To expand a bit on CPG's post:
A validated method should have a set of "system suitability" criteria. Those might include resolution between key peaks, plate counts, tailing factors, repeatability, retention windows, baseline noise, . . . etc. For a regulatory method, the rule is simple: if you meet system suitability, your results are valid. If you don't, the results are invalid and cannot be used.
In most methods, as the column ages, the system suitability results will gradually change (retention times may drift, peaks will get wider, tailing will get worse, etc.). If you chart the results over time, you can extrapolate forward to get an idea when you will drop below acceptability and make plans to change the column before you get to that point.