Uber Built the App for San Francisco. Accra Rewired It.

Based on the research ofAppau, Eckhardt, and Baako, "Global Ride-Hailing Platform Affordances and Developing Market Characteristics," Journal of Marketing, 2026

The Affordance That Assumed a Different City

Every ride-hailing app encodes assumptions about the world it was built in: that most riders carry a bank card, that a GPS pin is a usable address, that a driver's take-rate is a fixed number a company can simply set. Uber launched in Accra on June 11, 2016, its service available only in the capital at first, with a base fare of 1.70 Ghanaian cedis plus 1.30 cedis per kilometer, backed by a free-ride launch promotion and a formal handshake with Ghana's Ministry of Transport. Nothing about that launch looked improvised. It looked like the same product Uber had shipped in dozens of cities before it, dropped onto a new map.

That is precisely the setup Samuelson Appau, Giana Eckhardt, and Kingsley Baako examine in a 2026 *Journal of Marketing* study built on ethnographic fieldwork in Ghana, interviewing drivers, riders, investors, Uber executives, and government officials. Their finding cuts against the standard "market entry" story: platform performance in a developing market is not simply a matter of how well an app converts users. It is shaped by three collisions between the platform's built-in affordances and the market's actual conditions, what the authors call platform infrastructural deficiencies, (in)formalization, and affordance-based regulation. Ghana's roads, cash economy, and informal taxi sector did not just create friction for Uber. They forced local actors to respond with what the authors catalogue as acceptance, avoidance, adaptation, substitution, resistance, brand responsibilization, and subversion. The app was not simply used. It was fought over, worked around, and partially rewritten by the people using it.

Cash Was Never a Feature Request. It Was a Correction.

Uber's affordances assumed a card-linked wallet, because that is how the product worked in the cities where it was designed. Sub-Saharan Africa answered back almost immediately. Nairobi ran a two-month cash pilot as only the second city in Uber's global network to test cash as a payment method, and Uber's then sub-Saharan Africa general manager Alon Lits later told Quartz that the business grew more than three times over during that trial. Lagos followed with its own cash experiment in January 2016, the third African city to try it, and cash payments eventually became standard across Uber's markets worldwide. Ghana, Kenya, Nigeria, Uganda, Ivory Coast, and Tanzania all eventually carried the cash option built from that African pressure.

Read that sequence carefully and the causality runs backward from how a product roadmap is usually told. Uber did not anticipate the cash economy and build for it. Riders and drivers in markets where a card was the exception, not the rule, made the card-only affordance functionally unusable, and kept transacting anyway, in cash, off whatever rails the app allowed, until the friction was loud enough that Uber's own product had to catch up to behavior that was already happening. The "innovation" was really a company absorbing a workaround its users had already normalized.

The company did not design cash in. It designed cash out, and its users designed it back in.

The Address That Doesn't Exist

A ride-hailing app's core promise is that a pin on a map is enough to find someone. That promise assumes a city has been mapped in a way a pin can describe. Most of Ghana had not been. The country's government, aiming to formalize an economy in which studies put employment reliance on the informal sector as high as 88 percent, launched the GhanaPostGPS digital addressing system on October 18, 2017, a little over a year after Uber's Accra debut, precisely because the country lacked a formal street-and-number addressing system and relied instead on landmarks: a church, a colored gate, a particular tree. Riders in Ghana were routinely describing their location the same way they always had, by what was nearby, not by what an app pin could encode, and drivers navigated by the same local knowledge that predates the platform entirely.

That gap between "the pin" and "the actual place" does not stay theoretical once money and time are attached to getting it wrong. In neighboring Nigeria, Uber found drivers using a spoofing tool called Lockito to fake their GPS location and inflate fares, a violation the company said it treats as fraud and polices with automated detection and account deactivation. Whatever the intent behind any single instance, the underlying mechanic is the same one the Appau, Eckhardt, and Baako framework predicts: when a platform's location affordance does not match the ground truth of how a city is actually organized, someone on the ground finds a way to make the mismatch work in their favor. Riders and drivers also fall back on the oldest fix available: agreeing a landmark meeting point or a destination by phone call or WhatsApp message before, or instead of, trusting the pin. None of that is a bug report. It is the market quietly patching the software with its own local knowledge, in real time, every day.

The Commission Nobody Actually Negotiated

Affordance-based regulation is the paper's least intuitive finding, and its sharpest one: when a formal regulator is largely absent, the platform's own settings, like the take-rate it charges drivers, function as a kind of regulation by default, imposed unilaterally rather than negotiated. Ghana's Uber drivers understood that immediately. On April 25, 2018, roughly two years after Uber's Accra launch, drivers massed outside the company's offices for a sit-down strike over the 25 percent commission the platform retained from every fare, telling local press it was rendering their businesses unprofitable. It was not a one-off flare-up: Ghana saw further ride-hailing driver actions in the years that followed, and the same 25 percent figure, and the same grievance, echoed in strikes across Kenya and Nigeria, where drivers demanded, and in Kenya's case eventually won, a lower commission after sustained protest.

That is resistance in the paper's own vocabulary, market actors pushing directly back on a number the platform set unilaterally. But the more interesting responses sit one level down, in adaptation and substitution: drivers accepting off-app arrangements with regular riders to avoid the cut entirely, or steering trips toward cash to keep bookkeeping opaque to the platform, or simply picking up a phone call from a known customer rather than routing the trip through the app at all. None of this shows up cleanly in Uber's own trip data, because by construction, it happens outside it.

Design for the Workaround, Because It Is Coming Either Way

The instinct inside a platform company is to treat every one of these behaviors as leakage to be closed: crack down on off-app arrangements, tighten fraud detection on GPS spoofing, hold the commission line and wait out the strike. Ghana remains, as of Uber's own tenth-anniversary account this year, one of just four African markets where the company still runs ride-hailing at all, after retreating in 2025 and 2026 from Côte d'Ivoire, Tanzania, Nigeria, and Uganda. The four survivors, Egypt, Ghana, Kenya, and South Africa, are not the markets where friction disappeared. They are the markets where a decade of exactly this kind of local renegotiation between platform and users has been absorbed rather than eliminated, cash options and all.

Uber's own history argues for humility here. The company spent a decade learning, market by market, that cash, informal drive-to-own financing, and locally adapted vehicle types were not concessions but the actual product-market fit it needed in Africa, and it exported cash payments globally as a result. The lesson generalizes past ride-hailing to any platform exporting a developed-market affordance into a market with a cash-heavy economy, an informal complementor base, or a regulatory vacuum: your "neutral" default is a design choice tuned to a specific set of infrastructural assumptions, and the moment those assumptions fail, users will not simply churn. They will renegotiate your product for you, in channels you cannot see, using whatever tools, cash, landmarks, WhatsApp, are already lying around. Build for that renegotiation deliberately, and you keep the leverage. Ignore it, and you find out about it the way Uber found out in Accra in 2018: from a parking lot full of drivers who stopped driving.