A method was proposed to jointly optimize crowdsourcing service price and selective crowdsourcing distribution scheme. A crowdsourcing supply function about price was firstly constructed based on their relationship and then a mix-integer non-linear programming model was constructed to optimize the crowdsourcing service price, customer assignment scheme and distribution routes. The mix-integer non-linear programming model was further transformed into a more easily solved mix-integer linear programming model using the big-M method. To solve large-scale cases, an adaptive large neighborhood search algorithm was designed, combined with the problem domain knowledge-based local searches, saving algorithm, tabus search algorithm and simulated annealing algorithm. The performance of the adaptive large neighborhood search algorithm is superior to that of GUROBI solver, the earliest ready time rule and the saving algorithm. Numerical analyses show that the selective crowdsourcing distribution mode is better than no crowdsourcing distribution mode and full crowdsourcing distribution mode in lowering the distribution cost, the crowdsourcing distribution service mode is suitable for the scenarios with the high crowdsourcing supply service price sensitivity level and tight time window, and the distribution cost can be reduced by reasonably adding transfer points and widening time windows.