Video vs static ads for musicians

The claim, and why we are not repeating it

The most-cited version reads roughly: images work best for retargeting warm audiences; for cold audiences video wins almost every time. It is stated flatly, as if measured. There is no study behind it, no sample, no account data, no link.

It is also entirely plausible. Probably true and definitely unsourced is the hardest kind of claim to handle honestly, and the easy move is to restate it in your own words and inherit the authority without the evidence. That is how folklore spreads here. So instead: the mechanism that would make the rule true, an explicit label saying this is reasoning rather than results, and the test that tells you whether it holds for your act.

The mechanism that would make it true

A cold viewer has not heard the song

Music is an audio product. That sounds obvious and it is the entire argument. A still image of your cover art can tell a stranger a release exists, what it is called and roughly what world it belongs to. What it structurally cannot do is let them hear it.

So a static ad to a cold audience asks someone to click through on the promise of a song, based on a picture. A video ad can play four seconds of the song and let the thing you are selling do the selling. If the music lands, that is a far shorter path from stranger to interested. If it does not, the non-click saves you a wasted landing-page visit. Video also produces intermediate signals a still image cannot — watch time, partial views, completions — which become audiences you can retarget later. That is documented platform functionality, not a performance claim.

A warm viewer already knows the song

Now the job changes completely. Someone who visited your release page or saved the track does not need an introduction. They need recognition and a reason to act now. Recognition is what a still image is best at: cover art resolves in the fraction of a second someone spends scrolling, with no watch time, no sound and no hook to survive. It is also the exact image they already associate with the song, an advantage a new video does not have.

There is a production argument too, and on small audiences it becomes decisive. A warm pool of a couple of thousand people burns through creative fast — see the arithmetic on a small warm audience — so you need several variants, repeatedly. Static is far cheaper per variant. Even if video performed somewhat better per impression, the format you can refresh weekly may beat the format you can produce twice a year.

Why this is a mechanism and not a fact

Several things could make the reasoning above wrong for you.

• Your artwork might be unusually strong or weak. The rule treats “static” as a format. In practice it is one specific image, yours, competing against everything else in the feed.
• Your warm audience might not be warm. A pool built from broad video views or a long retention window contains people with only the faintest memory of you. For them the cold logic applies.
• Delivery optimises for your objective, not a format. Two formats can produce very different click rates and near-identical costs per actual outcome, which is why a comparison judged on clicks can point the wrong way entirely.

So the honest status of the rule is: a reasonable prior worth starting from, and worth abandoning the moment your own numbers disagree.

How to test it on your own campaigns

1. Change one thing

Same audience, same budget, same objective, same landing page, same offer. If the video also has a different call to action or points somewhere else, you have not tested format — you have tested two ads.

2. Do not put both in one ad set

This is the mistake that quietly ruins most self-run tests. Delivery is an optimiser, not a splitter: it concentrates budget on whichever ad it predicts will do best, often within hours and on very little data. You end up with one format having spent almost everything and the other nearly nothing, then read that as a verdict. Use separate ad sets with the audience split, or the platform's own A/B test tool, which exists specifically to hold delivery constant.

3. Measure the deepest event you can record

Not click-through rate. CTR rewards whatever attracts attention, which is not the same as whatever produces listeners, and the two formats earn clicks in different ways. Judge on cost per landing-page view at a minimum, better on cost per pre-save, email, ticket or purchase. That requires the event to reach the platform reliably, which is what server-side tracking is for.

4. Decide in advance what would change your mind

Write the threshold down before it runs. “If video's cost per landing-page view is a third lower, I move cold budget to video.” Deciding afterwards means deciding from noise.

5. Be honest about what a small budget can detect

Skipping this is why so much creative testing produces confident nonsense. Statistical power depends on how many outcomes you observe, and an artist spending modestly for a week observes very few. So you can detect large differences, not small ones. If video comes back twice as efficient, believe it. If it comes back slightly better, you have learned nothing and should not rebuild a strategy on it — accept the tie, run the format you can sustain, and spend your effort on a bigger question. Re-run it per release, too: the answer is about a specific song and image, not a law you establish once.

Production reality for an independent artist

Most advice here is written as though creative were free. The realistic starting inventory is cover art, some photos, and possibly one video — often a music video you cannot cut up quickly, or a phone clip. That constraint leads somewhere useful.

• Your one video is a cold-audience asset. If you only have one, put it in front of the people who have never heard the song.
• Cover art is a warm-audience asset with unlimited variants. Crops, colourways, a lyric, a release date, an “out now” treatment. Not infinite ideas, but enough to stop a small pool seeing the identical frame twelve times.
• Motion is not the same as production value. A slow push on a still, a waveform, a looping visualiser — all video as far as the platform is concerned, and all carrying the audio, which is the property that matters to a cold viewer.

The three-second hook

It has hardened into a rule: you have three seconds. Worth knowing where the number came from, because it is not a study of human attention — short-play thresholds exist as reporting and billing metric definitions in the platforms' own metrics glossaries, and the advice grew out of the metric. That is an origin story, not a measured optimum.

What survives is useful and specific to music: whatever you think the best moment in the track is, it needs to be near the front of the clip. A viewer who scrolls past before the hook arrives has functionally seen a silent picture — you paid video prices for a static ad. Do not save the good part.

Sound-off viewing, and the number we are not quoting

You will find a much-repeated statistic claiming a specific, very high percentage of Facebook video is watched with the sound off. It appears twice on one competitor's site. It traces to a claim from around 2016, predating enormous changes in how people watch video on phones, and no one links a methodology. We are not repeating it. The guidance survives without it, because it never depended on the number: video autoplays muted in several feed contexts, people scroll in offices and on trains, and a stranger is not going to unmute for you.

• The ad must work silent and be better with sound. If it is incoherent muted, you are relying on behaviour you cannot count on.
• Caption anything sung or spoken that carries meaning. Lyrics on screen are the cheapest way to hand a muted viewer what the audio would have delivered.
• Artist name and track title legible in the first frame, inside the safe area for the placement. Text under a caption bar is text nobody read.

Note that this argues for captions regardless of the true muted-viewing rate. If it is high, captions are essential. If it is far lower than the folklore suggests, they cost you nothing. That is why the guidance is safe to give and the statistic is not.

“I don't have a video”

This is the real blocker for most people reading this, and it is why the honest answer so often ends up being “static, because that is what exists.” Defensible for a warm audience; costly for a cold one, where the mechanism above says the audio is doing the work.

Our AI ad creator exists for that gap: cover art in, video ad out, priced per clip, so you can make a handful of variants and let the test above decide between them instead of deciding on principle. That is a claim about what the tool does, not a claim that AI-generated video outperforms anything — we have not measured that and will not imply it.

A note on numbers

There are almost none on this page, deliberately. No published, methodologically sound comparison of video against static creative exists for independent music advertising. The figures in circulation are self-attributed by companies selling something, borrowed from other industries, or stale — the sound-off claim being the clearest example.

What you get instead is the reasoning, labelled as reasoning, and a test you can run for the price of a week's budget. When we have first-party campaign data with sample sizes worth stating, it will appear here with a note on exactly what it measured. Until then: the mechanism is sound, the rule is a reasonable prior, and your own account is the only authority that matters.

Common questions

Where this fits

This is part of our guide to retargeting for musicians. If creative production is the bottleneck, the AI ad creator turns cover art into video ads priced per clip. And before you trust any format test, make sure the outcome you are judging on actually reaches the platform — that is what server-side tracking is for.