India's largest platform and marketplace for AI & Analytics leaders & professionals

Sign in

India's largest platform and marketplace for AI & Analytics leaders & professionals
Become a Thought Leader
Become a 3AI Member

Sign in

3AI Digital Library

< Go Back

Improving Customer Experience – AI/ML Driven Issue Resolution

3AI August 8, 2022

Author: Ashish Vashishta, VP – Technology, eClerx

Customers across industries typically face issues with their purchased products/services. A troubleshooting system which can use latest AIML technologies to resolve these issues faster and more effectively can lead to a much better customer experience.

As an example, a customer may have purchased a bundled product offering – A home security system along with internet connections, phone connections and cable TV connections. These devices may fail at times for numerous technical reasons during their lifetime. When they call customer support for resolution of these issues, customer support executives are not able to resolve all the issues at times.

For unresolved issues, a home visit of a technician is booked. This leads to multiple problems:

  • Degraded Customer Experience
  • Cost of Home Visit
  • Human Contact

The coronavirus crisis of 2020 made the matters worse. The customer experience had gone for a toss with non-availability of technicians and increased service delivery costs.

Service providers must solve the challenge of optimizing business processes and providing superior, efficient customer service at minimal operational costs to stay competitive.

It is important to solve this problem as it is applicable to virtually every B2C industry. If the customer issues can be proactively solved, it can lead to multiple benefits.

  • Improved customer experience – Proactive issue resolution over call leads to a much better customer experience rather than making the customer wait for issues to be solved, either through the home visit or by other means. This ensures that the customer doesn’t have to wait for engineer to visit when issue can be resolved remotely.
  • Ensuring minimal operational costs – An engineer’s customer site visit [Truck Roll] cost is 2X of what it used to be pre-covid times.
  • Ensuring customer and technician safety – AI/ML and Analytics driven issue resolution and visit avoidance ensures minimal human to human contact in these pandemic times.

DATA ENGINEERING PLATFORM

The answer all the questions listed in the previous section, lies in a holistic data platform that can not only ingest and analyse data but also has ML and Computer Vision components built in along with outbound dialler component.

CORE COMPONENTS OF THE PLATFORM

Data Extraction, Collation and EDA: The platform should be able to integrate and extracts customer trouble tickets from client systems. The integration can be a loosely coupled integration using REST APIs or tightly coupled using direct backend connectivity. This is governed by the systems the platform is integrating with. The platform then evaluates the data against quality dimensions – Accuracy, Completeness, Consistency and Uniqueness

For one client, where we deployed a similar platform, the monthly trouble call volume was 150,000+ tickets. This multi-dimensional data [Ticket meta data, Agents, Service, Equipment, Customer] had ~300 variables. We were able to ingest the raw data automatically using REST APIs

Computer Vision: The image and video analytics engine will then be responsible for capturing equipment images from the data management tool to analyze the equipment damages. Specifically, below aspects to be monitored:

  • Identify the equipment and its type – Is customer using client certified and approved device/equipment. Extract the device details.
  • Identify signal strength on device – Is device receiving adequate signal strength to operate at optimal level
  • Analyze physical damage to the device – Try to auto-analyze device image and compare with similar images in the model to see if there is physical damage to the device in which case truck roll cannot be avoided

Machine Learning: Logistic and Neural Net to assign ticket cancellation probability to tickets. The tickets with high cancellation probability then gets prioritized by the system for proactive issue resolution over call ahead of truck roll.

For the ones with low cancellation score are to be routed to downstream system that handles truck roll

Predictive Dialler: AI based predictive dialler to place outbound calls to the customer. This ensures that customer support agent’s time is utilized only when a human answers the phone. This leads to efficient operations as any calls which are going to voicemail or are unanswered are not routed to customer service agent.

SOLUTION DESCRIPTION

CONCLUSION

We believe that an integrated data platform that can address customer device issues and help resolve those efficiently will go a long way in improving the customer experience and thus leading to higher customer retention.

For one of our customer where we did a similar deployment we observed –

  • The customer experience had improved as was reflected by VOC (Voice of Customer) scores. On an average VOC scores improved by 8%
  • Faster resolution time with average time to close tickets reduced by more than 40%

Title picture source: freepik.com

    3AI Trending Articles

  • New Customer Expectations Driving Digital Innovation

    Featured Article: Author: Nitin Srivastava, Director – Data & Analytics, Advance Auto Parts India To ride this tide of disruption, it is crucial to understand the pulse of customer behavior and demand and integrate it into a digital first approach. One of the key changes brought about by the COVID pandemic was the acceleration of […]

  • Automated Quality Excellence Framework for Data Science Modelling

    Featured Article: Author: Abhishek Sengupta, Walmart Global Tech India Introduction:  A necessary but often tedious task data scientists must perform as part of any project is quality assurance (QA) of their modules so as to prevent any unforeseen incidents during deployment. While Quality Assurance and Excellence is quite prevalent in Data Engineering, QE in Data […]

  • Way to make object-recognition models perform better

    Adding a module that mimics part of the brain can prevent common errors made by computer vision models. Computer vision models known as convolutional neural networks can be trained to recognize objects nearly as accurately as humans do. However, these models have one significant flaw: Very small changes to an image, which would be nearly […]

  • Give AI a ‘positive’ spin: Google tells its scientists

    Google has reportedly been telling its scientists to give AI a “positive” spin in research papers. Documents obtained by Reuters suggest that, in at least three cases, Google’s researchers were requested to refrain from being critical of AI technology. A “sensitive topics” review was established by Google earlier this year to catch papers which cast a negative […]