We show that text-based indicators of supply and demand disturbances in commodity markets provide distinct information about future inflation movements relative to existing predictors, inflation expectations and survey forecasts. Specifically, we document that demand-side disturbances play a significantly larger role in prediction because they typically lead to uniform increases in quantities and prices of goods across the consumer basket, resulting in a clear and positive relationship between commodity prices and overall inflation. Supply-side disturbances matter in particular circumstances, for instance during the recent period of the pandemic and geopolitical shocks. In terms of magnitudes, the commodity-specific indicators reduce out-of-sample inflation forecast errors by up to 30 percent. We finally apply our indexes to the inflation decomposition framework of Blanchard and Bernanke (2023) and corroborate their finding that the bulk of pandemic-era inflation can be attributed to commodity supply disruptions, resulting in price increases in goods markets.