Trending audio on Instagram: the most repeated myth in the creator economy

You saved the audio. You added it within the first hour it went viral. Three days later the reel has two hundred views and the one before it got thousands. Same account. Same time of day. You look at both. The only difference you can see is that the first one used your own voice.

So you go back to the videos that told you to do it. Use trending audio. The algorithm loves it. Everyone says so. It is the single most repeated tactic in the creator economy.

We went looking for where it appears in Meta's own writing. It does not.

Here is what we read. Meta publishes a page for each part of Instagram. Each page explains what that part's software looks at when it decides what to show you. Meta calls them system cards. We read the one for Reels Chaining. That is Meta's name for the feed of reels that keeps going once you open one. We also read the cards for Feed and Explore. And we read what Adam Mosseri wrote about ranking. He is the head of Instagram. Five documents. Retrieved 17 August 2026.

The Reels card does list audio signals. It lists them twice. But the thing they measure is not what the advice says. The signals track whether you engage with audio. Not whether the audio itself is trending.

That gap is the whole problem. Creators are chasing a boost the system never promised. The advice turns on a guess about what "audio track" means in one sentence Mosseri wrote three years ago. Almost no one went to the page that breaks it down.

The breakdown matters. Every source below is linked, dated and quoted. Tactics like this one last because nobody can predict which posts go viral, and a sound you can copy feels like an answer.

What the Reels system card actually says about audio

The Reels Chaining system card was stamped 11 November 2025. That is the date it says "updated". Meta rewrites these cards without saying when. So every number and quote here carries that stamp.

The card lists what the software tries to guess about you. Meta calls these predictions. Each prediction has signals feeding it. Signals are facts the software is allowed to read.

The card lists ten predictions. Two of them mention audio. Both appear under the same heading. Here is the full text of that heading:

How likely you are to use the audio from a reel you're viewing in one you create

That prediction appears twice on the card. Both times it lists the same four signals. Here they are, verbatim from the card:

• Whether or not you've been viewing reels for more than 15 seconds
• How many times you've clicked the audio link in reels you've seen
• How many times you've clicked 'Use audio' after clicking the audio link compared to the number of reels you've seen
• How many times you've clicked 'Use audio' after clicking the audio link

Read those four. Every one measures something you did. Whether you viewed long enough. How many times you clicked the audio link. How many times you clicked "Use audio". The subject of every sentence is you.

None of them measures the audio itself. Not how many other people used it. Not how many views reels with that audio got. Not whether it is trending. The signals ask: does this viewer engage with audio as a feature? They do not ask: is this audio popular?

A second prediction mentions audio once. The heading:

How likely you are to watch less than three seconds of a reel

One of its signals reads:

How many times this reel has been viewed with the sound on

That counts views. It does not count which audio was playing. It measures whether people left the sound on while they watched. That is a fact about viewer behavior. Not about the audio's popularity.

Those are the only mentions of audio on the Reels Chaining card. No prediction asks whether the audio is trending. No signal measures how popular the sound is. The closest thing to "trending" is a count of how many times one reel was viewed with sound on. That is one reel. Not every reel using that audio.

The sentence that started it all

The advice comes from somewhere. It comes from one sentence Adam Mosseri wrote on 31 May 2023. His post Instagram Ranking Explained covers four parts of the app. Feed, Stories, Explore and Reels. Under Reels he lists the signals that matter most. The list is not in order. Under "Information about the reel" he writes:

These are signals about the content within the video such as the audio track or visuals in the video, as well as popularity.

That is the whole sentence. "The audio track" appears once. It sits in a list next to "visuals" and "popularity". All three are "signals about the content within the video". Mosseri does not say the audio's popularity. He writes "audio track" and "popularity" as two separate things.

The same post names one prediction about audio. Under what Instagram tries to guess, he lists these. How likely you are to reshare a reel. Watch a reel all the way through. Like it. And go to the audio page. Going to the audio page is described in brackets. It is "a proxy for whether or not you might be inspired to make your own reel."

So Mosseri names the audio track as a signal. And he names going to the audio page as a prediction. He does not say the audio's popularity boosts the reel. He does not say trending audio gets more reach. The phrase "trending audio" is not on the page.

Three years later the detailed card came out. It breaks "the audio track" into specifics. The specifics are all about viewer behavior. Whether you click the audio link. Whether you click "Use audio". How long you watch before you decide. None of them measures the sound's popularity.

The gap between the advice and the document is total.

What Mosseri said in January 2025

Mosseri posted a video on 22 January 2025. The post is at instagram.com/p/DFFyRp-pINJ/. We could not open it directly. Instagram posts are built by code in the browser. A plain fetch returns almost nothing. So we are relying on what Social Media Today reported about it, published the same day. Their article quotes him directly.

His words, as Social Media Today quoted them:

The top three signals that matter most for ranking are watch time, likes and sends. So when looking at your insights, pay close attention to average watch time, likes per reach, and sends per reach.

Three signals. Watch time, likes and sends. Sends means shares. He also said:

Likes are slightly more important for connected content, and sends are slightly more important for unconnected content.

Connected content is posts your followers see. Unconnected content is posts shown to people who do not follow you.

Audio is not in that list. Neither is trending audio. Mosseri named the three signals that matter most. He could have named audio. He did not.

That is the most recent statement we have from him on what drives ranking. It is eighteen months after the 2023 post everyone cites. Audio is still not named as a primary lever.

The claim no one can source

We went looking for a single on-record statement from Mosseri or Meta saying trending audio boosts reach. We found none. Not in the five ranking documents. Not in the system cards. Not in the January 2025 update.

We also looked at what gets repeated. One site said Mosseri confirmed in a July 2026 AMA that using trending audio does not hurt you. The quote given: "We're not going to punish you for posting content." We cannot verify that. July 2026 has not happened yet. The date is wrong. And no link to the original AMA was given. So we do not cite it.

The claim is everywhere. The source is nowhere.

Here is what we can show. The Reels card lists audio signals twice. Both times they measure your behavior with audio as a feature. Not the audio's popularity. Mosseri's 2023 post mentions "the audio track" once, in a list. His 2025 update names three top signals. Audio is not one of them. No Meta document we read says trending sounds get boosted.

What the system does measure about audio

The Reels Chaining card is clear about what it tracks. It tracks whether you engage with audio. As a person. Here is what that looks like.

You watch reels. You click the audio link on some of them. You tap "Use audio" on some of those. Or you do not. The software watches what you do. It builds a guess. This person engages with audio, or they do not. This person might want to see more reels whose audio could inspire them to create. Or they might not care.

The Feed Recommendations card covers posts from accounts you do not follow. It was stamped 29 June 2026. One of its predictions: "How likely you are to turn on the audio of a video post." Three of its signals for that prediction:

• How many times you looped other video posts
• How many times others turned on audio on the post
• How many times you turned on audio on other posts

Again the subject is you. How many times you turned on audio. How you behaved on other posts. One signal counts how many times others turned on audio on this post. That is the closest thing to a popularity measure in any audio signal we read. And it measures behavior. Whether people turned sound on. Not which sound was playing.

The Instagram Feed card covers posts from accounts you follow. Stamped 29 June 2026. One prediction: "How likely you are to turn the sound on for this video." Its signals:

• Whether the post is a recommended post
• Whether you turn the sound on while watching
• Whether the post is a video/Reel

All three are about you or about the post's format. None measures the audio's popularity.

The Explore card was stamped 22 June 2026. It lists twelve predictions. Not one mentions audio. Explore is photos and videos. The tab that introduced millions of people to accounts they did not follow. And its system card has no audio signal anywhere on it.

So four surfaces. Feed, Feed Recommendations, Explore and Reels Chaining. We read all four cards. Only two of them list audio signals. Feed and Feed Recommendations ask whether you turn sound on. Reels Chaining asks whether you engage with audio as a creator tool. None of them asks whether the sound is popular.

What "the audio track" could mean, and what it does not

Mosseri's sentence from 2023 is short. "Signals about the content within the video such as the audio track or visuals in the video, as well as popularity." One reading is this. "The audio track" means the system looks at which track it is. Then it checks how popular that track is. That reading makes the advice true.

The other reading is simpler. "The audio track" is one kind of content. Like visuals are one kind. And popularity is separate. This reading matches what the detailed card says three years later. The card breaks audio into signals. Every signal is about the viewer. Not the track.

We cannot prove which reading Mosseri meant in 2023. What we can show is what the system card published in 2025 actually measures. And it measures your behavior. Not the sound's popularity.

There is a second problem. Even if "the audio track" did mean its popularity, that does not say the direction. A signal is a fact the software reads. It is not a rule that says up or down. The card lists "How many times this reel has been viewed with the sound on" as a signal. More sound-on views could mean the reel is good. Or it could mean nothing until it is weighed against skips. We do not know. Meta does not publish the weights.

The 2023 sentence says "as well as popularity". Popularity of what? The reel or the audio? The sentence does not say. And if it is the audio's popularity, is that a boost or just a fact the model sees? The sentence does not say that either.

So even the most generous reading of Mosseri's 2023 sentence does not give you the advice. The advice says: use trending audio and your reel gets boosted. The sentence says: the audio track is a signal. Those are not the same claim.

Why this myth will not die

The myth survives because it has everything a good myth needs. It is simple. It is actionable. And it cannot be disproven by a creator in real time.

You posted with trending audio. The reel did well. The audio worked. You posted with trending audio. The reel did badly. You must have done something else wrong. The premise is never tested.

The alternative is harder to accept. Your reel competes in a funnel that narrows billions of posts to hundreds. One person at a time. It arrives at a stage that scores posts by history. Yours has none yet. Some reels break through. Most do not. Quality sets a floor and a ceiling. But between those, the outcome is unpredictable. That is what the research on cold start says. And it is what the random experiments show. The best rarely do badly. The worst rarely do well. Everything else is possible.

Trending audio gives you something to do. That is worth something. But the thing it does is not what the system cards measure.

The one thing trending audio actually does

Trending audio does one thing the system measures. It can get people to click the audio link. That is a signal. If your audience cares about trends, and they recognize the sound, they might click through. The Reels card lists "How many times you've clicked the audio link" as a signal feeding the prediction about whether you will use audio to create.

So if your reel makes someone click the audio link, that is a positive signal. For that viewer. On that prediction. Which is one of ten predictions the expensive model makes. We do not know its weight. We do not know if clicking the audio link matters as much as watching to the end. Or as much as sending the reel to a friend. Meta does not publish the weights.

That is the most you can claim. Trending audio might make someone click the link. That is one signal. Feeding one prediction. Out of ten. With an unknown weight.

The rest is speculation.

The cards name the signals, not the weights

That last point is general. Meta publishes what the software is allowed to read. It does not publish how much any one fact counts. So the gap applies to every claim about ranking, not just the audio ones.

Take the strongest statement Meta has made about interaction signals. Its own engineering post from 2023 calls user-item interaction features "usually the most powerful" facts a model can read. That is a direction. It is not a number. A post with some engagement beats a post with none, and Meta has never said by how much.

The academic work on early advantage is mixed on size. One randomised field experiment found diminishing returns. The first donor to a Kickstarter project moved the outcome more than four donors together did, per donor. That ran on Kickstarter, not Instagram. But it shows the shape need not be linear.

So the honest state of it is this. A direction is sometimes documented. A size is not. Anyone quoting you a multiplier, for trending audio or for anything else, is quoting a study that does not exist. Ask them for it.

Where this leaves you

You now know what the Reels card measures. It measures whether you engage with audio. Not whether the audio is popular. Mosseri's 2023 post mentions "the audio track" as one signal among many. His 2025 update names three top signals. Audio is not one of them.

No Meta document says trending audio gets a boost. The claim is everywhere. The source is nowhere.

That does not mean trending audio does nothing. It means the thing it does is not what gets repeated. If your audience recognizes the sound, they might engage. If the sound fits the reel, it might keep them watching. Those are real effects. But they run through the viewer. Not through a popularity boost in the ranking software.

Use the audio that fits the content. If a trending sound works for your idea, use it. If it does not, do not force it. The system is not scoring you on whether the sound is popular. It is scoring you on whether people watch, like, send and click.

The three signals Mosseri named in January 2025 were watch time, likes and sends. Those are the ones he said matter most. Audio was not on the list.

So the next time someone tells you to chase trending audio, ask them where Meta said that. Then ask them to show you the page.

What we could not determine

Where the "trending audio boost" claim originated. We can show it is not in Meta's system cards. We can show Mosseri's 2023 and 2025 statements do not say it. What we could not do is trace the claim to its first appearance. It is in hundreds of videos and blog posts. None of them links to a Meta source. We do not know which video started it. Or whether it came from a misreading of Mosseri's 2023 sentence. So we cannot say how it began. Only that it does not match what Meta publishes now.

Whether the July 2026 AMA quote is real. One article quoted Mosseri saying "We're not going to punish you for posting content" in a July 2026 AMA. The date has not happened yet. It is likely a typo for July 2024 or another date. But the article gave no link to the original. So we could not verify the quote. We do not cite it. If it exists, it says using trending audio does not hurt you. That is not the same as saying it helps you. And it does not name the signals the system uses.

What the weights are. Meta publishes the signals. It does not publish the weights. So every claim about which signal dominates is unproven. That includes claims about audio. And it includes claims about early engagement. The direction is documented. The size is not.

Whether "the audio track" in Mosseri's 2023 sentence meant the track's identity or just the presence of audio. The sentence is ambiguous. The system card that came out two years later measures viewer behavior. Not track popularity. We report both. We do not average them.

Whether Instagram's ranking treated audio differently before the 2025 card. The oldest saved copy of the Reels Chaining card on the Internet Archive is from 2025. We could not read earlier versions. So we do not know if audio signals changed. We only know what the current card says.

How we did this

Scope: this post reads published documents. No data collection. No experiment of our own.

How we read them: each system card was opened in a browser on 17 August 2026. Instagram's Transparency Center pages are built by code in the browser. Downloading the URL directly returns almost nothing. So all five cards were read in a browser and every quoted sentence was copied as it appeared. Mosseri's 2023 post was retrieved the same day from about.instagram.com. His January 2025 statement could not be opened directly for the same reason. Instagram posts are client-side rendered. We relied on the Social Media Today article published 22 January 2025, which quoted him. That article is linked in the Sources.

Limits: four of five primary sources are system cards. All from Meta. A system card is a description Meta chose to publish. Not an audited account. Meta rewrites these cards without saying when. So every quotation here carries the card's "updated" stamp. The one it showed on 17 August 2026.

Sources

1. Instagram Reels Chaining AI system card, stamped UPDATED NOV 11, 2025. Meta Transparency Center. Read in a browser. Retrieved 17 August 2026.

1. Adam Mosseri, Instagram Ranking Explained, 31 May 2023. About Instagram. The audio track sentence appears under "How we rank Reels", subsection "Information about the reel". Retrieved 17 August 2026.

1. Andrew Hutchinson, Instagram Shares New Insights into its Algorithm Ahead of 2025, 22 January 2025. Social Media Today. Source of the Mosseri quotes about watch time, likes and sends. The article links to instagram.com/p/DFFyRp-pINJ/ as the original post. Retrieved 17 August 2026.

1. Instagram Feed Recommendations AI system card, stamped UPDATED JUN 29, 2026. Meta Transparency Center. Read in a browser. Retrieved 17 August 2026.

1. Instagram Feed AI system card, stamped UPDATED JUN 29, 2026. Meta Transparency Center. Read in a browser. Retrieved 17 August 2026.

1. Instagram Explore AI system card, stamped UPDATED JUN 22, 2026. Meta Transparency Center. Read in a browser. Retrieved 17 August 2026.

1. Vladislav Vorotilov and Ilnur Shugaepov, Scaling the Instagram Explore recommendations system, 9 August 2023. Engineering at Meta. Source of the "usually the most powerful" quote about user-item interaction features. Retrieved 17 August 2026.

1. Arnout van de Rijt, Soong Moon Kang, Michael Restivo and Akshay Patil, Field experiments of success-breeds-success dynamics. PNAS 111(19):6934 to 6939, 2014, doi:10.1073/pnas.1316836111. Randomised field experiments on four platforms including Kickstarter. Source of the diminishing-returns finding. Full text read at PubMed Central, PMC4024896. Retrieved 17 August 2026.

Changelog

• 2026-08-18: First version drafted.