By Keegan Weber, Product Specialist - Engage
Over the past year, COVID-19 has forced consumers to rely more heavily on digital customer service channels – a shift that many large organisations weren’t ready for. We saw this when analysing South African banks’ social customer service during the early phases of lockdown, during which conversation volumes grew by 61% while banks’ response rates to customers fell by 39%.
In response to this increased demand for digital service and heightened consumer expectation for responsive service on social media, large organisations like banks and telcos have had to rethink their approach. The days of using community management tools to respond to social customer service queries are clearly behind us, but finding new tools to cut through all the social media noise is no easy task.
In the telecoms industry we found that only one in three mentions required a response from network providers. This means that two-thirds of online conversation about network providers was noise to social customer service teams, hindering their ability to prioritise the mentions which did warrant a reply.
For large-scale service providers, the reality is that sifting social media conversations from within all the noise is only half the battle won – the real challenge lies in identifying the ones that matter most. Generally speaking, these will be the mentions that pose a risk or consist of a service request; an acquisition opportunity; or a cancellation threat – all of which are seen as priority conversation and require a response.
Prioritising these customer interactions allows agents to serve the customers who matter most. For example, a premium subscriber or client who is threatening legal action or calling on others to boycott the business requires urgent intervention over a customer looking for a non-urgent and ordinary service request.
With this in mind, businesses need a tool that provides them with accurate data they can trust, which is where AI can often fall short. This is because AI has yet to master the complexities of online human conversations, which are typically filled with emojis, slang, sarcasm, innuendo and often multiple languages.
Until AI learns to deal with these sentiment-based complexities, a layer of human verification is required to manually review and tag each mention for its importance. This human touch ensures that users are delivered each ticket by order of priority and risk, which means less time for support agents to find the important mentions in a messy queue, and more time to respond adequately to customers who require a response.
By optimising workflow and identifying the priority posts that require an urgent response, tickets are also easily routed to the right agent in the team, saving the business time and ensuring that priority conversation does not go unanswered.