Finally, if I'm a customer and you have a freemium offering, I'm very likely subsidizing the usage of upwards of 50 other users. Is that fair?
EDIT: I agree that a non-paying customer is just a user (duh) but the goal of a product is to grow your customer base (to meet the real goal of increasing your profit). So anyone using your product for free needs to be doing so as a means to an end: Either they will convert to a paying customer at some point, or they will refer paying customers.
If you're building a fun side project, free users are groovy. But if you're trying to grow a profitable company, and your free users aren't working for you in one way or another, they're just a liability.
Actually, conversion rate is completely objective - it is the percentage of users you have who 'convert' and become customers.
What that means for your company might be different from product to product and industry to industry, but a 3% conversion rate for freemium is awesome.
You mean 3% conversion being awesome "is relative to"/"varies by" a given product or niche.
EDIT: "subjective" is traditionally used when someone's opinions or biases are influencing a decision or conclusion, so I see where the confusion is here and the fault is mine for using that word. My main point is that there is no one-size-fits-all "good" conversion percentage, it's entirely dependent on the product and market.
The best advice I can give regarding when it's appropriate to use freemium and free trials:
-Use Freemium when: your per user usage costs are trivial (Examples: Evernote, Basecamp, Freshbooks...sites that don't incur transactional usage costs like having to pay for data access to other systems or have high royalties to pay like Pandora)
-Use a Free Trial when: It costs you a non-trivial amount of money for someone to use your product (this also helps prevent free account abuse) or if your trial is able to answer someone's question that is preventing them from directly converting. An example would be that if your product's code benefit is that it creates beautiful reports for clients, you might offer a free trial to prove to product leads just HOW beautiful those reports are before committing to a purchase.
The problem I see with Freemium plans (and I have actual hard data to back this up that I plan on blogging in the future) is that they cause a fundamental shift in your customer's mentality which often times leads them to not convert to a paying customer. When a customer gets the seed planted in their mind that there is possibly a way to get your product for free (even if it's crippled in features or usage limits), they will morph their own needs to match the freemium plan rather than the other way around.
TL;DR: you can't just slap a freemium plan on your product and assume success, you need to look at your customer data and design it perfectly to fit in with historical trends (which is what the OP is arguing for here actually). Freemium can be detrimental if not designed and applied correctly and in my opinion should be avoided most of the time by SaaS startups.