Across the top face. The tape seam crosses that area on some accepted incoming packs. The demonstration cup landed on plain board, with no seam beneath it.

Then don't let them exclude the taped packs and call the concept representative. Ask what incoming condition they intend to support, with packing involved in that definition. A supplier can choose a restricted scope, but your workload still contains the cartons outside it.

Send identified filled samples covering the allowed surfaces before buying another tool. Agree what the trial records, including failed acquisition and doubtful holding, so unsuitable samples do not simply disappear from the result.

Unknown in the old demonstration. Our representative samples were filled and identified for the new supervised trial; one taped variant failed acquisition, while the plain-board examples completed the demonstrated transfer. The original concept is not accepted for the full incoming range.

Keep that failed variant in the next comparison. A changed cup position or tool would need its own assessment, not just a better-looking repeat on the plain cartons.

And record what the demonstrated failure response actually did. Failure to pick is useful evidence, but I would want to know whether the proposed workflow identifies and handles it without an operator having to infer it from an empty destination.