William - I'm not sure on the discrepancy in Goal Starts (i.e. experiment visits/entrances to the funnel), but everything else looks good. The important part is the conversion rate comparison for your two start pages; and that data is showing in your Content Experiment. I wouldn't worry too much about the entrances at this point; in fact, because both your A and B page are represented in the same funnel, the funnel can only show you aggregate conversion rate and aggregate funnel conversions (i.e. both A and B combined) and really only serves to confirm the conversion rate expressed in your content experiment.
I've tried running 2 funnels into 1 experiment (which would be ideal, because you could see how your A and B pages affect the entire funnel, individually), but ran into problems with my total conversions being overstated.