Resources
8 items
How to run a study. Then open one.
Guides, docs and the blog.
GuideVideo
How to read a result
The split, its range, each audience, and why.
GuideDocs
The user guide
Set up a study, run it, read the results and share them.
GuideDocs
The API
Find audiences, run studies and read results from your own code.
GuideDocs
Planetarium in your AI assistant
Connect Claude, Cursor or VS Code and run studies from a conversation.
Blog post5 min read
We forecast Amazon's October sale before it happened
Most accuracy claims are scored after the answer is known. This one was written down first.
Blog post5 min read
How we score an A/B verdict on 1,500 headlines we had never seen
A demo shows a tool at its best. A held-out test shows it at its average. This is how we tested Planetarium’s A/B verdicts, in plain terms.
Blog post5 min read
Why an audience states its split instead of answering 40 times
The obvious way to imitate a crowd is to ask one voice many times. We found a better way, and an honest list of where it falls short.
Blog post5 min read
What an audience is, and what it will never say about a person
An audience is a summary of real people, not a person. Here is how it is made, checked and limited.