Question, how many beta testers do you need as a minimum, if you want to do statistical analysis, such as:

2 versions of a product, with 50% of beta testers given one version, and the rest the other.

So you can compare and see which is used the most / least hours / etc.

That depends on how confident you want to be that the results aren't just statistical outliers. ;)

Also important: Have enough beta testers with each property combination contained in your target group.
Some basic properties are:
- Experience in the field your software is used in
- Sex, age, general tech affinity, main desktop OS, main phone OS, education
- Income (you may find surprising correlations for feature or UX design preferences)

@Oktokolo

Super confident. :-)

At an uneducated guess, my number is 20.

@Nanos
If the two products are in a testable state - why not add some telemetry and let it be tested by strangers wich fill out a form (and accept the EULA including telemetry) and can then use the software for free.
That way, you can have hundreds of testers easily...

@Oktokolo

I want to limit the number of testers, due to issues such as, only so much server capacity.

I'd still need to know the minimum number though, since what if it wasn't popular. :-)

@Nanos
Mathatically a group you can probe starts at 30.
So for two fields, I'd recommend at least 60 testers.