South Africa: R71bn Online, Takealot, and the Data-Light Creative Rule

South African e-commerce reached a record $4.07 billion — about R71 billion — in 2023, up 29% year on year, and the trajectory has continued: Black Friday 2025 online sales rose 43% over 2024, with 70% of those purchases made on a phone.

That is a market that has already crossed the threshold where creative volume decides who wins. And the supply of video is thin: studio production is priced for brand work, agency retainers are built around a handful of assets, and the sellers growing fastest on Takealot need twenty listing clips a month.

But South Africa is not a smaller version of the UK or Australia. It has constraints that change what the creative should be — and those constraints are the reason a per-asset, data-light pipeline is a genuine advantage here rather than just a cheaper option.

The constraint no one prices in: data

South African mobile users pay for data, and a meaningful share of the audience is on prepaid. That single fact reshapes video production more than any creative trend.

Three practical consequences:

• File size is a feature. A four-megabyte video that auto-plays into someone's data allowance gets abandoned or resented. Short cuts, efficient encoding and a strict payload budget are not technical niceties here — they are the difference between being watched and being closed.
• Shorter is better. Eight to fifteen seconds, not thirty. Every second of the wrong length is data spent on nothing.
• Captions carry the ad. With sound off and bandwidth constrained, burned-in text is how the message arrives. A caption-free ad is a silent, expensive mystery.

This is the opposite of the instinct most foreign creative teams bring, which is to add production value. In South Africa the discipline is subtraction: make it smaller, lighter and faster, then make it specific.

Load shedding deserves a specific mention. Scheduled power interruptions mean viewers drop out mid-session and sessions are fragmented. An ad whose payoff arrives at second fifteen is an ad many viewers will never reach — which is an argument for the shortest possible structure, not the longest.

The platforms

Takealot's scale is the anchor: reported survey use of 35.3% is more than double its nearest competitor, and its GMV grew 15% in FY2026. For a supplier, that means Takealot listing clips are the volume product in this market — exactly as Trendyol is in Türkiye and Allegro is in Poland.

Payment is the conversion problem

South African e-commerce has a documented checkout problem, and it is not the creative's fault — but the creative can mitigate it.

Card and debit cards remain the most used online methods, with instant EFT services such as Ozow carrying significant share (around 34% usage in recent surveys) and wallet apps like SnapScan in the mix. 3D Secure has been mandatory for card payments since 2014, which adds a step. And trade.gov notes that a large majority of potential customers abandon carts because of payment gateway failure — one of the highest abandonment causes in any market.

Three creative implications:

Name the payment method. "Pay with Ozow" or "Instant EFT" resolves the specific doubt a South African buyer has at checkout. Generic "secure payment" language does not.

Set delivery expectations honestly. Courier reality varies enormously between Gauteng, the Western Cape and rural areas, and pickup-point networks are a major part of how goods arrive. Over-promising delivery in this market produces refund requests, not loyalty.

Address trust directly. First-purchase risk is real, so a returns line and a contactable seller matter more than a brand flourish.

Language: English works, but not only English

South Africa has twelve official languages, and English is the commercial default. That makes it the cheapest market on our list to enter after Ireland — no localisation cost for the base asset.

But two things are worth knowing:

Afrikaans and isiZulu variants earn attention where they are used. Not as a national strategy, but as category-specific assets — Afrikaans for certain food, household and agricultural categories, isiZulu for specific regional retail. A variant is a text and voice layer on existing footage, so it is cheap, and the response in those segments is disproportionate because almost no brand bothers.

Accent and idiom matter within English. A neutral South African English voice reads as local; a British or American voice does not, in a market that is highly attuned to foreign messaging. If a voice actor is not available, keep the ad dialogue-free with captured text — a synthetic American voice is worse than silence.

One more detail: rand pricing with a space thousands separator and the "R" prefix — R1 299 — is the local convention, and prices should be shown VAT inclusive for consumers at 15%.

What sells

Takealot listing clips. Eight to fifteen seconds, silent-safe, product-first, no baked-in price. Bought in batches of ten to twenty and refreshed per promotion.

Feed ad variants. Ten hooks per product for Meta and TikTok. South African audiences are large enough to sustain volume, and creative fatigue is fast because the pool of active brands is smaller than in Europe.

Product page video for DTC stores. Higher margin, slower to sell, and the natural second order.

Seasonal batches. Black Friday has become genuinely large — the 43% online growth figure is not a rounding error — and the December/January window is compressed. Batch production is what makes it serviceable.

What a listing clip has to prove on Takealot

Unlike a feed ad, a Takealot listing clip has no scroll to stop. The buyer has already clicked, so the video's job is narrower and more practical: prove the thing is real, show its scale in a hand, and demonstrate the one mechanism the photographs cannot. Price, rating and delivery sit beside the player and are compared with competitors in the same view, so the video should not waste a second restating them. Eight seconds of honest demonstration beats twenty seconds of persuasion in this position, and it is the same discipline as every other marketplace in this roadmap.

Pricing in rand

VAT at 15% is additional for business buyers, and the invoice should show it separately. Rand pricing against a dollar-denominated cost base carries the same currency risk as the lira, which means repricing monthly and building the buffer in once rather than absorbing it quietly.

The route in

Cape Town and Johannesburg, agency-first. Both cities have a solid digital agency bench serving e-commerce clients whose creative demand exceeds production capacity, and both are English-speaking for business purposes, which makes the first conversation cheap.

One sample built for a real South African product — data-light, captioned, with Ozow and a named courier in it — will outperform a portfolio of beautiful foreign work, because it answers the two questions a local buyer is holding: does this person understand South Africa, and will my customer actually be able to watch it.

The broader commercial sequence is in from sample reel to retainer in 30 days, and the paid-social testing discipline is in AI UGC ads for Australian e-commerce — a comparable English-first market with similar mobile behaviour.

Encoding for a paid-data audience

The data constraint is easy to state and easy to get wrong in delivery. Four practical settings make the difference between a cut that plays and one that is closed.

• Keep the cut under fifteen seconds wherever the message allows. Length is the largest single driver of file size, and it is the one lever that costs nothing to pull.
• Deliver a compressed variant alongside the master. A full-quality master for paid social and a lighter variant for organic and WhatsApp distribution, where the audience is most likely to be on prepaid data.
• Front-load the first frame. No slow fade-in, no logo animation. A heavy opening frame is where a constrained connection stalls, and a stalled autoplay is a lost view.
• Test on a mid-range Android on mobile data, not on a desktop preview and not on an iPhone. This is the single most reliable QA step in this market and almost nobody does it.

The light cut is the one that wins word of mouth here, because it is the version a customer forwards to a friend. A brand whose ad cannot be forwarded has given up the cheapest distribution channel it has.

Where this fits

South Africa is the cheapest new market in this roadmap after Ireland: English-first, no new language capability, and a documented supply gap. It is also the entry point to a continent where mobile-first commerce is growing faster than anywhere else, which makes the first client a beachhead rather than a rounding error.

The mistake would be to treat it as a low-price market and compete on being cheap. The competitive position is data-light creative built for how South Africans actually use a phone — a specific capability that a foreign studio cannot produce and a local freelancer cannot scale.