Experiment Library search, filter and results

The experiment Library is also a record of what your testing program has learned. You can find past tests by what they were about, see how each one turned out, and read the thinking behind it, all from the list.

Search more than names

The search box looks through an experiment's:

• name
• hypothesis
• notes (description)
• expected mechanisms, including any you wrote in yourself
• variant names
• outcome (for example "won" or "lost")

So searching for "delivery" finds a test about delivery dates even if its name doesn't mention them.

Filter by outcome and mechanism

Click Filter, next to the search box, to narrow the list:

• Outcome: Won, Lost, Mixed, Inconclusive or No data
• Expected mechanism: the mechanisms set in the experiment's hypothesis, plus Custom mechanism for any test with a written-in mechanism

Each option shows how many experiments it would leave. Choose an option again to remove it, or click Clear to remove both. The button shows how many filters are on, for example "Filter · 2". Switching between views such as Running or Ended clears the filters.

The Result column

On the All, Ended, Deployed and Archived views, the Result column shows each finished test's outcome and its lift, for example "Won +4.2%". It uses the same result as the Program page, so the two always agree. A test that was won by deploying a variant shows "Won · deployed".

Sort by Result to put winners with the biggest lift first. Running and draft experiments have no result yet.

Open a row to see what a test taught

Click the arrow at the start of a row to open it in place. You'll see:

• the hypothesis and any notes
• the expected mechanism
• the result, with the winning or losing variant, its lift and the range it's likely to fall in
• the variants, with any that were deployed marked
• a link to open the experiment

Open as many rows as you like; they stay open while you page through or re-sort the list.