Urban freight has become a central concern in smart-city governance because rising delivery intensity increases congestion, emissions, and service unreliability in dense metropolitan areas. This study examines whether third-party logistics (3PL) providers can contribute to urban freight management not only through execution, but also through predictive coordination. Using operational data from an international 3PL operator serving Warsaw and Wroclaw, the paper evaluates a forecasting workflow that combines warehouse management system (WMS) and transport management system (TMS) data to predict delivery intensity at both city and postcode levels. The dataset covers four months of daily operations and contains approximately 200,000 records. The predictive engine dynamically selects among auto.arima(), nnetar(), and ets() on the basis of weighted forecast-error testing. Results show high forecast accuracy for palletized flows and weaker but still decision-useful performance for parcel flows. At city level, mean absolute percentage error (MAPE) is 0.36% for Warsaw pallets, 17.47% for Warsaw parcels, 3.78% for Wroclaw pallets, and 4.03% for Wroclaw parcels. Postcode-level error profiles further show that pallet forecasts remain comparatively stable, whereas parcel deliveries exhibit materially greater volatility. The delivery network is also highly irregular: most destination postcodes in both cities are served on no more than 25% of working days. On that basis, the paper proposes a practical smart-city coordination model in which 3PL forecasts are integrated with intelligent transport system (ITS) traffic data and postcode priorities to support delivery-slot prioritization. The study demonstrates that predictive 3PL capabilities can form a viable operational interface between logistics providers and urban traffic governance.