Seven books for the person who ends up writing the words inside the product.
Interface text is usually written last, by whoever is free, and it is the part of a product people read most closely. These books treat words as a design material with a job to do, covering how to find the user need behind a page, how to write titles, buttons, errors and empty states, and how to get involved early enough to change what is being built. The first two give you a method, and the rest support it with structure, evidence and argument.
Winters established the content design practice at the UK Government Digital Service, and this is the method, running from user needs and search evidence through pair writing and testing content with real users. It includes practical material on working with the legal and policy teams who own the words you want to change.
Podmajersky gives you a repeatable process, deriving a voice from product principles and then working through titles, buttons, empty states, errors and permission requests. The same problems are solved for three fictional products of different sizes, so you see the method at both ends of the scale.
Metts and Welfle on getting involved early enough to matter, with a long section on collaborating with designers, engineers and researchers. Also strong on accessibility and on the ethics of persuasive wording.
Covert on naming things, which is where most content problems begin. Short chapters, an exercise at the end of each, and no jargon, so it suits a team that has inherited a vocabulary nobody agrees on.
Krug on how little of your text anybody reads, and on making a page self evident. His cheap usability test is also the fastest way to show a stakeholder that the wording is the problem.
Labels are content, and this is the standard reference on labelling, navigation and search systems, including thesauri and controlled vocabularies. Read it when your terms drift apart across a large product.
Minto on structuring an argument so the recommendation comes first. That is the skill for the document arguing for the content change, which is often harder to win than the writing itself.