I didn't understand this quote in the first line: only charge for things that are hard to do

I've found this to be absolutely untrue. Don't let implementation difficulty factor in the decision whether to charge for a feature or not. The only exception to this is when choosing not to do something altogether because it's too hard.

On Obsidian Portal, we charge for some of the simplest things (like the ability to send e-mail notifications of changes), while giving away some of the hardest things for free. In my opinion, it's about providing a useful product for free, while constantly presenting a premium product that would be more useful. The decision about what to charge for should only be made in terms of how useful it is to your users, not how hard/easy it is for you to do.

"The decision about what to charge for should only be made in terms of how useful it is to your users, not how hard/easy it is for you to do."

This was a lesson it took me a bit of time to learn. As a contractor/consultant I needed to figure out what to bill. I started out thinking in terms of how hard something was; if I had to bust my ass, then I was going to charge for it, and if something is easy, how can I justify a high rate?

Which, as it happens, is idiotic.

The real metric is, How much value am I providing?

Realistically there are some constraints, such as what are competitors charging (a concern if you don't want to chase potential customers away), but personal effort is not a main factor.

There's a belief that hard work should be rewarded, but that just encourages people to work hard regardless of the end value of their efforts. Valuable work should be rewarded. If you can do it with out breaking a sweat, more power to you.

You're kind of screwed if you go about thinking that because you worked hard on something that people will ultimately find it valuable and reward you. That only happens in some politician's stump speech fantasy.

Amen, Micah. Just between us geeks, I charge for a one-line if statement. My customers certainly don't perceive it that way, though: they see the paid version as uniquely allowing them to accomplish the task they've set out to accomplish.

Here's my problem with Freemium.

There is the mindset that you'll win the lottery, get bought out, and become rich. It's all about users, not revenue. It becomes a lottery, where a few get very wealthy, and the rest make nothing. That's not Freemium, of course, that's "free", as in twitter, facebook, etc.. However, to me, Freemium is almost as bad a bet.

With Freemium, I end up burning so many resources with scaling issues, trying to service all of these non-payers, hoping I'll "win the lottery", by getting enough actual payers to finance things. The examples they gave were good cases in point of services that struggled under the load of the non-payers for a few years.

I would like to just charge from the get-go, and avoid the scaling issues that freemium implies.

I think there's a fine line between freemium success and profitable failure.

If you give away too much, there's no incentive for users to pay for upgrades.

If you don't offer some sort of free account, it can be hard to get traction.

In the case of dropbox, it seems a lot of users are content with 2GB for free.

What would happen if they dropped that to 256MB. Would they get more conversions? Or would users simply move to a 2GB freemium competitor?

Hard to know what's best.

I wonder what would happen if they offered a 10-15 GB option at half the price of the 50GB option. I know a few dropbox users who keep hitting the 2GB limit, but don't want to upgrade because they think 50GB is more than they need and they'd feel like they're paying for something they're not using.

Freemium does not necessarily imply scaling issues. If you're doing a commodity business with marginal costs like file storage, you might actually worry about cost of goods quite a bit, but if you're primarily charging for scales-to-the-moon-IP rather than server resources, it is highly likely that you'll never have serious scaling problems and the scaling problems you do will be tractable to well-understood solutions. (Most of what makes Dropbox a great backup solution is the non-commodity part: the user experience, which a) rocks and b) is hard to duplicate and c) can be offered to the entire world for free. Then there's the portion of the business that is "save gigabytes of data to a spinning metal platter" -- that is a commodity with actual marginal costs.)

Take a look at 37signals, StackOverflow, or FogCreek, for example, in terms of the scaling needs of real, stupidly profitable businesses. I think you could probably fit all three companies' hardware in my closet and have room left over.

Or, at the lower end of the scale, take my business. On a very good day I might get a thousand people to sign up for the software. That's a huge number for me (more people than I know in the entire world!), but it is a very little number for a computer, even a modest fraction of a modest server sitting in Slicehost's rack.

or Seth Godins' 'you can get my book free as an email.'

I basically felt the same way until someone pointed out that if you don't have a free version, your competitors probably will and they will have access to a wider audience as they roll out new features, etc. (through their large email list)

So, I'm still debating as my app is currently paid-only but this is a pretty important point to consider.

One way to look at it, if the idea of "free" is too troublesome, is to consider that at the outset people are paying with their time and attention.

You need to do something to attract users, and allowing non-paid usage may be cheaper than the cost of additional advertising that gets people to sign up and pay.

Good point. This is also where lifetime value vs. cpa will become a really important metric to watch.

Everyone would like to charge from the get go, but the trick is convincing customers to pay. Paying something is a huge psychological barrier, and it is so much easier to get people on board for free.

All I read from these case studies is that unless your product is targeted at the masses, you should avoid Freemium. 0.5% ratio of premium users is simply too low.

A ratio for 'freemium' as a whole isn't really all that informative and you shouldn't make any decision based on it. I think you can expect completely different results depending on your specific case. 30% & 0.03% are both possible in different circumstances.

Also, since the definition of 'user' is pretty loose, you can't really use these numbers to make any assumptions about the cost of serving them. If a free user stops using a service, he doesn't cancel his subscription. A paid user does. Freemium users are also hard to distinguish from trial users since the two are often amalgamated.