Ask your system architect if it's useful to have a PLL vendor with a huge range of flexible, off-the-shelf PLLs. Does it matter if the vendor has a dedicated design team available both to help you design the simplest possible clocking scheme, and to design custom PLLs if your problem resists a simple clocking scheme?

Consider for yourself the nature of the licensing terms, whether it obligates you to lingering royalties or special multiple-use pricing. Does the vendor need a test chip? Yours, perhaps?

Ask your vendor qualification people if you can switch foundries without requalifying your PLL vendor. Such changes are only possible if the PLL vendor's complete range of products are available to you on a broad range of processes from several foundries.

Ask the person responsible for timing closure about the value of a very low-jitter clock source. Ask that person how much it matters if the PLL vendor has fully characterized their design. Ask that person if they even believe the performance specifications claimed by your current PLL vendor.

Ask your bring-up team what it's like to work with a prototype with an unreliable clock source that has to be fixed in a second spin of the chip. Does it help if the PLL works well even outside the specified frequency range, even in bad packaging, or in the presence of unexpected noise?