Run a message test

Measure a video's effect on a randomized panel before the airtime is bought.

Most video budgets are spent on instinct. A message test measures a video's effect first: a randomized panel sees one video or a placebo, then says how they would vote.

Set one up

Open Tests in Survey. A test carries:

• its id, and the title respondents see
• its state: Draft, Approved, Open, Closed or Reported
• the target completes per arm
• the placebo video, and the arms: library clips with a video file, and every recording kept in Clips
• the entry questions, the explanation shown after the video, the ballot question and the comprehension question
• the panel vendor's redirects

Field it

• Field kit: the entry link a panel vendor uses, with their respondent id in place of {PID}.
• Monitor: completes per arm against the target.

To field to your own contacts instead, ask the Pollster in the Agentic Inbox: it mails personal links, reminds, and closes the test at its quota.

Close it

Close weights the respondents and computes each arm's lift. The result reads in Lift.

What respondents see

A bare page with nothing else around it. The arm is assigned before the first card shows; from a Pollster link, it was settled when the links were made, balanced within the group. Then one card at a time: the entry questions, the video (Continue unlocks only after it has played), the explanation, the ballot, the comprehension question and a completion code. A draft reads as closed to anyone but an admin, and only an Open test records answers.