A test screening is watching your cut with fresh eyes before you lock it - not to ask "did you like it," but to find where attention breaks and comprehension fails. Filmmakers have run them in theaters for decades; a password-protected review link makes the same mechanism available to a 90-second corporate piece without renting a room. The method below follows the conventions documented by working filmmakers (No Film School's process breakdowns, John August on questionnaires), adapted for remote review.
When to screen
Late enough that the cut is complete and watchable end to end - usually at fine cut, before picture lock - and early enough that you can still afford structural changes. Screening a rough cut wastes the audience on problems you already know about; screening after lock means you paid for information you cannot use. The ladder context is in the editing-stages guide.
Strangers, not friends
Friends and family forgive - they fill comprehension gaps from goodwill and history, exactly the goodwill a real audience will not have. The consistent convention across filmmaking guides is a small audience of strangers (no magic number is authoritative; more strangers means more signal, and even three is better than your roommate). What matters is they owe the cut nothing.
Same-time screening vs review-link screening
| In-room screening | Review-link screening | |
|---|---|---|
| What you observe | Live reactions - laughter, fidgeting, phone-checking | Recorded comments at exact timecodes |
| Watch-through proof | You can see the room | View counts on the link confirm it was watched |
| Scheduling | Everyone at once, one room | Asynchronous over days |
| Notes quality | Verbal, needs a note-taker | Written, pinned to moments |
Remote screening has one trap the room does not: you cannot see who actually pressed play. Use a link with view counts (no account needed for viewers), and a password so the cut does not leak while you collect notes - the link-security basics are in review-link security.
Ask written questions, not "thoughts?"
The questionnaire is the convention precisely because "so, thoughts?" produces taste and "did you understand X?" produces diagnoses. Ask questions with checkable answers:
| Type | Example question | What it catches |
|---|---|---|
| Comprehension | "What is this video about, in one sentence?" | A muddled spine |
| Attention | "At what point, if any, did you lose interest?" | Dead stretches |
| Expectation | "What did you expect to happen right after the intro?" | Sequencing problems |
| Recall | "What was said about pricing?" | Whether the key message landed at all |
Collect answers in writing, individually, before anyone discusses the cut with anyone else - group discussion produces consensus, and consensus hides the disagreements you paid to see.
Stay silent
The rule every filmmaker account repeats: do not defend the cut. If a viewer misunderstood something, that is data about the cut, not about the viewer. The moment you explain what a scene "was supposed to do," you have converted your test audience into your critics, and the remaining answers are contaminated. Ask neutral follow-ups ("what made it confusing?") or nothing.
Triage the notes
Aggregate before reacting: three strangers flagging the same 40 seconds is a pattern; one person's idiosyncrasy is not. Sort into must-fix (comprehension failures), should-fix (attention dips), and consider (taste). That is the same triage shape as consolidating client notes, and the must/should/could tagging in the comment templates works unchanged. Then decide which findings belong in this round versus the next - revision rounds are a budget, and the rounds guide is how to spend it. If you want help clustering, AI summarization can draft the groups for human confirmation (the honest version of that).
FAQ
How many test viewers do I need?
There is no authoritative number - the convention is a small group of strangers, with more viewers meaning more signal. What matters more than count is that they are strangers and they answer in writing, individually.
Can I screen remotely, or does it have to be a room?
Remote works and scales better for short-form and corporate work: a password-protected review link, view counts to confirm watch-through, and comments pinned to timecodes. You lose live-reaction reading; you gain exact-moment notes.
What do I do with contradictory notes?
Treat contradiction as signal, not noise - it usually marks a segment that different viewers read different ways, which is a comprehension problem to fix, not a taste tie to break. Look for what both groups needed that the cut did not provide.
Should I screen the cut with music and final sound?
Yes, as close to final mix as you can - temporary sound changes how scenes read, and you want the audience reacting to the thing you will actually deliver. On the audio side, run the loudness pass before screening; the loudness specs are the QC numbers.
Do I need a test screening for a 60-second corporate video?
A scaled-down version, yes: three to five strangers, three written questions (comprehension, attention, recall), one password-protected link. The mechanism shrinks cleanly; skipping it entirely is how muddled message videos ship.