Some of them do not realise that the logistics problem (of picking up stuff and delivering them) can't be solved by common routing.
Most of them work mostly on their mobile apps.
Some of them realise that the logistics problem they have is an NP-hard problem and try developing the solution in-house. Most of the time it results in failure (I believe some other firms that fell apart, I believe it was some Google acquired team cleaning services, particularly mentioned they couldn't solve the logistics problem in time and just when they did the funding disappeared).
Some realise that they can outsource their optimisation to services like:
+ open source: optaplanner, OpenVRP, jsprit, mixed integer programming solvers
+ proprietary services (ascii sorted): ArcGIS, ClearD Optima, DirectRoute, Foxtrot, Graphhopper, Optimoroute, Paragon, Roadnet, Route4Me, RouteXL, Routific, Routyn, Satalia, jOpt
All designed for different variants with some features that miss in others, available in all or some.
For some business the quality of routed paths isn't that important, for some it's crucial, especially if one wants to bring down the price.
Oh yeah, the problem is called Vehicle Routing Problem (or multiple travelling salesmen), it has variants that include vehicle capacities, pickup and delivery (one location preceding the other), pickup or delivery time windows, lunch breaks that don't have real location and other features.
That stuff is hard to solve if you have to run a business like Washio.