Dave has it absolutely spot on - control your process and do not rely on output testing.
Ensure those gauges, indicators, whatever you wish to call them, that you use to set the process are actually giving what you are asking for - is a setting temperature of 140 degrees the actual temperature achieved? Are you getting 180 or 110 instead?
Also, this could change over time so setting you machine to 140 when you started the job might have given you (pulls a figure out of the air) 145 degrees. Six months later the same setting of 140 could be giving you 120 degrees. How would you know that if the gauges weren't calibrated?
Do some simple tests to prove the correlation between the temperature, pressure, and adhesion strength characteristics. Use design of experiments or simple scatter charts to prove correlation of the effect of changes in those parameters on adhesion strength. If the results prove there is no correlation between them (unlikely) then this will justify you not needing to control them or calibrate your equipment.
If you can prove the correlation, then control the process parameters with SPC, you can eliminate costly destructive testing.
We never get scrap from the adhesion process, I mean, if the final product looks "loose" because the glue was not melted correctly, then we just "iron" that part for 20 more secs, and the part is done, but that doesn't happen that frecuently.
Have you measured the extra cost of that rework? The extra 20 or so seconds on cycle time could have a significant impact on the product margin.
How effective is the visual check on detecting poor adhesion? Have you performed an
FMEA on this process? Are the RPN's high?
What would happen if GM found seat covers coming loose in the field - can your company stand the cost of a recall at the sort of numbers of vehicles that could be affected by a simple failure mode that could be avoided?