The thing that made the length matter much less: grant extensions to anyone who asks, immediately, and treat each request as a research call.
A short trial with a generous extension policy gets you the urgency without the cost. Somebody who is genuinely evaluating and ran out of week hits a wall, emails you, and you say yes in one line. They are now a person you have spoken to, which is worth more than the fourteen extra days.
What it also gets you is the reason. Every extension request tells you exactly where people get stuck, waiting on a colleague, could not get the data in, was away. That is a list of your real onboarding problems, collected for free, from the people who cared enough to ask.
Two practical notes:
Make asking easy and obvious. A line in the expiry email. If people have to hunt for it, only the most determined ask and you lose the data.
Do not automate it into a self-service button. The value is the reply, and a button gets you the extension with none of the information.
The wider point for your situation: a trial length is not really a growth lever, it is a forcing function. What converts people is reaching the useful part. Anything that shortens the path there beats any number you pick.