Customer self-service is not a new phenomenon; it’s just much easier than in years gone by, writes Kelvin Brown, customer operations executive at Telviva.
Yet, despite this, everyone reading this will relate to being “trapped” in a seemingly never-ending loop in contact centres, pressing “1” for this, “2” for that, and when they finally, eventually, get to a human, they have to introduce themselves and their problem all over again.
When it comes to self-service, the South African banking industry is advanced compared to global standards, with managing cards, virtual cards, opening accounts, and fraud prevention amongst other functions. In the insurance sector, self-service is currently used for policy retrievals, claims submissions, and agentic-based call routing to replace complex IVRs. Internet Service Providers utilise portals that allow customers to manage password and account resets, perform feasibility, manage and track internet service orders, and change account and invoice information.
All companies delivering a service or product should already have some kind of self-service facility in place. Salesforce states in its “State of the connected customer” that 61% of customers would rather use self-service channels for simple issues, and that an average of 54% of customer issues can be solved by organisations that use it.
The rise of conversational artificial intelligence (AI) and Large Language Models (LLMs) may tempt businesses to rush to deploy new tools and virtual agents, but if it is not done carefully, they risk falling into a dangerous trap: using AI to build higher, more frustrating walls around their business, rather than building better bridges to their customers.
The true differentiator for modern self-service is no longer just having a digital interface. The key to a good CX is how seamlessly that interface mimics human interaction – not just in experience but also in efficacy. Traditionally, self-service has played a supporting role to traditional customer service, but this is shifting. When a customer can message or speak to an AI virtual agent and be served effectively, the boundary between self-service and traditional support blurs.
What does AI in self-service look like? Imagine a customer wanting to reset their password for an account. Traditional self-service may have provided a few channels for the customer to choose from, such as articles, a web chat or some kind of IVR-style menu choice on various platforms. All these would have delivered a rules-based experience with no doubt a few “I don’t understand your request” errors.
Adding AI changes this. When the customer calls in, he or she would be answered by a virtual agent on WhatsApp, voice or text. This virtual agent greets the customer in natural language and asks how it can assist. The LLM takes the customer’s request and, using automation tools integrated into the backend systems where the account and password reside, facilitates a security verification, an OTP and allows the customer to reset their own password.
Another great example of AI being used in a very practical way to enhance self-service, and one we have deployed and tested in our own Telviva ecosystem, is the use of our digital agent, Viva. This is not an automated bot, but rather an agent with the ability to offer self-service via the phone and, most importantly, to direct a customer to speak with the right person or team without the IVR headache of multiple layers of menus.
It is important to understand that AI is only as intelligent as the data feeding it. If the internal data is not organised and clean, the virtual agent will simply get it wrong, hallucinate and confidently deliver incorrect information.
Businesses should seek out partners operating within governance frameworks such as ISO27001 and who ensure the virtual agent’s “source of truth” is strictly structured, version-controlled, and clean. Businesses should ensure that the LLM is tightly restricted to drawing only from verified, public-facing company data.
When it cannot understand a query, a virtual agent should seamlessly hand the interaction over to a human agent. When this occurs, that agent must instantly receive the full context of the conversation. Nothing erodes customer patience faster than making customers repeat their problem all over again to a human after trying to resolve the issue with a machine.
The role of humans in a contact centre will change as virtual agents take on more of the high-volume, repetitive tasks. Humans will need elevated emotional intelligence and analytical skills to handle complex escalations and investigate non-standard issues. In addition to this, they will also need to master the art of querying AI assistants in real-time to find facts while critically verifying the outputs.
Measuring success
“First call resolution” and “speed of contact” metrics are outdated. When AI agents handle easy queries, and humans deal with more complicated tickets, penalising increased call times is counterproductive.
Rather, in an AI-augmented contact centre, success should be measured on:
- Zero Call Resolution: Measuring the percentage of interactions successfully resolved start-to-finish by self-service tools.
- Segmented CSAT: Tracking customer satisfaction scores for AI and human channels separately to avoid a blended view that masks systemic friction.
The future will no doubt be anchored in AI-augmented automation, yet customer service must remain deeply human. The organisations that earn customer loyalty won’t be the ones with the flashiest AI tools; rather, it will be the businesses that use smart, AI-augmented automation to respect their customers’ time and human empathy to respect their intelligence and frustrations.