From Reflections on freemium by Joel Gascoigne:
If you don’t get your free limits right in freemium, it just doesn’t work. However, most think about this purely in terms of being too generous, and harming the free to paid conversion rate. I think the opposite is also a big risk: limiting the free plan too much, that you don’t provide enough value to drive genuine consistent free plan usage.
Any company that has success with freemium will always be at risk of the temptation to restrict the free plan more, or even drop freemium completely. The short-term A/B test will always favor dropping freemium, as if you test free plan vs free trial, the free trial variant will always likely win within any timeframe you could do an A/B test. The short-term revenue bump is hard to resist. The longer-term consequences are often not thought about enough.
Notes:
(1) For Seeking Alpha, freemium works for our core product. However, when we A/B test tighter restrictions on free usage, they always result in a higher short term conversion rate to paying subscribers. Yet we know that restricting free usage doesn’t maximize long-term subscriber growth. So our key challenge is to find a methodology to determine the right demarcation line between free and premium to maximize long term subscription growth. Ideas, anyone?
(2) Cf. Freemium works if 4 conditions are met.
(3) Cf. The 3 conditions of a great freemium business.
A Founder's Notebook 