There are many issues that impact customer retention for businesses and require warranty data analytics. The cost of neglecting these can be enormous and in many ways is directly proportional to the size of a business. For example, for large automotive companies which sell millions of vehicles per year, a single percentage point decrease in customer retention can mean millions of dollars of lost business and untold loss in brand image. But this does not mean that small and medium manufacturers do not face this problem. While the absolute scale of the problem (in dollar figures) may not suggest a massive issue, relatively speaking, these same problems could devastate a smaller company.
On the other hand, small and medium businesses (SMBs) also do not have the internal and external complexities that plague bigger companies. For example, take the issue of warranty data analytics: large companies have to deal with dozens of current and legacy claims processing systems. If they sell globally, they have to customize their products (called homologation) and that results in numerous variations in assemblies and subsystems which need to be monitored. Further, when it comes to extracting insights from customer or service technician feedback, they have to deal with tens of different languages and cultures. All these making the process of applying unified analytics to extract key indicators all the more challenging.
So what can SMBs learn from their larger brethren when it comes to utilizing analytics for product quality? There are several commonalities and a few differences. Let us look at common factors.
No matter what the size of the business, the underlying problems (in this example let us focus on warranty data analytics), are structurally similar and can be addressed by the same tools and techniques.
1. Data types: There are really only two types of data – structured transactional data which is typically housed in databases and unstructured but rich data which comes in the form of customer comments, dealership service entries, and more recently social media exchanges (more relevant for a B2C company rather than B2B, business size does not matter).
2. Visualization: Almost always the first step is to express the data in visual terms. This simple process itself can address many issues which remain unseen at the data level. Simply visualizing data by region, for example can uncover answers to vexing issues. As an example, quick visualization by geography may explain why some components are more prone to replacement than others – perhaps they all come from the same manufacturing plant.
3. Classification, Association and Forecasting: These three techniques form a Pareto principle in analytics techniques, in that they can account for (or be applied to explain), 80% of warranty and recall problems. A time series analysis of a particular problem (example, a failing window seal) may reveal seasonal patterns – colder winters may be less friendly to some particular types of products and thus we would see seasonal spikes in the number of instances reported. Association analysis (popularly called market basket analysis), may be used to understand (and predict) problems that typically occur together. Another example, early wear out of clutch plates required for manual transmissions, in urban markets due to heavy stop-and-go traffic.
4. Text Mining: Compared to structured data, text mining is more challenging to successfully apply, but it can be richly rewarding. Document clustering and similarity mapping can reveal a wealth of hidden information from service logs and customer complaints. But it takes a slightly higher level of analytics maturity to use this to the greatest advantage.
At the end of the day, SMBs have access to all of these tools and techniques and today are in a very strong position to take advantage of them. Assuming that the data already exists, adopting these methods is a gradual and incremental process, but something that is relatively easy to get started.