WeBuyCars settlement: the trust test for online sales

R2.5 million. That’s the administrative fine attached to the National Consumer Commission (NCC) settlement with WeBuyCars—plus R3.42 million in refunds to 31 buyers, and a binding commitment to change how the business sells.

If you run an online store, a marketplace, or any digital service in South Africa, you should read that as more than “a used-car story”. It’s a loud signal that digital-first companies are being judged on the same thing customers judge them on: trust, clarity, and the quality of the post-sale experience.

I’ve found that most businesses obsess over acquisition and treat complaints as an operational nuisance. Regulators don’t. Customers don’t either. And this is exactly where AI in e-commerce (and digital services) can move from “nice-to-have automation” to real risk management.

What the WeBuyCars settlement actually tells the market

The point isn’t only the money. The point is that accountability is becoming enforceable, measurable, and non-negotiable.

The settlement—confirmed as a consent order—requires WeBuyCars to:

• Pay a R2.5 million administrative fine
• Refund R3.42 million across 31 consumers
• Revise terms and conditions to align with the Consumer Protection Act (CPA)
• Run a consumer awareness programme on rights and obligations in pre-owned vehicle purchases
• Create 300 jobs over five years to improve customer service capacity

That mix matters. It says: regulators want better outcomes, not just a once-off penalty.

The used-car angle is obvious. The digital commerce angle is bigger.

Used cars are a high-complaint category for a reason: information asymmetry. Sellers often know more than buyers. The same dynamic shows up in:

• Online electronics and appliances (hidden defects, warranty disputes)
• Subscription services (confusing renewals, cancellation friction)
• Marketplaces (third-party sellers, unclear accountability)
• Digital financial services (opaque fees, “it wasn’t us” blame shifting)

When contracts or processes try to limit statutory rights, you don’t just get churn—you invite scrutiny.

A modern e-commerce brand isn’t judged by its homepage. It’s judged by what happens after something goes wrong.

Consumer rights are now a customer experience requirement

Under the CPA, consumers buying from a dealer are entitled to goods that are reasonably suitable, of good quality, and free of defects. If the product fails those standards within six months, consumers can demand a repair, replacement, or refund—and sellers can’t contract out of that.

Here’s the practical implication for online retailers and digital services: your returns, refunds, repairs, and complaint pathways are part of your compliance posture.

Where companies get this wrong

Most problems don’t start with “bad intent”. They start with systems that are designed for the business, not the buyer:

• Terms and conditions written to reduce refunds rather than clarify them
• Support teams measured on ticket deflection instead of resolution
• Slow handoffs between sales, ops, finance, and service
• No consistent evidence trail when disputes escalate

When the experience is messy, customers escalate. When escalations stack up, regulators notice patterns.

December reality check: peak sales + peak disputes

It’s 24 December. South African businesses are sitting in the most intense period of the retail calendar: promotions, delivery delays, and “I need this sorted before I travel” urgency. This is when:

• Return windows get tested
• Product issues surface (especially with big-ticket items)
• Contact centres get flooded

If your customer service can’t keep up in December, you’re not just losing revenue—you’re accumulating risk.

How AI helps prevent the complaints that turn into consent orders

AI won’t “solve compliance” on its own. But it can reduce the exact failure modes that trigger disputes: confusion, delay, inconsistency, and missing documentation.

1) AI makes policies understandable (and harder to misapply)

Most policy disputes happen because customers don’t understand the rules—or staff interpret them differently.

Practical AI uses:

• Policy copilots for agents that answer, “What’s the correct remedy under our policy and the CPA for this scenario?”
• Plain-language rewrites of warranty/returns terms (with version control and approvals)
• Dynamic FAQs that adapt to what customers are actually searching and asking

The goal isn’t to hide behind the policy. It’s to make the policy predictable and consistently applied.

2) AI triage shortens time-to-resolution (the metric that really matters)

Customers don’t escalate because the first reply took 30 minutes. They escalate because nothing gets resolved after 10 days.

AI can:

• Classify tickets by urgency and likely remedy (repair vs replacement vs refund)
• Detect high-risk cases (repeat contacts, legal language, social media escalation)
• Route disputes to senior agents early, before frustration compounds

A simple stance I like: speed is nice; certainty is better. AI supports certainty by recommending next steps and ensuring nothing falls through the cracks.

3) AI creates an evidence trail that stands up in disputes

When a complaint becomes formal, your strongest asset is a clean record: what was sold, what was promised, what was disclosed, and what remedy was offered.

AI-enabled workflows can automatically attach:

• The product listing as seen at time of purchase
• Inspection notes, images, condition reports (for marketplaces and refurbished goods)
• Chat transcripts and call summaries
• Timeline of actions taken and offers made

This isn’t about “winning” against the customer. It’s about proving fairness and consistency—two things regulators look for.

4) AI improves transparency before purchase (where trust is won)

In the used-car context, “undisclosed defects” is the recurring theme. In e-commerce, it’s often “not as described”.

AI can reduce that gap by:

• Flagging risky product descriptions (missing specs, exaggerated claims)
• Detecting inconsistent images and mismatched variants
• Prompting sellers to disclose common defect categories (battery health, refurb grade, missing accessories)

If you operate a marketplace, this is especially important: the platform is often where blame lands, even if the seller is third-party.

A practical playbook for South African e-commerce teams

If you want the short version: build for fewer disputes, faster remedies, and clearer records.

Step 1: Audit the “contract layer” like a product

Your T&Cs, returns policy, warranty language, and cancellation process are product surfaces.

Run a quarterly audit:

• Are remedies stated clearly (repair/replacement/refund)?
• Are there clauses that could be interpreted as limiting CPA rights?
• Do support scripts match the written policies?

If your documents and your frontline behaviour disagree, customers experience it as dishonesty.

Step 2: Measure complaint health like you measure conversions

Add operational metrics that predict regulatory pain:

• Time to first meaningful action (not just first reply)
• Time to resolution by category
• Repeat contact rate within 7 days
• Refund cycle time (approval to paid)
• Escalation rate to managers or external bodies

Then use AI to identify the top three drivers of repeat contacts. Fix those first.

Step 3: Put AI where the friction is (not where it’s fashionable)

AI should sit at the bottlenecks:

• Ticket triage and routing
• Agent knowledge and policy interpretation
• Refund/returns automation with human approval gates
• Quality assurance (spotting inconsistent or risky agent responses)

If your team is small, start with one use case: AI-assisted triage + templated remedy flows. It’s usually the fastest ROI.

Step 4: Treat customer service capacity as a growth constraint

The settlement’s requirement to add 300 customer service jobs over five years is a clue: regulators see service capacity as part of consumer protection.

AI helps teams do more with less, but it doesn’t eliminate the need for humans—especially in high-value disputes. The winning model is:

• AI handles routing, summaries, knowledge retrieval, and repetitive updates
• Humans handle judgment calls, empathy, and exceptions

In South Africa, where WhatsApp support and voice channels remain critical, that hybrid model is how you scale without losing trust.

What customers will expect next (and why that’s good news)

This settlement encourages a healthier pattern in the digital economy: companies will compete on clarity and remedy, not just price.

Expect customers to demand:

• Faster refunds and cleaner communication
• No “fine print surprises” on warranties and returns
• Proof of condition/quality for pre-owned and refurbished goods
• Consistent outcomes across channels (email, call centre, WhatsApp, in-store)

For businesses, the upside is real: when you design a complaint system that actually works, you reduce chargebacks, negative reviews, and support costs—and you earn repeat purchases.

Where this fits in our AI and e-commerce series

This series is about how AI is powering e-commerce and digital services in South Africa—not just through marketing automation, but through the parts of the customer journey that decide whether you keep the customer.

The WeBuyCars case is a timely reminder: AI isn’t only a growth tool. It’s a trust tool. When you use it to improve transparency, shorten time-to-resolution, and keep a defensible record of fair treatment, you build a business that’s harder to disrupt and easier to recommend.

If you’re operating a digital commerce platform going into 2026, here’s the practical next step: map your top five complaint categories, then identify where AI can reduce ambiguity and delays in each one. Which customer promise would you be comfortable defending—verbatim—under regulator scrutiny?