It's the incoming end face on the sketch, and the blank drawing allows length variation. We didn't give samples individual IDs on the sheet. That's frustrating.

@PriyaChan1090 Use labelled bench samples and measure overall length plus the intended seating reference. Comparing exposed lengths can then help identify seating differences to investigate.

@BethBennett0759 That still sounds a bit too generous to exposed length. A short blank sitting proud could match a longer blank seated correctly.

@MinaBennett0772 You're right to qualify it. I meant comparing the paired dimensions, not passing parts on exposed length alone. The reference must distinguish stock allowance from a gap.

Must the surface touching the stop also be the finished-part datum?

@RosaBaker0466 They don't have to be the same surface, but their relationship needs defining. Ask the drawing owner how the incoming reference supports the machining datum.

@BethBennett0759 The downstream datum is a shoulder rather than the incoming end. Their relationship is missing from our planning sketch, which makes the disagreement more understandable.

We had a different sleeve project with two perfectly respectable spreadsheets that couldn't be matched. Putting IDs on the physical samples made the measurements useful.

Did you link end-face pictures to the sample labels? A few edges look rough here, though photos shouldn't become a substitute for dimensions.

I'd attach them to the measurements. That lets another person locate the same edge or contact mark without assuming it's responsible for the variation.

@PriyaChan1090 Show where the stop really contacts that face. Naming the surface alone can hide a design that bears only against an edge.

Inspection pictures of the removed workholding show contact over only part of the face. They're not tied to sample IDs, so they can't explain individual readings.

Useful clue, but uneven witness marks don't tell you whether chips, geometry or an old contact condition produced them.

The measurement approach is clearer, but I still can't account for the seating variation. I'll keep the tooling decision open while the dimensional and reference questions are checked.

You've improved the investigation without claiming the cause is settled. Keeping the tooling decision open follows from that.