[ysapir]

A usage-based trial is really just the same thing as a time-based trial. There is a certain pace at which users are going to use up their usage quota, and that translates to a time period. Generally, though, you want the time period to end after the user has seen the benefits and before they have exhausted their initial enthusiasm or encountered any bugs. Giving a user four months to make their choice probably puts you on the wrong end of all those considerations.

After the time-based trial, there are going to be different packages based on different usage quotas anyway. KISSmetrics is probably catering to very different and more established customers than the author, and their pricing model is structured accordingly.

[cooperadymas]

Slightly tangential, but the KissMetrics 14 day trial makes little sense to me regardless of their target customer base. With an Analytics package you need time to set it up, then time to gather data, then time to analyze it to see how beneficial it is. 14 days just isn't enough to do anything of the sort. That said, I am a satisfied user of KissMetrics. Both their sales and their support team are incredibly helpful and responsive.

The author compared KissMetrics to MixPanel, two services that offer relatively the same thing. MixPanel gives a free account up to 25000 events instead of a 14 day trial. This is a much, much more useful way of introducing the product to a new user.

[toumhi]

> Generally, though, you want the time period to end after the user has seen the benefits and before they have exhausted their initial enthusiasm or encountered any bugs.

Corollary: do not end the trial, whether it is time-limited or usage-limited, if the user couldn't verify the expected benefits before it's over.

For example, at my company, we help vacation rental owners to publish their property on multiple websites. We've learnt that 1 month was, in fact, too short a trial for people to fully appreciate our service (for reasons that would take a long time to explain here). YMMV.