For a credit fund in the 3 to 7 year duration bucket, it would not. The risk sits on a very narrow part of the curve. Whether you have five or 15 KRD buckets, the DV01 would remain the same and the allocation wouldn't move. Extra tenors add more lines to the report but don't necessarily create a meaningful shift.
Custom tenors become more interesting in long duration strategies. 40 year gilts, ultra long swaps - those are the types of instruments that a pension fund or a life insurance company might want to evaluate with custom KRD buckets. A standard set up goes to 30 years and the 30 year bucket incorrectly becomes a catch-all. This is a modeling problem and it hides curve risk in the most convex region of it.
Relative value is the second case. A PM running a barbell (short the belly, long the wings) needs enough granularity to distinguish the 7-year from the 10-year bucket. A grid that jumps from 5 to 10 will show the barbell as flat. Add a 7-year point and the short position in the belly becomes visible.
Now there is another frequent oversight: modelling custom tenors without updating the hedging universe in the model. If the grid includes a 15 year point but the only available instruments are 10 and 20 year futures, the model will consequently show exposure to the fund manager that they cannot act on.
KRD's suffer from an innate interpolation problem. Key rate durations are computed by shifting the curve point and observing the price change. The shift gets interpolated between tenors: Linear, cubic, log-linear. The methodology, sometimes more sometimes less, will affect the outcome but enough to produce different key rate profiles for the same underlying portfolio.
Set key rate tenors at the points where you have instruments to hedge. If you hedge with 2, 5, 10, and 30-year futures, those are your key rate points. Adding more subdivides the risk into pieces your Portfolio managers cannot act on.