If age is highly correlated with virtual fitting room, yes, this will not be a good model to estimate the treatment effect. If they are not highly correlated, we need to consider the tradeoff between the added explanatory power and multicollinearity.