Monitoring of reef restoration efforts and artificial reefs (ARs) has typically been limited to coral fragment survival, hampering evaluation of broader objectives such as ecosystem recovery. This study aimed to determine to what extent AR design influences the ecological recovery of restored reefs by monitoring outplanted coral fragments, benthic cover, coral recruitment and fish and invertebrate communities for two years. Four AR designs (16 m2), unrestored controls and natural reef patches as reference (n = 10) were established in Mkwiro, Kenya. ARs consisted either of concrete disks with bottles, layered concrete disks, metal cages or a combination thereof. A mixture of 18 branching coral species (mainly Acropora spp.) was outplanted on ARs at a density of 7 corals m-2. After two years, 60% of all outplanted fragments had survived, already resulting in coral cover on most ARs comparable (though Acropora-dominated) to reference patches. Coral survival differed between ARs, with highest survival on cages due to the absence of crown-of-thorns sea star predation on this design. In total, 32 coral genera recruited on ARs and recruit densities were highest on reference patches, moderate on concrete ARs and low on cages. ARs and reference patches featured nearly twice the fish species richness and around an order of magnitude higher fish abundance and biomass compared to control patches. Fish abundance and biomass strongly correlated with coral cover on ARs. AR, reference and control patches all had distinct fish species compositions, but AR and reference patches were similar in terms of trophic structure of their fish communities. Motile invertebrates including gastropods, sea urchins, sea cucumbers and sea stars were present at ARs, but generally more abundant and diverse at natural reference patches. Taken together, all studied ecological parameters progressed towards reef ecosystem recovery, with varying influences of AR design and material. We recommend a combination of metal cages and layered concrete ARs to promote high fragment survival as well as natural coral recruitment. Ultimately, a longer period of monitoring is needed to fully determine the effectiveness reef restoration as conservation tool to support coral reef ecosystem recovery.