Branching enzymes (BEs) confer to α-glucans, the primary energy-storage reservoir in nature, a variety of features, like slow digestion. The full catalytic cycle of BEs can be divided in six steps, namely two covalent catalytic steps involving glycosylation and transglycosylation, and four noncatalytic steps involving substrate binding and transfers (SBTs). Despite the ever-growing wealth of biochemical and structural information on BEs, clear mechanistic insights into SBTs from an industrial-performance perspective are still missing. Here, we report a Rhodothermus profundi BE (RpBE) endowed with twice as much enzymatic activity as the Rhodothermus obamensis BE currently used in industry. Furthermore, we focus on the SBTs for RpBE by means of large-scale computations supported by experiment. Engineering of the crucial positions responsible for the initial substrate-binding step improves enzymatic activity significantly, while offering a possibility to customize product types. In addition, we show that the high-efficiency substrate-transfer steps preceding glycosylation and transglycosylation are the main reason for the remarkable enzymatic activity of RpBE, suggestive of engineering directions for the BE family.