You may want to import rather your Zernike measurements into a Zernike fringe phase surface located at STOP position, then optimizing your decenter/tilt parameters, so to get low RMS wavefront error.That may work better

Your table of measured coefficients does not give the normalization radius for the polynomials. Check with your supplier what this value is if you don’t have it already. There are cases where a smaller inaccuracy (1%) can result is a completely different value even for low degrees coefficients so guessing/handwaving is not good enough.
See for example the article below where the authors found that

“In our example we obtain a relative sensitivity [...] of 11, indicating that the NA value should be specified to 10−3 accuracy when the required coefficient accuracy is 1%.”

https://www.spiedigitallibrary.org/journals/Journal-of-MicroNanolithography-MEMS-and-MOEMS/volume-5/issue-3/030501/Concise-formula-for-the-Zernike-coefficients-of-scaled-pupils/10.1117/1.2345672.pdf

Example using the Zemax doublet sample file, comparing wavefront Zernike standard coefficients for the EPD of 20mm (default) and 19.8 (1% difference). The coefficients are totally different (piston and defocus ~doubled, primary spherical went from negative to positive and the magnitude got 36 times higher.

The use of a Zernike fringe phase surface suggested above likely helps to reduce this sensitivity.