Crowdsourced　on-demand　delivery　is　being　required　by　many　industries　with　the　increasing　demand　of　online　shopping　and　on-demand　food.　However,　it　is　difficult　to　ensure　the　delivery　efficiency　due　to　the　uncertainty　of　delivery　time　for　crowdsourced　couriers　in　total　delivery　process.　An　ant　colony　algorithm　based　on　total　logistics　delivery　efficiency　metrics　is　proposed　in　order　to　ensure　the　efficiency　of　crowdsourced　on-demand　delivery.　Firstly,　two　metrics　are　designed　to　describe　the　total　logistics　delivery　efficiency　quantitatively.　The　average　delivery　time　of　each　order　and　the　number　of　orders　completed　on　time　are　calculated　in　total　logistics　delivery　efficiency　metrics.　Secondly,　a　total　logistics　crowdsourced　on-demand　delivery　model　(TLCODM)　is　built　considering　the　transport　efficiency　and　order　punctuality.　This　model　serves　as　the　fundamental　support　for　crowdsourced　on-demand　delivery.　Finally,　an　ant　colony　algorithm　based　on　total　logistics　delivery　efficiency　metrics　(ACO-TLDE)　is　proposed　to　obtain　the　optimal　scheduling　solutions.　The　total　logistics　delivery　efficiency　metrics　are　used　to　guide　the　update　of　ant　colony　algorithm.　The　effectiveness　of　TLCODM　and　ACO-TLDE　are　verified　by　experiments.