It is really hard to get the kind of exponential efficiency gains in this space that a 'blitzscale'-focused startup needs in order to really succeed.
Trucking is an industry that relies heavily on relationships (trucker <-> broker, broker <-> shipper, broker <-> facility, etc), and you're not going to change that (required for true automation) without already being at an absurdly large scale.
Convoy spent too much effort on chasing things that had the potential to give the company those exponential gains without addressing the high-touch nature of the industry (but were just not a good fit, and ultimately fell flat).
And Convoy—conversely—had very little incentive to spend most of its effort on iterative improvements to the core brokerage (which would have probably led to profitability, but not the kind of returns that VCs ultimately want)