The vanishing human in customer care
As companies race to automate customer support, speaking to a real person has become a luxury—sometimes even a paid one
When was the last time you tried calling customer care? Was it hard just to locate ‘customer support’ on the brand’s app or site? Did you have to jump through hoops just to get the option to speak to an agent? Select random issues because yours wasn’t on the list? Click through so many automated menus that by the time you reached someone, you had forgotten how you got there, to make the process less painful the next time around? And when you did succeed, were you frustrated even further because it wasn’t a human agent but an AI voicebot you spoke to on the call? Welcome to customer service in 2025.
It’s a routine Yash Bhardwaj knows all too well. The 27-year-old tech entrepreneur from Mumbai regularly finds himself in this maze with hyperlocal delivery apps. “I probably have to pick 3-4 false issues for some kind of ‘talk to agent’ option to even appear."
His recent encounter with Uber’s customer care system, however, made food delivery apps seem efficient in comparison. “I took an Uber rickshaw ride and paid via card, but the driver said he hadn’t received the money. I opened a ticket through the app, but it had a very AI response. Nobody got back to me, and I couldn’t find a way to contact them," Bhardwaj says. The only feature designed to draw immediate attention was the SOS button, but his issue, he felt, fell more in the “annoying" category. Ironically, even the SOS button doesn’t connect you straight to a person. It leads you through multiple prompts, sharing your live location with authorities and offering options to either call the police or connect with a support agent on the safety line.
How did we get here?
Less than five years ago, calling customer care and speaking to a human agent was the norm. Today, companies have turned it into the last resort, with many of them deliberately making it harder to reach a real person. Some are even turning direct human assistance into a premium feature. Consumers are now paying for Swiggy Black, Uber One, and even Meta Verified, not entirely for faster delivery or social media perks, but for something that was once free: quick and reliable access to human support.
How did we get here? “Contact centre human agents are expensive. Then there’s the issue of training and high churn," says Gaurav Singh, founder of Verloop, a customer support automation platform. A regular customer support ticket ends up costing a brand between ₹40 and ₹50, he says, which includes multiple overheads. For online travel agencies, the cost can shoot up to ₹130– ₹300 per ticket.
There’s one more factor at play here than just cost. “Some people just don’t want to talk," Singh observes. The tech-savvy customers prefer efficient, automated interactions. “When sector margins are low, the tendency to hide away from direct human contact gets higher," Singh explains. “And when customer literacy is low, companies avoid calls altogether because most of them are about very basic issues." He cites the example of an e-commerce company catering to India beyond the metros, where nearly one-third of calls are simply: “Have you received my order?"
In e-commerce, Singh notes, 18–20% of orders in newer companies lead to customer support queries, a figure that drops to around 12% for more mature businesses. “In the near future, sub 10% of all queries will require human intervention," he predicts, with Gen AI expected to drop current standards by another 50-60%. Singh sees the future in voice chatbots. “Eight out of 10 clients prefer voice chatbots over text chatbots now. Customers are dealing with a Whatsapp chat fatigue and prefer voice," he says. The technology is getting so sophisticated that soon “less tech-savvy users will not know it’s a machine they’re talking to."
But there’s a catch. Shaheena Attarwala, a UX design leader from Bengaluru, points out that “training efficient, contextually aware AI agents is also time- and cost-intensive". The current AI-led solution isn’t mature enough to handle real context, she says, and in some cases, voicebots actively block access to human support.
Attarwala recently called Airtel Xstream Fiber broadband service about an internet issue. Every time she requested to speak to a human executive, the bot replied: ‘You are talking to our customer service executive’. “I couldn’t give any details about my problem, and I couldn’t speak to a human at all."
Prasanna Venkatesh, director of product design at a consumer internet company in Bengaluru, points to Amazon as another company that is particularly frustrating in this aspect. “There is absolutely no human involvement. I have had incidents where my items were not delivered, and there is no way for me to raise a request for missing items."
The absence of human support becomes most pronounced, he notes, during moments of genuine crisis. “A woman finding her cab driver’s behaviour unsafe and dangerous has no direct access to a human support agent and needs to go through the usual automated process." The problem is particularly acute for older users, adds Venkatesh. “For products catering to senior citizens, human access becomes a deciding factor."
Venkatesh believes there’s a deeper issue at play. “Tech founders live in a bubble where emergent AI and technology practices will help them build companies with as few people as possible, and they leave no stone unturned to have a lean team to build products." He’s been part of projects to build tech-first support systems, and “the pace at which near-zero human intervention has come into the picture has baffled me frankly," Venkatesh says.
The cost of losing the human touch
Not all companies can afford to abandon the human touch, though. “The higher your LTV (lifetime value) for the brand, the faster the human support because retaining you matters," says Singh from Verloop. At IKEA India, “the customer goes through three simple IVR (Interactive Voice Response) steps- choosing a language, selecting the purpose (sales & resolution), and moving directly to a live agent queue," says Sunil Kumar, the company’s country manager for remote customer meeting point. “This ease in experience is also the reason why our remote channel today is not limited to after-sales support but also a growing sales and design channel, contributing about 5% to total India sales and nearly 10% to our overall B2B sales," he adds. Mint reached out to every brand mentioned in the story. Only IKEA responded till press time.
“In the banking, finance and insurance space, Gen AI bots are limited because the sector depends on strict rule-based decision-making; you can’t afford the hallucinations AI is prone to," says Jacob Joseph, VP of data science at CleverTap, a customer engagement and retention platform. He cites Klarna’s 2024 experiment as a cautionary tale: after replacing 700 agents with an AI chatbot, the Swedish fintech company rehired humans when nuanced financial queries required empathy and judgment. “The best model is hybrid," Joseph says. “For instance, in a wealth management app, a human advisor can use a bot to analyse a user’s risk profile and offer more meaningful assistance."
Ultimately, the issue isn’t just about efficiency or cost, it’s about trust. As Attarwala puts it: “Humans trust humans. We know someone is accountable, and we can put a name to the person we spoke to. With an AI agent, whom do the customers hold accountable?"
Customers may not actively notice when reaching support is easy. But they are likely to remember when they needed help and couldn’t get it. “I genuinely feel companies who are substituting stuff like this to cut-costs probably produce low quality products that have a high support need," says Bhardwaj from Mumbai.
In the race to automate everything, companies risk learning an expensive lesson: it only takes one moment of terrible service to break a customer’s trust forever. And unlike an AI voicebot, that’s one thing no algorithm can fix.
topics
