The first salesperson many shoppers meet now lives in a chat window, and it does a better job than many call centre scripts.
When budgets are tight, people do not browse politely. They ask an AI what to buy, what to skip, which option is cheaper, and which brand is bluffing. Shopping starts here for more and more buyers, and the latest CX data shows the shift is already visible in the numbers.
The 2026 South African Customer Experience report, released recently and co-authored by Rogerwilco CEO Charlie Stewart, Ovatoyou founding director Amanda Reekie, and Julia Ahlfeldt of Julia Ahlfeldt CX Consulting, has tracked online and offline consumer behaviour since 2019. This year it surveyed 2,000 people online and 56 business leaders. Its bluntest finding is that 72% of consumers say their financial position has stayed flat or got worse over the past year, which explains much about the rise of AI shopping research.
Shoppers are already using AI like a comparison clerk
The biggest use for AI among consumers is plain information hunting. People use it to research products, compare prices, and pull together facts before they spend a cent. This is becoming the default first step.
The report says 23% of consumers use tools such as ChatGPT, Gemini, Claude, and Perplexity to compare products and prices, hunt for deals, and ask open-ended questions. That same behaviour showed up in Discovery Bank and Visa’s SpendTrend26 report earlier this year, which found that 40% of surveyed shoppers rely on AI weekly when deciding what to buy. The report called this trend disciplined and strategic consumerism. The reality is harsher: people are trying to stretch every rand.
AI use is also spreading into practical help. Consumers turn to it for writing support, troubleshooting, advice, and other tasks that used to be split between Google, a friend’s WhatsApp voice note, and a frustrated call to customer service. Charlie Stewart’s point is simple: buyers now expect information to arrive fast, in language they can use, and with enough detail to answer the question in front of them.
This directly affects brand visibility. If the answer is being shaped before a shopper reaches a website, the company’s digital footprint is no longer a marketing extra. It is part of the buying experience itself. A brand that cannot appear clearly in AI-generated answers is already losing the race before the contact centre phone rings.
Ahlfeldt puts it neatly: “AI is more than just another Google.” She is right. Google still matters, but for many shoppers it is no longer the first move. It is the backup plan.
The call centre is losing easy wins
One in four consumers now uses AI for help solving a problem. Business leaders are not seeing that clearly enough. The report says only 7% of businesses recognise that pattern, while 32% still assume customers would rather call a contact centre.
That gap should worry anyone running support, sales, or BPO operations. A chunk of the questions that once filled queues are now being filtered through a chatbot before a person ever gets involved. If the AI can explain the issue, compare the options, or suggest the next step, the human agent only gets the cases that are messy, emotional, or expensive.
The report also points to the next stage of this shift, which it calls agentic commerce. In plain language, AI does not stop at advice. It starts handling actions on behalf of the user, sometimes with very little human input. According to the report, 67% of consumers would be comfortable letting AI fill a shopping cart, 62% would trust it to place a meal-delivery order online, and around half would let it book travel or even a medical appointment.
This is a warning shot for customer service teams that still treat AI as a novelty sitting in the background. Once people are happy for software to take over routine purchasing and booking tasks, the old idea of the contact centre as the first stop starts to break down.
Human agents are not disappearing, but their work is shifting. Routine troubleshooting, basic order checks, and repetitive explanations are exactly the kind of jobs AI is beginning to absorb. The remaining calls will be harder, stranger, and more emotional. That requires a different skill set and different hiring.
Companies need to write for machines as well as people
Three-quarters of the businesses in the report do not have an active plan for how they show up in AI-generated answers. That number should make a marketing director look at the calendar and then at the budget line for content.
The problem is not only technical; it is editorial. If product pages are vague, pricing is out of date, FAQs are thin, and support articles are buried, AI systems have less useful material to work with. The machine will still answer, but it may answer with the wrong version of your brand.
Companies cannot rely on quick fixes when shoppers are already asking AI to compare prices, surface discounts, and explain the difference between one model and the next. The work now sits in the boring middle: cleaner product data, sharper descriptions, better structured FAQs, clearer delivery information, stronger review management, and a consistent presence across the web. That is where the answer gets shaped.
The old separation between marketing, customer service, and sales is getting messy. A product page is now a sales tool, a support tool, and an AI input at the same time. If those teams are not talking to each other, the customer still feels the gap.
The broader signal is already there. AI is becoming less like a shiny add-on and more like an everyday decision-support tool. It is front-facing for shoppers, not just a back-office toy for retailers. Brands that keep treating it as someone else’s problem will discover that the customer has already moved on.
Job seekers need a different shortlist
For people trying to get into support, BPO, or customer experience, this change is not a disaster. It is a filter. The easiest work will shrink, but the better work will be more visible.
The roles that grow from here will need people who can do a few things well. They will need to write clear content for FAQs, product pages, and help desks so AI can read it properly. They will need to watch digital performance and spot where a brand is being misrepresented online. They will need to read data, notice patterns in AI search behaviour, and turn that into better service design. They will need to handle exceptions, complaints, and escalations when a machine hits its limit.
That points to a different kind of contact centre candidate. Not just fast typing and a friendly tone, but clean judgment, structured thinking, and enough tech literacy to work alongside AI tools without panicking when the script changes.
For school-leavers and entry-level job seekers, the basics still matter, but they are no longer enough on their own. For mid-career agents, the smart move is to move up the value chain. Learn how content affects customer journeys. Learn how AI answers are built. Learn where a human still has an edge, especially in complicated or sensitive cases.
The labour rules do not vanish here. The BCEA still sets the floor, the CCMA still hears the disputes, and the National Minimum Wage still applies. But the people who get shortlisted for better roles will be the ones who can work inside this new shape of customer service, not the ones waiting for it to go back to the old one.
January and mid-year hiring spikes will still pull in applicants across Cape Town, Durban, and Johannesburg. The difference is that the shortlist will increasingly favour candidates who understand AI-assisted service, digital content, and escalation handling. The old “good voice, good attendance, learn the script” profile is being replaced by something tougher and more useful.
This is the job market shift hiding inside the shopping trend. Shoppers are training themselves to use AI before they ever speak to a brand. Companies are still pretending the call centre will catch everything. It will not.
