“Can we get the stems?” is not a delivery plan

In advertising production, stems are useful only when they serve a versioning plan. Asking for “all the stems” at the end of post can feel safe. In practice, it often creates a folder of inconsistently named files, with levels that do not match the approved mix, edits that no longer land on picture and a media team still waiting for platform-specific masters.

The better question is sharper: which audio elements need to survive cutdowns, vertical formats, market adaptations and legal approvals?

That question applies to original score, production music, sound design and final mix. It should be asked before picture lock, or at least before the final audio pass. Otherwise the music becomes a fixed block, while the campaign continues to multiply into 60, 30, 20, 15, 10 and 6 second versions, sometimes with voiceover, sometimes without, sometimes sound-on, sometimes watched muted.

If you are planning a campaign with several deliverables, this page sits alongside our guide to original music for advertising and our practical note on music and sound design in postproduction.

Stems, multitracks and versions are different things

A stem is a usable submix. It is not necessarily every microphone, MIDI lane or synth layer. For a commercial campaign, stems are usually grouped by function: drums, bass, harmony, melody, textures, impacts, risers, musical sound design, and sometimes sung vocals if they exist and if the rights allow them to be separated.

Multitracks are more granular. They are useful when someone needs to reopen the production, change the arrangement or rebuild part of the cue. They can be essential on complex work, but they are heavier to check, archive and share. Asking for them without a clear reason can create more risk than control.

Versions are different again. They are editorial masters: 30 second TV, 20 second online, 15 second vertical, 6 second bumper, no voiceover, no music, international, local tag, end-card variant, or a market version with different supers.

A serious delivery plan separates these three levels. First, the approved masters that are ready to traffic. Second, the stems that help mixing and adaptation. Third, work elements, but only if a named team actually needs to reopen them.

This distinction avoids a common mistake: assuming stems can replace a strong edit strategy. They cannot. Stems create flexibility only when the musical architecture has already been designed to flex.

Cutdowns should influence the composition, not just the export

A piece of music built only for a 60 second film may collapse in 15 seconds. The introduction is too long, the lift arrives too late, the sonic signature lands after the logo, or the most ownable musical gesture disappears in the short version.

When the future cutdowns are known early, the score can be designed with multiple entry points. A motif can appear before the first cut. A texture can work without the top line. A build can be shortened without sounding amputated. An impact can support the product moment without crushing the voice.

Concrete example: a brand needs a 45 second hero film, a 20 second pre-roll, a 15 second vertical social cut and a 6 second bumper. If the music spends its first 12 seconds setting atmosphere, the bumper will keep only a decorative fragment. If the composer knows from the brief that a bumper is required, they can write a two second gesture that works at the logo, then expand it across the longer film.

This is where a music agency becomes a production partner, not simply a track supplier. It anticipates use cases, prepares clean edit points and checks whether the musical identity still reads in the shortest formats.

For the briefing stage, see our guide on how to brief a composer for an advertising film.

Platform specs are not just technical footnotes

Specs vary by publisher, territory and media plan. Google Ads documents video requirements around formats, ratios, file size and recommended durations. Meta’s Reels guidance is built around vertical video with audio, while still depending on the user’s sound setting. EBU R 128 provides a widely used European framework for loudness measurement, including integrated loudness and true peak descriptors.

The point is not to memorise one universal number. There is no universal number. The point is to obtain the delivery specs before the final export session.

Within one campaign, the TV mix may need a broadcast loudness target, while social versions need intelligibility checks on a phone speaker. A cinema master, if required, is not the same decision as a mobile-first social export. A connected TV asset may sit between broadcast discipline and digital delivery rules.

That does not mean every cutdown must be mixed from scratch. It means the delivery list should identify the required masters, loudness target, codec, language, voiceover status, picture ratio and exact duration.

Without that list, the final day becomes an emergency export queue. With it, the mix can be prepared, checked and named in a way that production, agency and media teams can all understand.

File naming is a production decision

Clear naming prevents expensive mistakes. A file called `final_v8_NEW_reallyfinal.wav` helps no one. A simple convention should identify project, duration, format, language, status, audio type and date.

For example: `PROJECT_FILM_30s_EN_TV_MIX_FULL_20261008.wav` and `PROJECT_FILM_15s_INT_SOCIAL_MUSIC-STEM_20261008.wav`. It is not poetic, but it is legible. In international delivery, legibility matters more than elegance.

Folder structure matters too. One folder for approved masters. One for stems. One for references and reports if required. One for superseded versions if they must be retained, but never mixed with current approved exports.

You also need to define who can approve audio. An editor may request an export to test a cut. A producer may ask for a no-voice version for client review. A media partner may request a technical format. But the approved master should be signed off by a clear person or approval chain.

This sounds administrative. It protects the creative work. It reduces the chance that the wrong version is trafficked, localised or used as the basis for a new edit.

The serious limitation: stems do not solve rights or weak thinking

More files do not automatically mean more safety.

If the territory, term, media or rights holders are unclear, tidy stems will not fix the legal problem. If a sung vocal includes a sensitive lyric, an instrumental version may help, but it does not replace rights review. If the musical idea is generic, cutdowns will not make it ownable.

Stems can also create a false sense of freedom. A local team may rebuild the music without understanding the intention, move an impact, cut a decay or use an isolated stem as if it were a finished master. That is why usage rights, modification limits and versioning rules should be written into the brief and delivery recap.

For multi-territory work, this connects directly to our article on music rights for an international advertising campaign.

What to send before final mix

Before requesting exports, send a short delivery grid. Exact durations. Picture ratios. Countries and languages. Media. Launch date. TV, cinema, online, social, DOOH or event use. Voiceover status. Need for no-voice, no-music or no-sound-design versions. Technical specs from the publisher or broadcaster. Named approval contact.

Also say what is not needed. No open session if nobody will open it. No music stems if the campaign has no adaptation path. No instrumental version if it is neither useful nor permitted.

Good audio delivery is restrained. It gives the campaign enough material to adapt, without flooding teams with files they will never use.

If your campaign has multiple films, markets or short-form assets, Sul Tasto can help organise the music, stems and versioning before the mix becomes a bottleneck.