Your trial period should be reflect how long it typically takes for the prospect to derive some sort of value from your product.
Tools like HitTail and KissMetrics take time to prove their worth. They need to collect a certain amount of production data before being able to deliver value to the customer, so if I ran either of them I’d probably go for a 30 day trial period.
I’ve settled with a hybrid trial model with my app, Planscope, which is a project management app.
When you sign up, there’s actually no trial. But the signup process is pretty frictionless, and they’re locked into a sandboxed sample project which gives onboards them into a typical use case of the software. In truth, it’s more of an interactive “tour” page which also gives me the prospects email address.
But after you create your first client project, a 14 day trial kicks off. If someone is still using Planscope on a real project after a week or so (they’ve invited their client in, they’re tracking time, etc.) then they’re most likely going to convert — so I’m able to have a shorter trial period. Hell, I could probably try out a 7 day trial and the results wouldn’t change all that much.
I tried the model Jason referenced at MicroConf last year (see: http://planscope.io/blog/proposal-killing-the-free-trial/) and it bombed. The idea was that I’d do the perpetual no-limit sandbox project access, and capture CC details and charge immediately once their first client project is created. I think the “no trial” model works fine for companies like WP Engine because with hosting it either works… or it doesn’t work. Most SaaS products require “experiential buy-in” — the customer needs to like your workflow and design decisions, etc. And if they can’t try your product against the way they work first, they probably won’t be willing to buy upfront.
@Shpigford tried axing the free trial for his more traditional SaaS, Temper.io… Josh?