Actually, if I were inclined to "do the math", I would set the variables like this:
A = % of SPECIALTY kit buyers only willing to order the Dasher version.
B = % of SPECIALTY kit buyers only willing to order the Dancer version.
C = % of SPECIALTY kit buyers who intend to order both versions.
D = % of SPECIALTY kit buyers who will only buy one version, prefer the Dasher version, but would be willing to order the Dancer version if that were the only viable option.
E = % of SPECIALTY kit buyers with a preference for the Dancer version but would order the Dasher version if the Dancer version weren't available, or would switch their order to the Dasher version if the Dancer version didn't look like it was going to tip.
F = % of SPECIALTY kit buyers with a preference for the Dancer version, but would order the Dasher version if the Dancer version weren't available, and would not switch to the Dasher version if the Dancer version didn't look like it was going to tip.
We have to imagine a scenario where given a total of x orders of SPECIALTY kits, that Ax + Cx + Dx + Ex is less than about 22 and where adding Fx to the total brings it up to at least 22. I'm guessing that the "danger zone"--where we split the vote between E and F--is when x is somewhere around 25-27. Fewer than 25 total orders and even throwing in Fx won't help the Dasher version tip anyway, whereas 28 or more and Fx isn't needed to help the Dasher version reach 22 orders. That's a pretty narrow window of risk.