How South African e-commerce brands can take advantage of AI search
AI treats SA e-commerce as a sub-category of global e-commerce. Amazon comparisons crowd out local context.
SA-AEO-Bench v1.2 — pre-registered on OSF before data collection. 19,020 successful AI responses across GPT-5, Claude Sonnet 4.5 and Gemini 2.5 Pro. 232,254 cited URLs classified against a published source taxonomy. 100 South African brands, 12 industries. Re-run quarterly. Methodology and analysis code public at osf.io/w4az2.
This page is one industry’s view of that dataset. The per-brand cut is in the Scorecard, the full per-industry tables in the Q2 Index Report.
The pattern.
When the question is about buying something in SA, AI looks at SA sites — sometimes. BusinessTech is the site every engine reaches for on "where to buy" and "best online shop in SA" queries. MyBroadband carries the tech-product weight. TechCentral and htxt sit alongside as supporting signal. Hellopeter sits underneath all of it, shaping what the engines say about delivery and returns. That is the SA pool. A brand cleanly placed across the first three plus a clean Hellopeter footprint shows up in the answer. A brand quiet on those four does not.
The harder pattern is the Amazon-versus-Takealot default. ChatGPT defaults to Amazon on global queries and still names Amazon ahead of Takealot on a meaningful share of SA-specific ones — the corpus pre-dates Amazon SA launching here in 2024 and the model has not fully caught up. Gemini behaves better on local intent because it reads recent SA tech-press coverage of the Amazon SA rollout. Claude hedges, often naming both. Perplexity privileges whatever launch story ran last. Same query, four different recommendation sets.
Logistics is the second-order signal nobody is watching. Aramex, Courier Guy and PostNet integrations get cited in SA tech-press round-ups as the trust marker that separates a real SA e-commerce brand from a drop-shipper. Brands with named logistics partners read as legitimate to the engines. Brands without read as risky. The gap between best-positioned engine and worst-positioned engine on "is this site safe to buy from in SA" is wider than the marketing team thinks. The per-engine breakdown is in the SA AI Visibility Index Report. The picture for your brand is in the free Scorecard.
Where AI gets its info.
Three to five SA sites carry most of the weight in e-commerce. The names matter. The counts are reserved for the Index Report.
When you ask any of the four engines about online shopping in SA, BusinessTech turns up. It is the default category authority — "best online shops in SA", Black Friday round-ups, Takealot-versus-Amazon coverage. A brand left out of BusinessTech round-ups starts every answer behind the brands that made the cut.
PS5, smart TV, laptop, router, phone — when the question is "where do I buy a [tech thing] in SA", MyBroadband is the engines' default. A brand covered well here picks up a disproportionate share of high-ticket electronics intent. A brand missing from MyBroadband forfeits the consideration on the products that pay for everything else.
TechCentral is what AI reaches for on the corporate-news shape of the question — Amazon SA launching, Takealot leadership changes, Naspers and Prosus movements. The engines treat it as the senior commentary source. A brand named in TechCentral commentary lands harder than the same brand named in a marketing-press release.
The Q2 Index Report names every SA domain we measured for e-commerce — the full source taxonomy classified by role (default authority, freshness layer, complaint signal, long-tail specialist) with per-engine citation share for each one. The long-tail is where the cheap wins live. Download the Q2 Index →
For the source map specific to your brand — which domains currently shape AI’s answer about you — run the free Scorecard.
How the engines differ.
Each AI reaches for e-commerce sources differently. One strategy almost never works across all four. The per-engine picture for your brand is in the Scorecard.
ChatGPT reaches for Amazon on most "where to buy" queries, including SA-specific ones it should know better on. Takealot gets named — often second. Loot, Yuppiechef, Bash and Superbalist show up only when the prompt is explicitly category-specific. Worst engine for new SA entrants. Best engine for a brand already cemented as a SA category default.
Claude reads the editorial coverage of the Amazon SA launch more carefully than the other engines and is the most likely to name both Takealot and Amazon SA in the same answer with a caveat about delivery coverage. Best engine for brand-positioning questions. Weakest for transactional "where do I buy this right now" intent.
Gemini pulls Hellopeter, MyBroadband forums and Reddit threads the other engines ignore. It is the most dangerous engine for a brand with delivery-complaint volume — and the engine that powers Google AI Overviews, which is what most SA shoppers actually see when they search. Most likely to elevate Bash, Superbalist and the specialist brands on category queries.
Black Friday SA pieces, launch coverage, new-product reviews — Perplexity catches them first. A brand that fed htxt and MyBroadband in the last fortnight reads as the current authority. A brand that went quiet decays out of this engine first. The Black Friday SA cycle in particular reshuffles Perplexity answers every November.
Two ways e-commerce brands play this.
Brands in this category split into two camps. Which camp you are in tells you what AI-visibility risks you actually carry.
The category default
Takealot. Cemented across a decade of SA tech-press coverage as the answer to "where do I buy this in SA". AI engines name it by reflex on the queries that match. The risk is not Amazon SA — the risk is the engines still defaulting to the US Amazon on queries that should resolve locally. The defensive move is constant editorial freshness: Takealot stays in the answer because the SA tech-press keeps writing about Takealot. The day the coverage slows, Perplexity drops it first and Gemini follows.
The specialist play
Yuppiechef in kitchen, Bash and Superbalist in fashion, Faithful to Nature in natural goods, Loot in books and media. These brands won AI visibility by anchoring a category cleanly in SA tech-press round-ups and lifestyle coverage. The play works — they show up on "best [category] in SA" queries far more confidently than their general-traffic share would predict. The risk is breadth: a specialist that tries to broaden its AI footprint into general "best online shop" intent almost always loses to Takealot, and quietly weakens its category anchor in the process.
What changed this quarter.
Questions to take to your next meeting.
Each one is specific. Each one is answerable. We answer them for your brand in the Scorecard and for the category in the Q2 Index Report.
- 01When ChatGPT answers "where do I buy a PS5 in South Africa", does your brand get named — or does the answer default to Amazon?
- 02Where does your brand sit between ChatGPT (Amazon-default, conservative) and Gemini (Hellopeter-aware, specialist-friendly)? The gap is almost never small.
- 03Which Hellopeter delivery thread is currently shaping Gemini's answer about you? Have you ever read it?
- 04For your closest competitor — Takealot, Amazon SA, or your category specialist — which engine treats them most generously, and what coverage explains the lift?
The Q2 Index Report includes the full diagnostic question set for e-commerce — every meeting-ready question with the answer cross-referenced to the per-engine measurement. Plus the SA brands currently ahead, behind, or improving fastest on each one. Download the Q2 Index →
For the answer specific to your brand — where you sit on each question, on each engine — run the free Scorecard.
What to do about it.
Three categories of move lift e-commerce AI visibility. Which specific tactic is right for your brand depends on where you stand today — that conversation lives in the 30-minute walkthrough.
Audit your placement across the five SA e-commerce sites.
BusinessTech, MyBroadband, TechCentral, htxt and Hellopeter form the citation core. A brand cleanly placed on the first four and quiet on the fifth reads very differently to AI than one with delivery-complaint volume. The Scorecard names where your brand stands on each, and which placements have decayed since last quarter.
Watch the Amazon SA default before you spend a rand on awareness.
AI still names the global Amazon on a meaningful share of SA-specific queries. If your brand is named third behind two Amazon entries, awareness spend will not fix it — only fresh SA tech-press coverage will. The Index Report quantifies the gap across the category. The Scorecard quantifies it for your brand.
Re-check what AI thinks your brand actually is.
Entity confusion in e-commerce — your brand confused with a foreign namesake, with a sibling app, with a discontinued service — breaks more answers than the sector tracks. Wrong entity, wrong answer. The brand-specific confusions for your nameplate are in the Scorecard, with the impact of each one.
Brand-level scorecards for e-commerce land in the next Index Report. Reserve your copy →
Common questions,
answered straight.
Why does ChatGPT keep defaulting to Amazon on queries that are obviously about SA?+
ChatGPT learned from a corpus where Amazon is the default answer to almost every "where do I buy this" question worldwide. The Amazon SA launch in 2024 is still working its way through the model's training. Gemini, which reads recent SA tech-press more heavily, behaves better on local intent. The structural gap between ChatGPT and Gemini on SA e-commerce queries is the widest we measure in any sector. The full per-engine breakdown is in the Q2 SA AI Visibility Index Report.
My brand is on BusinessTech but not on MyBroadband. Does that actually matter?+
Yes — and it matters differently per category. BusinessTech is the "best online shop in SA" authority. MyBroadband is the "where do I buy this tech product" authority. Missing from one means missing from a specific class of question rather than from search overall. The Scorecard names which queries you are currently winning and which you are losing because of the gap.
Does Black Friday SA actually move AI visibility, or is it just noise?+
It moves Perplexity heavily and Gemini noticeably; ChatGPT and Claude shift more slowly. Brands that fed htxt, MyBroadband and BusinessTech with deal-coverage in late October and November reshape answers across the freshness-sensitive engines through December. Brands that sat the cycle out lose ground they do not recover until the next event. The cycle-by-cycle analysis is in the Index Report.
How fast does AI-visibility data go stale in SA e-commerce?+
Faster than most consumer sectors. Perplexity drops you within a couple of months of a quiet press cycle, especially on deal-led queries. Gemini holds a bit longer. ChatGPT and Claude decay over many months. Around Black Friday all four refresh aggressively. The freshness state for your specific brand is in the Scorecard.
Where do you get this data?+
We run roughly four to six thousand structured prompts per quarter across ChatGPT, Claude, Gemini and Perplexity. Five replications per prompt. Latin-square order-reversal to control for position bias. Methodology pre-registered on OSF. See the full methodology for details.
Keep going.
How SA e-commerce brands win AI search
The categories of move that lift SA online retail in AI answers.
Best online shopping in SA, AI-ranked
How the engines actually rank SA online retailers — and where they disagree.
Methodology — SA AEO Bench
Pre-registered measurement design and source taxonomy.
E-commerce is the category.
You are the question.
The free Scorecard runs the same e-commerce measurement against your specific brand. You get the per-engine gap, the SA sites currently shaping your AI answer, and the three moves that matter most for you.
- Dataset: sa-aeo-bench-v1
- Snapshot: 2026-05-19 · 17 days old · budget 270 days
- Pre-registration: osf.io/w4az2
- 14,826 responses · 188,877 citations · 100 brands · 10 industries
- POPIA-compliant · Information Officer registered
- Data residency: SA + EU only
- Methodology: pre-registered before data collection · audit trail per record
- Models measured: OpenAI GPT-5 · Anthropic Claude Sonnet 4.5 · Google Gemini 2.5 Pro