Frequently Asked Questions
Current Warranty analytics
1. What is warranty data?
Warranty data is created when a vehicle or piece of equipment experiences a part failure or reduction in performance while under warranty. After the claim is submitted, the component is repaired and a log of all repair details is made.
2. What influences warranty claim behaviour?
Several factors including the timing of a claim submission, which is influenced by the type of component failure; the driver or equipment user; proximity to a dealership; the time to the next scheduled service; the time to the end of warranty period.
3. Why do I need a warranty analytics system?
Because it will enable you to extract valuable insights from large bodies of warranty data that will allow Quality, Warranty and Engineering personnel to detect problems rapidly and so prioritise these issues effectively. Better use of warranty data can ultimately create significant cost savings.
4.What’s wrong with our current warranty analysis methods?
Most warranty systems tend to be retrospective, looking at historical failure frequency or warranty spend to determine what the biggest issues are likely to be. However, they fail to forecast systematically to understand how established and emerging issues will develop.
5. Why does this make it harder to detect issues early?
Because components experience most of their warranty activity later in the warranty period. Using existing methods, issues with these components remain hidden until the current failure frequency or cost become large enough to stand out from the rest – but by then it’s too late. From just a handful of claims, we have made it possible to detect problems as they emerge so your Quality team can take action.
6. Why do current warranty analysis methods make it hard to prioritise issues effectively?
Component issues are currently prioritised according to failure frequency or spend to date, rather than by undertaking predictive analysis of how an issue will develop. The failure to forecast systematically across the entire vehicle and component portfolio means that emerging issues go unidentified for too long and cannot therefore be prioritised effectively.
7. Our warranty reporting is inconsistent – can you help?
Yes, by creating a simple-to-use reporting system. Too many Excel spreadsheets can cause confusion and inconsistencies across different factories within the same company. We make the process clearer and easier for all stakeholders to follow.
A new approach
1. What can high-quality analysis tell me about my data?
It will enable you to detect issues as early as possible so you can prioritise correctly. Once an issue is detected, systematically analysing the data for a country, region or climate zone, as well as examining correlations with weather and other data sets, will let your engineers identify the root cause more quickly.
2. Why is forecasting so important for early issue detection?
In order to stop problems escalating your system needs to identify issues to flag up based on very small numbers of claims. If historic activity is the only data used to look for issues then by definition there needs to be a large volume of claims before the issue can register. Systematic forecasting allows you to understand how a small volume of claims will develop and whether this will become a significant issue.
3. Why is systematic forecasting important for correct prioritisation?
To prioritise correctly, forecasts are required on all components to the same relative point in time so they can be compared and then effective Pareto analysis applied.
4. What are the limitations of parametric forecasting?
Parametric forecasting (curve-fitting) such as Weibull has no historical context so forecasting for components with low initial claims activity that only develop later into warranty issues. As a result Weibull underestimates the scale of the problem because it lacks a mechanism that can anticipate this type of behaviour.
5. Why is non-parametric forecasting suited to detecting issues early?
Because it uses historical claims behaviour and patterns to inform analysis of current claims activity. This enables forecasting with historical context and allows it to achieve accurate forecasts from a small number of claims.
How does We Predict deliver Warranty analytics as a service?
1. What is analytics as a service?
We don’t sell hardware or software so there is nothing to install or implement and no need for you to employ analysts or data scientists. Instead, we will use your data to build an analytics system that will answer the questions your business faces today – and that will evolve to answer the questions it will face tomorrow.
2. What data do I need and how often do I need it?
We require warranty claims and vehicle sales data, along with details of dealer locations, relationships and descriptions of parts and part numbers. Normally we will update monthly or weekly but daily is possible if required.
3. What can We Predict offer an OEM?
You will receive a comprehensive analytics service that will turn large volumes of data into valuable information that can be used in functions including Quality, Warranty, Financial Accounting, Supplier Recovery, STA, Dealer Management and Sales.
4. What can We Predict offer to a Tier 1 supplier?
You will receive a complete analytics service that turns large volumes of data into valuable information for Quality, Warranty and Purchasing teams.
5. What is Indico, where is it built, how is maintained and deployed?
Indico is our unique analytics platform which is a combination of a sophisticated SQL database and a HTML5 visualisation system. It is built and maintained by We Predict and is deployed in the cloud offering secure, discrete login for our clients.
6. What is We Predict’s creative client relationship?
This allows all We Predict clients to benefit from the ongoing development of Indico. Any system improvements or analytical developments made for one client are available to all others so that you continuously benefit from the most advanced systems and methods that We Predict offers.