Thermoelectric technology, which possesses potential application in recycling industrial waste heat as energy, calls for novel high-performance materials. The systematic exploration of novel thermoelectric materials with excellent electronic transport properties is severely hindered by limited insight into the underlying bonding orbitals of atomic structures. Here we propose a simple yet successful strategy to discover and design high-performance layered thermoelectric materials through minimizing the crystal field splitting energy of orbitals to realize high orbital degeneracy. The approach naturally leads to design maps for optimizing the thermoelectric power factor through forming solid solutions and biaxial strain. Using this approach, we predict a series of potential thermoelectric candidates from layered CaAl2Si2-type Zintl compounds. Several of them contain nontoxic, low-cost and earth-abundant elements. Moreover, the approach can be extended to several other non-cubic materials, thereby substantially accelerating the screening and design of new thermoelectric materials.