It's either one of two elements of the bullets, or someting with the press of die.
The most common cause of inconsistent OAL's is because we A. Measure from tip of the bullet to the head, should be the same every time right? Wrong, first of all the tip of a bullet, even the ballistic poly carbon tips will vary by .001" on a good day, they get flatten or deformed slightly during shipping and dayto day handling. And then if it's a lead tipped bullet, the number's your going to get are going vary by quite a bit, lead obviously gets very easily flatten or deformed just while getting made and packaged, not to mention shipping, handling, and while in the hands of the retail store you now have clerks and customers handling them.
And another for sure element that effects OAL is, inconsistencies of olgive location. your seating does not seat off of the tip but off of the tappered area of the bullet. So with that onconsistency being nearly impossiible to avoid, it becomes extremely frustrating. A bullet comparator will eliminate much of the problem, but the unavoidable issue of deformed bullets is just something we are forced to accept.
This is one of the reasons I prefer to seat my bullets to where they chamber nicely, but most or all of the bullets are seated right up to contact with the lands. As a matter of fact I just finished seating 20 Speer 110 gr. TNT's and the OALs varied by as much as .015" on the extreme end, from the tip. And average was probably .007"-.008" variance from the tip. Fortunately for me though I wasn't realizing any of this variance at the olgive. The variance at the olgive was only about .003" and I seriously doubt that a variance .015" OAL from the tip is going to effect my accuracy enough, to degrade my recent worst 5 shot group with these TNT's of .620".