No need for the remark about training, that's offensive as oppose to constructive. Not sure why you'd feel the need to add that comment at all.
I've had plenty of training and have been given a project to specifically develop a UV-VIS method *if* possible, obviously ELSD/CAD etc. being easier and more favorable.
First, a general comment: from a single post, it can be difficult to gauge the experience level of the poster in order to frame an appropriate reply (adequate detail without being patronizing). You didn't provide a lot of detail in the original post and, to be honest, it wasn't obvious to me that you were experienced and were trying the DAD on purpose instead of from naivete.
As to whether I have ever seen a "nice" peak only above a certain concentration, the answer is "yes". The first time was a data-processing artifact decades ago. The data system (actually, in those days, "digital integrator") would only show a signal above a certain threshold level. Apparently, the idea was not to show noise. Hopefully, not something you would see today, but you might check to see if what you're plotting is the "raw" signal or processed data.
The second is a more likely scenario if you are using a cation exchange column without any buffer in the mobile phase (your initial post *did* say "water", not "buffer"). Without any competing ions in the mobile phase, it's possible that any sample below a reasonable level is simply fully bound to the column. I've run into this with proteins on GFC columns that had residual active sites. At low levels, all of the protein was strongly bound. We had to "prime" the column with several high-level injections of BSA before we could use them.