Two-sided　platforms　include　two　parts　of　users,　customers　and　suppliers,　which　are　the　key　strategic　resources　in　two-sided　platforms.　Through　Customer　Intelligence　Classification　(CIC)　management　provided　in　this　study,　a　platform　can　distinguish　different　values　of　customers　to　allocate　resources　efficiently　and　maximise　customer　resource　value　and　enterprise　investment　returns.　Meanwhile　platform　can　also　distinguish　collusive　drivers　through　CIC.　Network　Data　Envelopment　Analysis　(NDEA)　is　applied　in　this　research　to　open　the　black　box　of　a　two-sided　platform　operation　system　by　dividing　it　into　two　sub-processes:　Marketing　and　Service　(MS)　sub-process　and　Value-Creating　(VC)　sub-process.　Four　categories　of　customers　are　classified　through　performance　evaluated　by　NDEA　from　data　of　an　on-demand　freight　platform　Z　from　March　to　August　in　2018.　In　addition,　six　influencing　factors　are　discovered　through　Panel　Regression　Model.　The　main　conclusions　are:　(1)　Customer　with　high　marketing　service　valuation　and　low　value-creating　valuation,　needs　proper　operation　methods　to　promote　their　order　quality.　(2)　Customer　with　low　marketing　service　valuation　and　high　value-creating　valuation,　needs　to　take　attention　to　collusion　between　customers　and　driver.　(3)　Shipping　fee,　client　number,　driver　number　and　coupons　per　order　should　become　customers＇　Key　Performance　Indicators　to　evaluate　their　potential.