Guidelines for developing optical clocks with $10^{-18}$ fractional frequency uncertainty

There has been tremendous progress in the performance of optical frequency standards since the first proposals to carry out precision spectroscopy on trapped, single ions in the 1970s. The estimated fractional frequency uncertainty of today's leading optical standards is currently in the $10^{-18}$ range, approximately two orders of magnitude better than that of the best caesium primary frequency standards. This exceptional accuracy and stability is resulting in a growing number of research groups developing optical clocks. While good review papers covering the topic already exist, more practical guidelines are needed as a complement. The purpose of this document is therefore to provide technical guidance for researchers starting in the field of optical clocks. The target audience includes national metrology institutes (NMIs) wanting to set up optical clocks (or subsystems thereof) and PhD students and postdocs entering the field. Another potential audience is academic groups with experience in atomic physics and atom or ion trapping, but with less experience of time and frequency metrology and optical clock requirements. These guidelines have arisen from the scope of the EMPIR project "Optical clocks with $1 \times 10^{-18}$ uncertainty" (OC18). Therefore, the examples are from European laboratories even though similar work is carried out all over the world. The goal of OC18 was to push the development of optical clocks by improving each of the necessary subsystems: ultrastable lasers, neutral-atom and single-ion traps, and interrogation techniques. This document shares the knowledge acquired by the OC18 project consortium and gives practical guidance on each of these aspects.