It will entirely depend on your protein. How much do you know about the protein? How many proteoforms are present? Are you trying to capture one proteoform or all of them? Are some peptides able to distinguish the phenotype better than others? An important thing to consider is that proteins are very complicated and on average there are ~100 proteoforms per protein coding gene in humans (https://www.nature.com/articles/nchembio.2576).
If something is known about the protein then we will specifically target portions of the protein coding gene sequence in an effort to capture the forms of the protein we are interested in. If nothing is known then we will often focus on 1) peptides that can be measured precisely and 2) are different between the conditions we are interested in.
Thyroglobulin is definitely a great example of an assay performed with lots of validation using a single peptide (https://pubmed.ncbi.nlm.nih.gov/26430076/). However, there are lots of cases where peptides from the same protein coding gene give very different results and you need to consider which peptide(s) reflect the biology you are trying to measure (https://www.biorxiv.org/content/10.1101/2021.01.25.428175v1.full). It is important to remember that in most cases it isn't ok to report a single quantity per protein unless you know that summarizing the results to a single measurement is ok for your protein.