When Adding a Middleman Changes Platform Economics

Every platform company learns the same rule early: as supply grows, your leverage does too. More drivers on Uber means drivers compete for rides, so the platform can take a bigger cut and pay out less. It is a reliable dynamic of the gig economy and has become a template that companies around the world are trying to replicate.

But a new study from Duke University’s Fuqua School of Business finds that rule breaks down—and can even reverse—the moment a middleman enters the picture. 

In the forthcoming paper, Sharing Platforms in Emerging Markets: The Role of Human Intermediaries*, Fuqua Professor Can Zhang and his co-authors—Priyank Arora of the University of South Carolina and Hello Tractor’s former data scientist Olufunke Adebola—studied sharing-economy platforms that operate where a standard app isn’t enough: markets with poor internet access, fragmented demand, and customers who can’t book or pay digitally. In those settings, platforms rely on local agents who go door to door, aggregate customers, and submit requests on their behalf.

When booking agents come in, many assumptions about platform economics need to be revisited. More supply available? Pay the agents more and lower your own commission. Demand surging? Raise agent wages. Paying your agents a fixed percentage across services and regions? You may be suppressing demand.

“The size of the market depends on the booking agents who collect demand,” Zhang explained. “When you have more supply, you want to incentivize them more.”

When an app isn’t enough

The researchers partnered with Hello Tractor, a sharing-economy platform that connects tractor owners with small farmers in rural Africa.

Zhang’s interest in the topic is personal. He grew up watching his grandparents farm without mechanization.

“I saw how people are farming without equipment,” he said.

Without tractors, farmers relied on animals and manual labor—work that was slower, more dangerous, and physically punishing. It also affected crop yields. Mechanized ploughing can turn soil more uniformly, improving soil structure and nutrient distribution. “It’s not only about saving labor,” Zhang noted. “It’s also about the quality of the job.”

Yet owning a tractor makes little economic sense for smallholders cultivating just one hectare of land. Sharing the equipment is the natural solution.

“In certain areas, especially in rural parts of emerging economies, digital access remains a major constraint,” Zhang said. “Someone may not even have digital data plans or payment options on their cell phones.”

The problem is that demand is fragmented. “If I’m a tractor owner, I’m not going to drive 10 miles to serve just one hectare,” Zhang explained. The trip simply wouldn’t be profitable. At the same time, latent demand may be abundant—farmers needing the service—but that need doesn’t show up in the app.

“There is a need,” Zhang said. “But that need does not always translate into demand submitted on the platform.” Limited internet access and the lack of mobile payments mean the platform may see little activity even when demand exists.

Booking agents fill the gap. They go from farm to farm, aggregate small plots into viable bundles, and submit combined requests to the platform—creating both volume and economies of scale.

The economics change

Using their model to examine Hello Tractor's business practices, they found that once booking agents enter the picture, standard platform economics change.

A commonly used method in platform settings—paying agents a fixed percentage of the customer price–turned out to be insufficient to appropriately incentivize demand.

“These agents are independent workers. They are not employees, so you must use financial incentives to motivate them,” Zhang said.

Conventional platform theory suggests that when supply grows—when more drivers join Uber, for example—the platform can take a bigger cut by raising its commission.

But with booking agents, that logic can flip.

As the provider pool grows, Zhang and co-authors show that platforms should often raise agent wages while lowering their own commission.

Why? Because supply doesn’t create demand automatically in these settings. Booking agents determine how much demand is aggregated and realized. With more tractors available, agents should be incentivized to aggregate more demand.

Raising agent wages encourages them to gather more customers. Even if the platform keeps a smaller percentage of each transaction, higher overall volume can increase total profit. 

“I am collecting a lower portion of the price now, but a lot more demand will be served. I will still gain a higher profit overall,” Zhang said.

The broader implication: platforms that rely on booking agents may be leaving money on the table if they pay agents a fixed percentage of customer price. Markets with fragmented demand or limited digital access may require rethinking how platforms use prices and wages to balance supply and demand.

When surge pricing shouldn’t reward providers

The paper also challenged traditional surge pricing strategies.

In ride-hailing, high demand typically triggers higher prices and higher driver pay to attract additional supply. But in markets with booking agents and significant fixed costs—such as hauling a tractor to a remote community—the calculus changes.

When demand rises, providers can serve more customers in one trip, spreading their fixed costs over a larger base. That lowers their average cost.

In those cases, platforms may optimally pair higher customer prices with lower provider wages and higher booking-agent wages.

Booking agents can collect more orders per visit. Paying them more helps the platform capture that demand surge. Meanwhile, providers don’t necessarily need higher wages because their per-customer costs are falling.

The same logic applies to any platform that relies on independent agents to reach customers that technology can't reach— agricultural services, telemedicine, and last-mile product or service delivery in underserved markets. The study suggests that incentives should reflect the role of booking agents in shaping both demand and the cost structure of providers.

A win-win, with conditions

But even this wage strategy doesn’t always produce the desired outcome, Zhang said. 

When demand aggregation is expensive—when it is difficult for agents to collect additional customers—increasing agent wages may not generate enough value.

The opposite problem can arise when agents can easily collect more demand—and therefore they are relatively inexpensive. While the platform may be tempted to raise the service price and extract more profit, this profit-maximizing price could rise too high for customers’ benefit.

To preserve shared gains, the platform may need to commit to a lower price than its profit-maximizing one, Zhang said.

“The incentives don’t naturally line up unless the platform is willing to sacrifice a little bit of its own profit to achieve win-win,” Zhang said.

From tractors to telemedicine

Although the research is motivated by agricultural services, its implications extend far beyond farming.

The authors highlight telemedicine platforms in Pakistan and fresh-produce platforms in Kenya that also rely on local agents to connect fragmented customers to providers.

In each case, the challenge is similar: customers lack easy digital access, and individual transactions are too small to justify the service.

The broader lesson is that in such cases, the Uber model is not always applicable. In underserved markets, human intermediaries can be essential complements to technology.

“And once you have this additional layer, you need to manage incentives in a different way,” Zhang said.

---

* Paper accepted for publication at Manufacturing & Service Operations Management