Same revision, identities checked. One bracket can settle in two positions with the clamp open.

Then repeatability is already a question within one piece. Record which contacts change between those positions under the bench inspection method; do not use extra clamp force to choose a winner.

Sam's observation also keeps sample variation from becoming the automatic explanation. You can examine different brackets afterwards without losing the two seats already found.

One corner lifts when the opposite guide is touched. Designer's sheet calls both features locators.

That names the conflict to review, not the feature you should grind away.

Ask which part datum each is supposed to establish and how the drawing tolerances are accommodated. Two labelled locators can still be a valid arrangement, but the designer needs to explain this observed alternate seating, not just repeat the labels.

Designer found the guide copied from an earlier bracket revision into this nest. Current parts were correct.

Has the nest changed yet?

No. Revised guide clearance proposed; drawing review first, then bench checks. Stronger clamp suggestion withdrawn.

Keep the observation that made the error visible in the revised tooling check.

And use the identified sample that found two seats when that check happens. A different, convenient bracket could conceal the problem while the drawing revision appears to close it.