The sustainability transition of the chemical industry hinges on the educated design of catalysts for reactions such as the CO hydrogenation, optimizing the materials for selectivity toward valuable products. So far, theoretical models have been used to predict reaction selectivity from the competition of elementary surface processes. Here, we provide an in situ experimental view of surface adsorbates during CO hydrogenation. We compare X-ray photoelectron spectra acquired at reaction conditions (200–325 °C, 150 mbar) over single crystals of Fe, Rh, Ni, Co, and Cu and infer which elementary steps decide the product distribution. We find that the chemisorption energies of C and O, as often used descriptors for catalytic activity, qualitatively predict the rate-limiting steps. They fail, although when reaction-induced carburization occurs on Ni and Fe, steering the selectivity toward methanation on Ni and hydrocarbon chain growth on Fe. For the noncarburized Co and Rh we show how the adsorbate distribution and the oxygen chemisorption energy allow for oxygenate production on Rh, but hydrocarbon chain growth on stepped Co. Ultimately, we show how in situ experiments provide a chemical and mechanistic understanding of CO hydrogenation selectivity, useful to tailor catalysts for a sustainable production of high-value chemicals.