Dated training records, authored procedures with revision histories, project and job records, internal certifications, and an organisational statement of how many people hold the same knowledge.
Records created during the work, not after it
Hypothetical example: The most persuasive evidence is contemporaneous, which means it has to be generated while the projects are running. A firm that expects to transfer specialists should build the habit deliberately: record who was trained on what and when, keep procedures under version control with named authors, retain job records showing which individual performed which proprietary step, and maintain a simple internal register of who is qualified on each capability. That register also answers the scarcity question directly.
Reconstructing all of this in the weeks before a filing is possible but visibly retrospective, and it tends to produce documents that assert conclusions rather than record events. A Cavan Monaghan business with several possible future transfers should treat this as ordinary quality record-keeping, since much of it is required for other purposes anyway.