Demand-Side Flywheel

Win demand first, pull supply after

Convoy's driver conversion rate went from near zero to roughly 95 percent, and the difference was whether the caller was offering an app or a load.¹ Dan Lewis recounts that the trucking marketplace launched on the supply-first hypothesis, that drivers would download an app for the good experience and loads would come, and that it "completely bombed," because truck drivers operating in a world of phantom load boards and untrustworthy brokers had no reason to download an app for theoretical future freight.¹ The pivot, in his account, was to win demand first: sign a shipper, get a real load assignment, then call a driver and offer a specific load in the next 24 hours. Seeing real work, the driver downloaded the app and completed onboarding to take that job.¹

Stated generally, the move is to bootstrap a business-to-business marketplace by first winning demand, a real customer with a real job, and then using that concrete demand to attract supply on demand, rather than the conventional approach of building supply first and expecting demand to follow. The account attributes the inversion to conditions specific to B2B services: suppliers will not join a platform with no volume, buyers will only commit to a platform with proven capacity, and the chicken-and-egg problem cannot be solved by either side joining on spec. Consumer marketplaces can build supply first because users discover the platform without a prior commitment, transactions clear in minutes, and trust comes from social proof and design; B2B services with long lead times and high trust requirements, in this framing, break each of those assumptions.

Why the inversion works here

Lewis ties the result to two properties he treats as characteristic of freight and, more broadly, of B2B services with advance notice. The first is long lead time: freight moves 24 to 72 hours after booking rather than in five minutes, and that window is what lets a platform onboard supply after demand is already locked. The second is a high trust requirement: supply will not join on spec, but will join instantly for proven, real work, so a concrete job functions as the trust signal that a pitch cannot supply.¹ This is the trust logic that Trust-First Design treats as a design default, met here through a real load rather than through onboarding.

Once the initial match is made, the account describes a compounding effect. More demand onboards more supply, which teaches the platform carrier locations and preferences; better location data improves matching, which lets the platform offer supply faster and win larger shippers; and accumulating lane density makes batching possible, lowering cost. In Lewis's framing the demand-first insight is the ignition, but lane density is the flywheel that sustains it. He also describes paying carriers within 24 hours of delivery, against an industry standard of 30 to 60 days, which both incentivized carriers to use the app throughout a job and produced the location, timing, and paperwork data needed to improve matching.¹

Scope and open questions

The account frames the pattern as applying where there is a long enough lead time to lock demand before supply is needed, a high enough trust requirement that supply will not join without real commitment, and a B2B context where supply is businesses optimizing for utilization. It sits alongside Celebrate the Activity as a demand-expansion pattern, but operates in the opposite condition: it applies when supply will not join on spec, whereas celebrating the activity applies when the category does not yet know it exists. Two questions are left open. It is unclear whether the pattern requires a specific lead-time range, since too short a window makes it impossible to secure demand before calling supply while too long a window may keep supply from committing that far out. It is also unclear how the trust requirement scales, since the pattern draws its force from supply having good reason to distrust on-spec platforms, and in markets with lower initial distrust the supply-first approach may still be easier.