Five years ago, the typical African online seller had a phone, a roll of bubble wrap, and a WhatsApp number that doubled as the entire business. A customer DM'd. The seller scrolled to find the price. The customer asked again three days later. Half the orders never closed. The other half got paid for via screenshots that sometimes turned out to be fake.
It worked. It also nearly killed everyone running it.
The 2010s playbook: hustle, repeat
African chat-commerce in the 2010s ran on heroic effort. The seller was the storefront, the salesperson, the cashier, the dispatch rider, and the customer service team. The tooling was a Notes app, a spreadsheet, and a private WhatsApp group with the bike riders.
It was the only way. Shopify did not understand the market. Jumia did not want small sellers. There was no in-between, so the in-between got built by hand, one DM at a time.
What broke first
The thing that broke first was scale. Sellers hit a ceiling around 30 to 40 orders a day — not because the demand stopped, but because the human stopped. There were not enough hours to reply to every "is this still available?" while also packing, dispatching, and finding new stock.
The next thing to break was trust. Fake bank alerts. Disappearing buyers. "I sent it, check now." Sellers learned to delay shipping until they personally saw the credit alert — which slowed the whole machine down.
The third thing to break was memory. The customer who bought four months ago was a stranger today. The product they kept asking about was forgotten. There was no way to know who was a repeat buyer until they reminded you themselves.
The workarounds that did not really work
Some sellers tried Shopify. Their customers refused to leave WhatsApp. The store became a beautiful, empty room.
Some sellers tried link-in-bio shops. The buyer clicked, looked at three products, and DM'd anyway. "Hi, is this still in stock?"
Some sellers hired a junior to handle the inbox. The junior left. The seller went back to typing. The cycle repeated.
None of these solutions met the customer where they actually were. They asked the customer to change. The customer never does.
The shift: AI-native storefronts
What changed in the last twenty-four months is that AI got cheap enough to do the typing for you. Not as a chatbot bolted onto a website, but as the storefront itself.
An AI-native storefront sits inside WhatsApp, Instagram, and Facebook — the apps your customers were never going to leave. It reads the product photos you already post. It answers "how much?" the moment it lands. It sends a payment link. It confirms the credit. It logs the order. It remembers the customer. It follows up next month.
The seller stops typing. The customer never notices the difference — except that replies arrive faster, prices are always there, and the order never falls through the cracks.
The new playbook for African vendors
Sell where the customer already is. Do not drag them to a website. Bring the storefront to the chat.
Let the AI do the unpaid labor. The 100th "how much?" of the day should not need a human.
Make payments verifiable, not vibe-based. Payment links and instant confirmations end the screenshot drama.
Treat memory as an asset. The customer who bought once is your cheapest next sale. A storefront that remembers them is worth more than one that does not.
Reclaim the hours. The point of automating the inbox is not to be lazy. It is to spend the freed time on the things only you can do — designing, sourcing, building the brand.
Where this goes next
Within five years, the African vendors who are still personally typing every reply will look the way the vendors who refused to use card payments looked in 2015. The market will have moved on without them.
The vendors who win will be the ones who built early on AI-native storefronts — the ones who took the chaos out of the chat without abandoning the chat itself.
The DM was always going to be the storefront. We just needed the right counter behind it.
Open your free Stur store at stur.africa and run your business inside the conversation.
