I have not used the assay myself, but Bhuiya and Liu "A cost-effective colorimetric assay for phenolic O-methyltransferases and characterization of caffeate 3-O-methyltransferases from Populus trichocarpa." Anal Biochem. 2009 384(1):151-8 [link] seems to have a relatively straightforward protocol.
Roughly, you make a 0.4% (w/v) solution of Gibbs reagent in ethanol, and then add one fifth volume of that to your enzymatic reaction mix. You then observe the solution spectrophotometrically.
The biggest issue I see is that it looks like each phenolic compound has its own absorption maximum, and widely varying extinction coefficients. In their protocol it seems they look at full range scans, rather than single wavelength measurements. You'll definitely need a standard curve for your particular phenolics in order to get quantitative results. (Also, any background phenolics or medium absorption in the 400-600 nm range may throw off measurements - use appropriate blanks.)