In Modern Support Experience, one of the greatest challenges that industries face is to identify the Root Cause and Symptoms of the high volume of issues coming in at an incredible velocity. With the gathered data from multiple data sources and immensely varied information, it gets harder to identify and establish the latent causes for problems.

Ascendo brings to you one of its most utilized and successful components, Auto Categorization, which is built on state-of-the-art Natural Language Processing techniques and Deep Learning algorithms.

Understanding the Problem:

Data in the real world is very convoluted and keeps evolving with time. As products and tools continue to improve, the number of issues and types of issues also increase. These issues keep changing with product changes and can evolve into something similar to what was faced before or extremely unique to how the product has evolved. Almost every industry today suffers from exponential growth in issues. The complexity of issues keeps raising the graph of data points. This pattern hardly stabilizes with growth in product stack and integrations.

Currently, support agents tend to spend more than 50% of their time evaluating the problems, let alone finding the right solutions. Historical information needs to be crystal clear to understand the data based on past information. Making past data expertly classified and immaculately accurate is a task that is extremely time-consuming and strenuous. It is not difficult just because of the enormity of the data, but also due to inconsistencies in human thinking. When issues are being described by the end customers, people describe problems in their own ways without adding any context to the components of those problems. Moreover, agents and other experts could think of multiple solutions, causes, and symptoms for the same problems which increases the complexity of normalizing your data, therefore introducing multiple ways to understand the same type of data points.

The standard way of solving this problem is to make use of expert knowledge. This method introduces multiple points of consideration, making the available solution hard to be implemented:

Thousands of historical data points are needed.

Each data point needs to be filled in with expert knowledge and accurately defined.

Even for a small support team, about 17.5 hours per week is spent manually classifying problems.

Multiple people talk about the same issue differently!

The list of problems keeps expanding as the product expands

Possibility of multiple unknown issues

Mapping new problems with the existing list of problems causes inaccurate classes of category assignment

Inaccurate learning

Manual work hurts getting the Voice of the customer feedback into the product in a timely manner

Not understanding issues in real time is an impediment to providing proactive support.

Ascendo Recommendation Engine:

Ascendo aims to make the support experience as easy as possible for its customers. The solution has to be simple, smart, efficient, fully integrated with data sources, and most importantly requires little effort and time for users. The solution also has to bring expert knowledge into learning and has to be a continuous learning engine. The feedback loop could simply confirm and enhance automatically created root causes that groups multiple problems into their own bucket.

Download the full whitepaper to read more on this.