Too long by what metric? Personally, I dislike short trials. If I get distracted by something more important before I have time to fully investigate the product, I'm likely to just forget about it when the trial prematurely ends.

We have an API product and thought about this a lot. The 7 day trial does not work great as the users require to learn about the API / test calls etc.

Currently thinking about having a 7 day trial that starts ticking after say 10 successful API calls. Do you think that is a good compromise?

I don't think this solves the problem of people getting distracted. If I make 10 API calls and then get called away to do Super Important Thing for the next 6 days, I'm probably pretty likely to give up on a service if my trial period is reduced to 1 day.

Have you thought about N API calls for a trial? That's what I've used for API-first products in the past that worked pretty well for me.

No metric, again just gut feel. Overly long trials can also fall by the wayside. We don't particularly want a long sales cycle, but if near end of the trial, when you inevitable will hear from us, you'll know if you need more time or not.

Startups are an experiment after all, so let's see what works.

My thought for shorter trials: If something more important derailed the trial then the product isn’t solving a problem with enough urgency for you and you were unlikely to convert.