Accurate sales forecasts are critical for managing logistics and production networks, especially in multichannel and omnichannel contexts. This study investigates (1) whether demand patterns differ across distribution channels, (2) how marketing campaigns affect these channels, (3) whether incorporating marketing information improves forecasting accuracy, and (4) whether different distribution channels require different forecasting approaches. We analyze fifteen weekly sales time series from direct sales, e-commerce, and retail using ARIMA, ARIMAX, and Random Forest models enriched with time-series and marketing features. Forecast accuracy is assessed using root mean squared error (RMSE), while feature relevance is examined through permutation importance, SHAP values, and ARIMAX coefficients. The results indicate heterogeneous effects of marketing campaigns across products and channels. Including marketing information improves accuracy in several cases and forecasting methods should be chosen individually per product but can be the same across distribution channels.