I agree with BrianN's answer: skip the registration-first step. Here is an extended explanation of why this works best, which I feel complements his answer.
You defined one of your goals as "get them signed up". But as a normal, white-hat business, this is not really your goal. Your goal is to get them to become regular users, whatever the definition of "regular" is in your field. An account which is not in use does not give you an advantage. The exceptions: 1) if you are building a collection of valid e-mail addresses and related you plan to use for marketing purposes, which can be anywhere from unsavory to illegal, and 2) if you are trying to impress investors and press with "we have a million users" slides. But any investor worth his salt will ask you for the percentage of active users and paying users among them. So, I will assume you don't care for either of the exceptions and state that, in the typical freemium model, the root of your goal tree is:
- Getting as many paying customers as possible, which is achieved through
- Getting as many active users as possible.
How does a non-user convert to an active user? Let's separate the users in three groups: decided-against using your solution, decided-for it, and undecided. You can't do anything about the first two, you are trying to convert the undecided visitors to your site to decided-for. From the user's point of view, the process looks like this:
- Get preliminary information.
- Do whatever needed to try out the product/service. At this stage, he may terminate the process if "whatever needed" requires too much in terms of effort, money, or loss of a feeling of security (e.g. by giving up sensitive information).
- Actually try out the product/service.
- Decide for or against further use.
- If decided for further use, do whatever necessary to incorporate the solution in his/her workflow.
In a try-first solution, this process is supported in its details. Sign-up happens at Step 5.
In a sign-up-first solution, the site does not fit the process so well. But users will act as users always do: they will bend the software to fit their process, not the other way round. So, they will still follow the process above. Sign up now occurs at step 2. And the decision for or against your software occurs at step 4. All the newly decided-against users end up with an account for your service/product. This does not make them decided-for users, and they will not return to your site later.
Can you tweak something about the sign up to make a decision-for more likely? No, you can't. If you make the sign up very low effort in the terms described above, it doesn't create any commitment in the user. If you raise the effort required, you increase his commitment, but you also increase the proportion of users who will terminate on Step 2. And while I have no numbers for that, the industry trends of decreasing trying-out barriers (which probably prompted you to create the question) seem to confirm that the increase of terminating customers is disproportionally high.
And does the try-out-first method hurt your relationship with the users who were decided-for before they came to the site? No, it doesn't. They actively want to sign up; the only way a try can stop them from doing that is if they find out that they were mistaken and your solution is not good for them. Those people would have become inactive after signing up anyway.
Bottom line: You can't force users to do something they don't want to do. If they don't want to be your active users/customers, they won't be, no matter if they signed up or not. So keep the sign up for those who have decided to stay with you, and let the others try first.