If shilling sites are more likely to be SEO-optimized, yes. One of the implications of Bayes' Theorem is that if observing A makes B more likely, then observing B also makes A more likely. A worked example:

Given:

P(shill) = P(!shill) = 0.5

P(optimized|shill) = 0.8

P(optimized|!shill) = 0.6

Then:

P(optimized) = P(shill) * P(optimized|shill) + P(!shill) * P(optimized|!shill) = 0.4 + 0.3 = 0.7

P(shill|optimized) = P(optimized|shill) * P(shill)/P(optimized) = 0.8 * 0.5/0.7 ≈ 0.57

P(shill|!optimized) = P(!optimized|shill) * P(shill)/P(!optimized) = 0.2 * 0.5/0.3 ≈ 0.33