And, on top of the other great comments, I would also add that it will be drastically impacted by the weight of your brand and/or impact of your product. I have seen more than one case where large companies paid to be part of initial trials. The caveat is the offering companies were HUGE brands who could offer significant advantages down the road if the trial companies participated, even if they paid to.
For most, including start-ups, who have no significant brand or lever to apply, free (or close to it) is almost expected. I always suggest that if it is a software or hardware product merely getting into the client's test bed for evaluation is a huge win in itself. Delivering over and above in that phase will likely make the actual trial participation, and terms, that much easier.
What Chris said. the "Freemium" model has the advantage in that it increases your user base. OTOH the "conversion" rate is often very very low. For example we currently are using a FreeCRM because we just don't have enough staff to make a paid version worth it.
When we get to where our sales team is large enough that the Freemium version of the solution isn't enough, we are going to look very seriously at all full fee CRM solutions, and if we opt for another solution, we will simply delete all the accounts, pay for one month of use of the "Pro" upgrade and export all the data.
So the step from Free to Pay is a challenge.
We ourselves have an "intro trial" that we charge for - but for which in larger client engagements we will discount the trial fee.. But the point is that free is also sometimes a disincentive to use. I'm currently registered for a "free" trial for a piece of mgmt. software that I've not deployed because I haven't paid for it and because it doesn't cost me anything I haven't had the incentive to do the deployment