The question is not only creative
Yes, you can test music for a commercial before you buy the final rights. But the permissions, files and audience need to be controlled.
This comes up constantly in production. The edit is still moving. The director likes one track, the agency wants a second route, the client asks for proof, and the research company needs a rough cut by tomorrow. At that point, music is no longer just inspiration in the edit suite. It has become a use that needs a boundary.
The risk is practical, not academic. Putting a track against picture can involve copying, syncing, exporting and showing the result to people outside the immediate production team. In France, representation and reproduction are protected uses. In the US, copyright owners have exclusive rights that include reproduction, derivative works and public performance, subject to specific limitations.
That does not mean every research cut needs a full campaign license. It means the team should separate three things: searching, presenting and exploiting. When those uses are separated early, creative choice becomes easier and the final rights process is less likely to damage the film.
If you are still weighing an original score against library music or a famous track, it is worth reading original score vs production music for advertising before you build the test.
Separate internal mock-up, closed test and pre-release
The first category is the internal mock-up. This is the version shared between editor, agency, director, producer, music supervisor and a small client group. It exists to judge direction, not to exploit the work. Even here, the file should be traceable. A ripped stream, a mystery MP3 or an unnamed download will cause problems later.
The second category is the closed research test. A group of consumers watches a rough cut, often inside a research platform, with limited access for a limited time. This is not a public campaign, but it is no longer only a production conversation. Before using a track, check whether presentation, research or testing is allowed.
The third category is pre-release. This includes retailer presentations, sales meetings, internal conferences, distributor previews, client portals, unlisted links that can be forwarded, festival style presentations and anything that moves beyond the people making the spot. In production language, it may still feel private. In rights language, it may not be.
Before you show any film with unlicensed music, write down who will see it, where they will see it, how long access will last, and whether the file can be downloaded or forwarded. Those four answers usually tell you whether a temporary permission is needed.
For a deeper look at temp tracks in commercial edits, see temp music in advertising.
What to clear before a music test
A useful music test starts with a small clearance note. It does not need to be legal theatre. It needs to answer the right questions.
Where did the track come from? A production music catalogue, a publisher, a label, an independent artist, a commissioned composer, a stock platform, an AI tool? Can it be placed against advertising picture for evaluation? Is the test limited to the agency and client, or will consumers see it? How many exports will exist? Can the client keep the file after the test?
For an existing song, separate the composition from the recording. The composition is the song itself, including music and lyrics. The recording is the master. Testing with the original master may be harder than producing a temporary cover to judge tempo, tone and structure. But a cover does not remove the need to deal with the underlying song. It only changes the master question.
For commissioned music, the composer agreement or purchase order should say whether demos may be used for agency presentation, client presentation, closed consumer research and internal case films. Without that language, people often assume the demo is cleared because it has been paid for. Payment alone is not a usage map.
For AI-generated music, document the tool, terms, important prompts, available training data assurances and transfer limits. Do not let an AI file enter the test as if it were an anonymous piece of production music.
A concrete example: three routes for a fragrance film
Imagine a 30 second fragrance film. Picture is not locked. Three music routes are alive.
Route A is a famous song. It has instant emotional value, but it is expensive and slow to clear. Route B is an original hybrid score with restrained strings, granular synthesis and breath-like sound design. Route C is premium production music, fast to license but less distinctive.
The weak process is to cut all three, send them to the client, wait for the favourite and then start rights work. If Route A wins emotionally and then fails legally or financially, the project does not simply go back to zero. It goes back to zero with a disappointed client, an edit shaped around an unavailable track and less time left.
The stronger process is to give every route a status. Route A can only be tested if the publisher and label allow limited presentation or research use. Route B is covered by a demo agreement that permits closed testing and includes an option for final production. Route C has temporary evaluation terms confirmed by the catalogue.
Then make the test fair. Same duration. Same perceived loudness. Same voiceover position. Same export quality. Otherwise the research is not comparing music. It is comparing volume, mix finish and editorial polish.
If the spot carries voice, read voiceover and music in advertising before trusting a winning test cut. A cue that wins without final copy can fail when the VO arrives.
How to avoid a false result
Music testing is easy to distort.
The first distortion is level. Louder often feels better. Before a research export, align the versions with a consistent approach and keep enough dynamics for the idea to breathe. Do not crush one route to make it feel finished.
The second distortion is editing. If one track forces a very different cut, you are no longer testing only music. That can be legitimate, but the question should be framed as a choice between film territories, not a soundtrack swap.
The third distortion is familiarity. A famous song brings existing memory. Sometimes that is exactly what the brand wants to buy. But research should separate immediate liking from long-term brand usefulness. An original cue may feel less spectacular in one viewing and become more valuable across cutdowns, radio, social, retail, events and sonic identity.
Do not judge the music only in the 30 second hero film if the campaign will not live only there. Ask for the hero cut, a short cut and an audio-only passage if radio or podcast will matter. The right answer often changes when picture is removed.
The serious limitation: some music should not be tested
Testing is not automatically wise. A film with a strong directorial point of view can be weakened by research that rewards the safest cue. A deliberately unfamiliar music territory can lose to whatever sounds most recognisable. A famous track can win a preference score because people already know it, not because it serves the brand best.
There is also a hard rights limitation. Some rightsholders will not permit testing. Some will charge for it. Some catalogues allow agency presentation but exclude consumer research. Some artists do not want their work associated with a product category, even temporarily. That answer is frustrating, but it is better to hear it before the edit becomes dependent on the track.
So a music test is not an insurance policy. It is useful only when the routes are lawful, comparable and still creatively open.
What to prepare before the export
Before exporting a test film, create a short cue note: track title, source, known rightsholders, licensing status, permitted use, test dates, territory, approximate audience size, platform, deletion date and responsible producer.
Add a creative note as well: tempo, role of the cue, relationship to voice, key musical behaviours to keep if the track changes. This is what allows a music team to replace a rejected cue without losing the idea.
If the route is original, keep stems or pre-stems available as soon as possible. If it wins, the team will need to mix, shorten, adapt and deliver versions quickly. For that next step, see stems, cutdowns and versioning for advertising.
If the test will influence production spend, bring the music team in before the first research export. A small rights frame at that moment can protect the idea, the budget and the final mix.