We have moved

Please go to our new website and update your bookmarks

Churn Heat Map

As markets become more mature, competitive and undifferentiated, many companies find themselves struggling with high customer attrition (churn). Companies try to cover up by acquiring new customers by means of profit-eroding attractive acquisition offers, which in many cases, encourages churn behavior as competitors adopt similar acquisition strategies. In order to address the root cause of customer attrition, it is vital for companies to identify the “dissatisfiers”. The “dissatisfiers” may result from several factors (dimensions); for e.g., poor products, service, lack of trust, company policies, perception/brand issues or dissatisfied segments and in most cases – a combination thereof.

Since the “dissatisfiers” can exist on several dimensions, analysis of historical data can provide invaluable insights. However, the real challenge of churn related data analytics does lie in the execution of the analysis, but rather in the generation of insightful interpretation with a big-picture view of things and actionable remediation plan. Going by the 80-20 rule, companies stand to benefit the most by identifying the “dissatisfiers” that have maximum impact and are most addressable. Due to complex nature of such analysis, a structured and comprehensive approach to data analytics is required.

Diamond has developed a technique, “Churn Heat Map”, which is a useful tool that can allow companies to identify churn drivers in an efficient and reliable way. The tool is used to analyze historical customer attrition data to generate a color coded heat map of churn rate modulated by the severity of churn and volume on a grid of churn drivers and customer segments. The churn drivers and customer segments are chosen from standard attributes in order to address specific needs of a problem. The rules for color coding are also customizable (for e.g. red color may indicate above industry churn for a segment). The “Churn heat map” is a useful in customer attrition remediation projects in a variety of industries – telecommunications, financial services (credit cards, banking, brokerage etc.), media, and many more industries plagued by churn/attrition problems. Such a tool provides the following benefits:

• Allow rapid hypothesis generation
• Help identify relatively addressable and significant pockets of churn (low hanging fruit)
• Serve as a reporting tool for visually monitoring the efficacy of churn remediation activities
• Leverage experience and knowledge across products, markets, and geographies by constantly enriching the model with additional categorization variables

Several off-the-shelf software such as SAS Cube, SAS Enterprise Guide, SQL Server, SGI MineSet 2.5 can be customized and leveraged to implement such a tool with limited development effort.

A sample application of such a tool when applied to a telecom churn data analytics is shown below:

The above figure shows the logical flow of transformation of historical marketing data into a churn heat map, which can then be readily leveraged to generate hypotheses, which once validated can yield actionable recommendations leading to early wins in a churn remediation initiative of a company.

Future of Business Intelligence

An interesting response to an article on the future of business intelligence that came out last week. The blog talks about issues along the BI value chain and the need for innovation. However one issue might be the highly fragmented value chain itself and therefore some vertical integration might also make economic sense.  One interesting factoid that jumped out:

Most data warehouses are built by hand, which suits systems integrators just fine (all those yummy billable hours) but does not serve customers well. TDWI reckon an average data warehouse takes 16 months to deploy, USD 3 million to build and costs 72% of its development costs in support every year.

Customer Education as a Retention Strategy

Companies are pouring resources to achieve better customer service, which ultimately translates into increased loyalty. However, if the product fails to meet customer’s needs, satisfaction and hence the loyalty will be limited. We recently helped a client in diagnosing the high churn levels of newly acquired customers. After performing customer research on this segment, we realized that only about a quarter of the customers understood how their chosen product features compared to the spectrum of available offerings by our client.

Thus, many customers who were dissatisfied by their specific product features switched providers (rather than moving to a different plan with their existing provider) as they were unaware of the alternatives. This highlights the need of better customer education and experience during the sales process which is aligned with meeting the needs of the individual customers.

Loyal Customers

“When making a purchase, a consumer has a choice between using frequent-flier miles, cash, or some combination thereof. Which will he or she choose? Another consumer has an opportunity to participate in a special program to get a free car wash after paying for a certain number of washes. What’s the best way for the car-wash owner to motivate the customer to participate?”

Here is a very good summary of the latest in the area of customer loyalty in academics. Article and associated papers talk about efficacy of ‘combined currency’ (dollars and miles) programs, ‘artificial advancements’ and ’status’.

As marketers use more complex loyalty schemes and techniques, the importance of testing marketing campaigns becomes very important. Proper campaign management processes and governance structure within the marketing department helps the companies to structure test and roll out the most profitable marketing campaigns to the customer base.

Location, Location, Location

When Scottish hotelier Charles Forte was asked about the secret of success of his hotels, he is reputed to have said, “Location, location and location.” . This credo also seems to apply to almost all companies who do some form of brick and mortar distribution. Be it a retailer who wants to improve its productivity or companies who want to optimize the spend on distribution and associated trade marketing.

Companies have historically focused on removing inefficiencies from supply chains by optimizing inventory and logistics. Store location which falls at the intersection of operations research and marketing has only recently received some focus. The major challenge in being good at locating new stores or evaluating the future potential of existing stores, is integrating the customer segmentation strategy with the distribution data.

 

We recently helped a client to evaluate the retail distribution strategy and came across this problem. We solved this problem by creating a market potential index, by identifying the concentration of client’s most profitable customer segments, for every county within client’s footprint. We could then map the number of points of sale (POS) per each county to evaluate whether the client stores are in ‘desirable’ locations.

We found that 24% of client’s retail investments were made in locations that were less than desirable, and realignment of POS to desirable locations would save the company millions of dollars a year, even by conservative estimates

This call may be monitored

Despite all of the money spent on customer satisfaction and CRM systems, many companies are no closer than they were 20 years ago in understanding the specific drivers of customer dissatisfaction.

Poorly conceived self-help automated customer care strategies, without a proper understanding of the root causes of customer calls, have further exacerbated the problem and consumers have started protesting- sometimes in unique ways:

The gethuman project is a consumer movement to improve the quality of phone support in the US. The most popular part of the gethuman website is the gethuman database of secret phone numbers and codes to get to a human when calling a company for customer service.

Very few call centers have properly designed tools in place to identify the root cause of customer calls, at a level which is actionable. Without understanding the root cause of customer calls and quantifying their impact, any self-help strategy project is starting off on the wrong foot. At Diamond, we have been very successful in designing and implementing a statistically valid call monitoring module to many of our projects. It ihelps to dentify and quantify the various major buckets of root causes of customer calls and forms the starting point of further analysis to seek actionable solutions (see example)

We have been using this data collection technique in problems which need us to get to the bottom of a certain customer reaction (churn, downgrade, buy). Sometimes, listening to a customer call can provide the insight which a lot of smart graphs can miss.