The company offers a suite of solutions that enables brands to manage customer interactions through multiple channels, thereby improving overall customer experiences, loyalty and ultimately, Customer Satisfaction (CSAT) scores.
THE CHALLENGE
In today’s social media world, brands have to constantly stay connected to their evolving base of global customers. While the traditional channels of customer communication such as manned telephone helpdesks and emails are still used to gather feedback and complaints, people are increasingly switching over to social media to interact with their preferred brands. This happens through people following brand accounts, following influencers or brand ambassadors, sharing brand content, liking posts, writing reviews, sending messages to company accounts, tagging companies in posts, reporting issues, or replying to others who have reported various issues etc.
Based on global figures, the average person has 8.4 different social media accounts across all platforms. Facebook has over 2.9 Bn Monthly Active Users (MAU), while Instagram has over 2 Bn MAU. In comparison, Twitter has a user base of 368 million MAU.
The company had a portfolio of offerings which manages customer interactions for various brands through chat, mobile applications, social media, smart speakers, and messaging / SMS. The average annual volume of these managed customer interactions is over 1 billion. The social media component of their end-to-end solution allows brands to collect live social media customer sentiment, while also providing a workflow for their agents and teams to further interact with their customers. However, this solution had some drawbacks. The set of tools used to build the social media management solution was not in line with their enterprise technology stack. The solution was also severely constrained by the data volumes which could be concurrently handled. This limited the overall value proposition of their combined portfolio of customer interactionssolutions.
Hence, the company needed a technology partner to rebuild this social media management application on their modern technology stack and integrate it with their overall solution suite.
T HE SOLUTION
The team of architects, data engineers and data visualization specialists from Prescience Decision Solutions analysed the existing social media management application to identify any gaps in its overall functionality. For each brand related message on Facebook, Twitter or Instagram, the solution handled it through the following components,
- Information Collection
- Information Processing
- Information Handling
- Reporting & Analysis
For the Information Collection component, the team built the required connectors for live feeds of all brands related content from Facebook, Twitter, and Instagram. Since both Facebook and Instagram are owned by Meta, the team developed a new application with the required authorizations from the parent company. For Twitter, the team created a paid account and then developed a new application to get the data feed.
For the Information Processing component, all the live social media data is consolidated into a platform queue. This pipeline is fed into the company’s proprietary Machine Learning (ML) model on which sentiment analysis is performed. This Machine Learning model automatically assigns the required tags to each message in the queue.
For the Information Handling component, the tagged messages in the queue are transferred to the specific teams of agents belonging to various geographies for the Marketing department, Customer Care department etc. These teams will then continue the conversations with the corresponding customer on the social media platform and will address any concerns or collect their detailed feedback, as required. The messages which do not require any action are tagged accordingly, so that their agents do not unnecessarily spend time on handling them.
For the Reporting & Analysis component, our team built a set of management reports and operational reports. These management reports capture Key Performance Indicators (KPIs) on trending topics, current customer sentiment, platform-based interaction volumes etc. The operational reports capture KPIs on agent performance including number of successfully handled interactions, escalations etc.
The overall solution was built on,
- Kafka
- RabbitMQ
- Java
- Elasticsearch
- Looker
- Kubernetes
THE IMPACT
The team from Prescience Decision Solutions successfully rebuilt the data handling and reporting components of the social media management application, using the same technologies as the rest of the company’s enterprise product suite. With this combined offering, the company now offers a robust end-to-end customer interaction solution which connect to all major social media platforms while handling large volumes of live data feeds.
