a report after each run on what level of cleanliness it was able to achieve on the input data. Based on this feedback, the tool can also be improved in future.

This example just gives a fair idea about how simple data mining techniques could be used to increase BPO efficiencies manifold. And things like this will soon be a standard, if not already.

This is it for now. In the next post, we’ll talk about how clustering / segmentation schemes and other descriptive analytics techniques are used in the BPO industry followed up by predictive tools and solutions.

Feedback/ comments most welcome.

- posted by Mudit Chandra

Image courtesy- good_day@flickr


  • Share/Bookmark

Tags: , , , , ,
Posted in Analytics in BPO Industry, Uncategorized | 4 Comments »

Analytics in Customer Life Cycle Management

November 16th, 2012

So now the account has been sourced and the bank has a new customer. What happens next varies depending on the type of product/ service. But there are a certain things that may happen irrespective of all that. For example- the account may churn out after a limited time thereby denying the bank precious revenues and a relationship, the account may go delinquent on its payments resulting in losses for the bank. There are also many hidden opportunities that the bank has no way to figure out, for example –potential cross sell/ up sell opportunities- it might happen that the same guy needs some other banking services and competition’s marketing tapped into it before our bank could. The bank was at an advantage of already having a relationship with the customer thereby enabling them an access to precious information about him. All these opportunities directly result in revenue leaks and direct losses, and analytics can help fix these.

Churn Reduction

To go through all the above discussed cases one by one and how analytics could save the day, we start with Churn Reduction. With the help of predictive analytics on relationship data and profile information, we can develop churn prediction models to help identify customers with high possibility to churn in the near future. We can then treat them to segment specific retention programs and contain churn. This information can also be used to develop lifetime value prediction models which can be clubbed with churn prediction models to identify potential high revenue customers who have high propensity to churn. They can be exposed to targeted programs to earn loyalty and maximize lifetime revenues.

These are just some examples of smart usage of analytics in the banking operations. New ideas are emerging everyday on how to best leverage analytics at each and every node of business process flow and a discussion of this broad nature can never cover everything. The objective is to give an idea of the applicability (and necessity!) of data mining and analytics in the day to day operations of a bank.

Collection and Recovery

Moving on, we come to delinquency management aspect of credit risk management. Delinquency prediction scorecards are pretty standard in the banking industry now. Such model scorecards help in identification of potential delinquent accounts and that enables the delinquency management teams to timely treat these accounts with appropriate collection strategies to minimize delinquencies and to use collection resources optimally. A new advancement in this area has been to help identify the reasons behind a delinquency. To explain- a delinquency might be because of many reasons- intentional, habitual or genuine problem among many other things. Analytics and data mining are being used to segregate these groups and to treat them differently. For example, an intentional fraud customer (should ideally be identified at the time of application itself) should be exposed to the most severe collection treatments as soon as possible and the credit exposure should be minimized. While a genuine delinquency customer (lost a job, hit by recession etc.) wants to make the payment but is unable to. This customer needs to be treated differently- probably given comfortable payment options; he will bounce back and be a good and loyal customer again. By using analytics and data mining on historical collection data, banks also fine tune treatment strategies, for example – which segment should be exposed to what treatments (message/ calls/ personal visits etc.), and with what intensities / at what time of the payment cycle etc..

Analytics also comes handy in the non performing assets (NPA) management/ recovery processes. By leveraging historical data and decisioning tools, banks can identify ‘juicy’ pieces from their NPA portfolio and maximize recoveries and ROI on recovery budgets.

Cross selling

Now, coming on to cross sell and up sell opportunities. In today’s tough business environment, acquiring a new customer is way more difficult (read costly) than selling to an existing customer. And given that you already have a relationship with the guy, you have access to a lot of information about him that your competition doesn’t, and this gives you a clear cut advantage. Analytics helps banks tap into this opportunity and make the best of an existing relationship. Banks use cross-sell models to sell their products to their other portfolios, and to up sell their higher range products to existing customers. Analytics helps them identify who among their existing customers are most likely to buy a different service- based on customer profile, preference and service features. This way you don’t need to market your service to the entire universe, but only to those who are most likely to buy. This helps in – marketing ROI maximization, enhancing customer loyalty and unleashing the full potential of a relationship.

Also, a lot of banking is now real time just – ‘a swipe of card’, ‘a click of  button’ based allowing no time to the bank for judicious decisions –whether to authorize the transaction or mark as risky/ fraudulent and stop it. Analytics helps institutions to develop tools to assess transactions in real time and take a call on whether the transaction is fraudulent or not. This comes under Fraud Risk Management function of the bank.

With this, we come to the end of application of analytics in the consumer banking industry- the first part of this series. We have broadly tried to touch upon most of the important dimensions of tactical and strategic decision making process and how analytics is being used to gain a competitive advantage. As we said earlier, this is not all, in fact far from it. But it gives you just a fair amount of idea on how analytics is shaping the way business is run in today’s world. And how it continues to evolve every day. And with high emphasis on ‘focus on the core business’ and let the specialists do the rest, it only makes sense that more and more businesses will look out for outsourcing tasks from their data analytics and data mining functions.

Hope you liked it. Do post feedback/ comments/ questions.

Stay tuned for more!

Relationship Churning Prevention Campaign

- posted by Mudit Chandra


  • Share/Bookmark

Tags: , , , , , , ,
Posted in Analytics in banking | 4 Comments »

Analytics in Retail Banking

November 9th, 2012

Banking industry has always been a very early user of new technologies, often breeding ground, and analytics is no exception to this. Banks were the first to leverage their historical data –by applying statistical techniques and using computing resources to solve real business problems- in short, using analytics. Retail banking is the business of banking services and solutions aimed at the individual. Since everything at a retail bank takes place on large data sets, it is the perfect breeding ground for analytics.

Let’s go through this from a bank’s customer life cycle perspective- starting with when the bank markets its services to a person walking down the road, to that person coming at a branch to sign up, then enrolling as a customer, going through the relationship life-cycle based on the relationship/ product, then may be churning/ going delinquent or termination of relationship.

Marketing

Marketing is the first logical entry point for analytics into any industry simply because analytics lets you unlock the full potential of your marketing spends, track and optimize your ROI from campaigns. Banks collect a lot of customer data during their interactions. By analyzing that data, they understand customer buying patterns, linkage with customer profile and demographic details and then use this information to tap into cross-selling opportunities, prospective new market segments and align offerings to the group of customers most likely to buy them. All these improve your marketing efficiency and increase ROI.

Sourcing/ Acquisition

Once the bank has made a successful sales pitch to a person, he walks into a branch and applies for enrollment. If the product or service involves credit, the bank would want to make sure that the guy has enough credit worthiness to cover the expected exposure. The bank collects a lot of information about him as a part of the application process. Then using analytics on his historical behavior, profile, earnings, borrowings and all such data, they assess his credit worthiness and fraud propensity leading to either his enrollment as customer or rejection. This comes under the application fraud risk management function of banks.

Credit worthiness assessment is also used to determine specific features to be offered (e.g. credit limit in case of a credit card) at the time of sourcing. Credit worthiness is a dynamic attribute about the customer and it keeps getting updated with new information coming in during the relationship thereafter and remains a very important tool for the credit risk management function of the bank.

Hope you enjoyed the post. In the next entry, we will cover various applications of analytics during the customer relationship life cycle.

The analytics dude

- posted by Mudit Chandra


  • Share/Bookmark

Tags: , , , , ,
Posted in Analytics in banking | 9 Comments »

Analytics is a necessity

November 7th, 2012

In the last decade or so, data analytics as a function has crossed over from being a nice-to-have business practice to a need-to-have one.

In a technology driven, global market place, information has come at the forefront of retaining competitiveness. And it’s not just robust data management practices and easy access to data warehouse that will do the trick for you. However important and prerequisite these are, but it is analytics-the science and art of making sense of that data- that will give you a competitive edge.

Normal Distributed Random Stuff

With ever growing competition and globalization, business processes and decision variables are getting more complex, and decision making is becoming more complex every day. And this is where analytics has taken a leading role in assisting decision making process by highlighting hidden trends, predicting future events and unlocking the full value of a company’s collective history to take right strategic and tactical decisions.

As the cost of data storage and data management keeps reducing, the variety of data producing sources keeps increasing, data analytics is set to permeate each and every step of decision making in business process flow.

At the same time, outsourcing of data analytics services has also slowly graduated from the “concept selling” stage and come at the forefront of knowledge process outsourcing (KPO) industry. And analytics outsourcing is set to dominate the growth in KPO market for the next decade.

In light of such a backdrop, we thought it appropriate to launch a series concentrating on application of analytics in the decision making process, industry by industry. We will try to cover as many industries as we can, and we will start with banking industry- the early users of analytics- with a focus on retail banking.

Feedback/ questions most welcome. We will try to take mid course corrections during the series based on the feedback.

- posted by Mudit Chandra


  • Share/Bookmark

Tags: , ,
Posted in General | 15 Comments »

The RIGHT to target the RIGHT customers

August 3rd, 2010

Are there really ‘right’ customers for a given product or service ? And if yes, what about the ‘right’ of the wrong customers to be targeted as well.

One of my friends started his career with a renowned analytics company. His first project was to find out the best customers from a pool of many physicians to be targeted for a newly launched drug. His first reaction was – “How mean! What about the doctors who will be left out?”. Notions like ‘Business exists for society’ and ‘Treat everyone as equals’ filled his mind at the time when he should be most concerned with running the regression equations. As he became sad, the urge to put his best in the work made him do some readings and meetings with wise people. That ultimately did dispel the dark clouds from his life and though he is now living happily (the ‘ever after’ way) with the regressions and correlations of the world, it was realized that many wanna-be statisticians in the business world face similar questions and thus remain stuck somewhere.

In an attempt to increase happy number crunching in the world, we dedicate this blog to our version of the big picture. Let’s start from the basics of the basic:

Business Analytics 101

  1. Business occurs when a person (or animal involving at least one human being) is in need of some goods. Some other person (or plant/animal/an inanimate object involving at least one human being) produces the desired good. The two meet and the producer gives away the goods to the needy in exchange of some ‘money’ (referred from now as the Big M).
  2. Money making is good- So long as the needs exist, producers will produce and thus will give the goods for:
    (a)  free as a noble gesture,
    (b)  free in the fear of getting killed/molested,  or
    (c)  the want of tons of the big Ms. However -
    - Not everyone will at all times be utmost noble so society cannot survive on case (a) alone.
    - The law will restrict case (b) to the minimum.
    - Hence, case (c) will always exist. As it is based on mutual consent so is not evil.
  3. Money saving is good – Producers will always need raw materials to produce goods. These raw materials will require the M to get acquired. Furthermore, people involved in the production will be distributed some of these Ms to celebrate their contribution and to manage their greed that will erupt from time to time. So there will be fixed costs that producers will always have to incur. Besides, there will be uncertainity in the world which the producer only has to cope up with. As the wise man plans for the worst time, so the producer will want to have some of the Ms placed somewhere safe. Furthermore, because of need/greed more people will acquire skills to produce goods hence there will be competition. Competition will lead to innovation. Innovation will require hiring and tolerating bizarre people who want more Ms . Saving money is, thus, a good idea.
  4. Cutting costs -  Less consumption is saving.  Saving money on people will make them unhappy which not only will increase unhappiness in the world but will also lose talent. In the new era, production costs are almost optimized so there is less room available (though analytics has done wonders here as well). Luckily there is one caveat – producers go to the customer and advertise. This consumes lots of Ms. Hence any money saved in this piece without losing on customers is a fantastic idea.
  5. Right/Wrong Customers – As marketing is an en masse exercise, so fundamentally there will be people who will either:
    (a) use your product whatever happens , or
    (b) not use your product whatever happens, or
    (c) are confused and thus are most viable to marketing attempts.

    The wise producer will make category (a) happy , will not go to category (b), and will try to convince category (c). While right customers will come from categories (a) and (c), the category (b) will certainly be the wrong customers. Hence there is a big need to save the M in spending efforts on category (b) customers.

  6. The art in Analytics - Some boys and girls have studied books on consumer psychology, statistics, and probability. They know that people repeat their behaviors even when they change.  They work hard and segregate (a)s, (b)s, and (c)s from the customer pool and study their past performances of buying goods on advertising attempts. They have also realized that (c) forms the largest group and there are sub-groups which exist within. They will strategize efforts to the right customers and will (i) cut costs and (ii) increase revenue.
  7. Save the world – Money is saved, right customer is informed and she has got the goods, wrong customer is not troubled and the time is saved. Everyone is happier – the world is now a better place to be in.

____

Appendix (The big M) – There is a field in data analytics that deals with charging the optimal price from the customers so that the overall profits are maximized. A quick count of the keyword ‘M’ in the article above will demonstrate possibilities of optimizing money transactions in a general business scenario. Now incorporate possibilities of vendors comparison, profit sharing, retention bonuses, cross-regional expenditures and currency conversion, open market, new technologies, etc. This will give a glimpse of different ways to optimize money.

- posted by Hemant Kathuria


  • Share/Bookmark

Posted in Uncategorized | 3 Comments »

Gmid Associates: From an Idea to Reality

July 8th, 2010

Things culminated for real when we were asked to provide services for a company that was into a lot of data. They had basic things in place – systems, people, reporting MISs. Still the senior management had the bigger picture missing of ‘what exactly is happening’. This had to be found out because the organization was working in an area famous for it’s tight margins. Fixing leakages would make sense as that will more than double up the profits.

We agreed, worked part time, and went deep into the system to deliver our first project – a product that integrates different sub-units and uses data for the management to see the bigger picture. It was a hit, and the value it created led to us getting introduced to other areas of their work with the expectations to replicate magic. We then pooled more members into these projects and stretched ourselves as best as we could. Now that we were beginning to play our flutes in the various projects it started making us more convinced about the power of Analytics in transforming businesses.

While we were having extraordinary experiences in this direction, there were interesting stories happening elsewhere. This was a time when all four of us were doing fabulous in our otherwise professional lives. One of us got promoted in a super fast way that too in a set up where only rare people can manage to do so. Another one cracked his summer internship with an Internet major and was thinking about placements he would choose at IIM. The third partner scored phenomenally well in the GMAT and was exploring MBA schools for his applications.  However, the will to create an organization, and not just work in one, allowed us to look deeper in our hearts and dream big. And when we did that we realized that a company focusing on developing ‘Analytics Tools and Practices’ is the way to go.

It was clear to us that the commerce in India has now come of age to contain the Analytics piece as a ‘must have’ entity. This is different than the earlier scenario when the rules of probability and statistics were only buzz words with the big consulting companies whose focus was to deliver one-offs assignments at an exorbitant price. We knew that in the globalized world, where only the best will survive, solutions of these kind need to lie at the heart of the corporate machinery. We also realized that, for the work to succeed in this industry, it is vital to seamlessly integrate the products with the existing processes. The Predictive Decisioning should not just be the icing on the top but the baking powder itself that gels the cake.

The first assignment came in September 2009 after which one of the partners quit his job to test the waters. With coming of more assignments, other partners soon followed and June 3rd was the day we registered ourselves as a company. The mission has begun and we believe in the power of what we do. Where it is existent, we enhance. Where it is not, we build from scratch.

We consider ourselves fortunate as we work with two very senior members of the financial world. They also happen to be our mentors and are extremely helpful in guiding us on the projects we pick. The journey so far has proved to be very exciting as we got to touch new frontiers in both work and learning. And while we zestfully tread the path, we keep in mind that the best is yet to come.

- posted by Hemant Kathuria

especially now when in the globalized world only the best will survive

  • Share/Bookmark

Posted in General | Comments Off