My 2009 ‘Music’ folder was 1.2GB of naming-conventionless MP3s, 13 songs that were secretly other songs, and 1 virus.

I'd queue a LimeWire download before bed and find out in the morning whether "W0nderwall.mp3" was Wonderwall.

Last week we broke down how iTunes got the world to pay 99 cents a song.

Apple's offer was basically: give us a dollar and you can skip the parts of the album you didn’t want anyway, along with the viruses + guesswork + ‘am I gonna get busted’ factor that came with torrenting.

Plenty of us still torrented away. Most of us ended up on Spotify.

Today's piece is on how Spotify ended up beating iTunes and piracy at the same time.

Enjoy.

— Tom

Omnisend now works with ChatGPT and Claude. Ask it to score your email setup based on campaign cadence, automation coverage, and catalog completeness, then get your top three improvements in seconds.

Not on Omnisend yet? Brands average $79 back for every $1 spent - migration is free in under 5 days, and pricing runs up to 35% lower than Klaviyo's.

Hunting for AI roles takes hours you don't have. Careers pages, group chats, internal referrals, a dozen job boards, all updated at random. The Athyna AI Job Board does the watching for you.

Set up a profile once. The matches come to you.

_{Thank you for supporting our sponsors, who keep this newsletter free.}

The what: A TLDR explanation of the strategy

Blink once.

Knock a domino over.

Say the word “Spotify”.

Each of those things take more time than it takes for Spotify to start playback of a song.

In 2009, a Wired writer described the Spotify beta as "a magical version of iTunes in which you've already bought every song in the world."

Not bought. Not downloaded. Just there, the instant you hit play.

→ That feeling was the product.

Spotify launched in 2008 into a market that had already lost. The fastest way to get a song was to steal it, and no legal company could beat the convenience of a file downloaded straight to your hard drive, for $0.

Spotify founder Daniel Ek stopped competing with iTunes and started competing with The Pirate Bay, on the one important axis that piracy hadn’t quite nailed: speed.

The obsession was utterly total.

They got a song to start in a median of 265 milliseconds, faster than most brains register a wait.

Behind-the-scenes, they assembled an engineering masterpiece, with only 8.8% of the audio coming from Spotify's own servers.

The rest came from your computer’s cache and, for years, from strangers' laptops.

You read that right. Let’s learn how.

The how: The strategic playbook boiled down to 3x key takeaways

Every music startup in 2008 was trying to out-iTunes iTunes. That would mean a bigger catalogue, a slicker store, better price.

Spotify ignored all of it.

They noticed that the real winner in the digital music battleground was trending towards another player entirely: piracy.

And while piracy had a limitless catalogue and unbeatable price, users did have to spend a lot of time hunting for safe seeders, then wait for files to download and import into their media player of choice.

So Spotify built against that. A streaming engine that felt like the file was already yours.

Spotify's own product and engineering lead put it bluntly:

Napster and LimeWire had already trained a generation to expect all music available at their fingertips. Spotify couldn’t argue with that expectation.

So it met it → full catalogue + artist pages + playlists on top → a near-perfect playback experience.

Piracy had already written the spec for user behaviour. Spotify just had to legally build and ship it.

You can't obsess over milliseconds without someone who can shave them. Spotify's answer was to buy the best low-latency engineer in the world.

In 2006, they acquired µTorrent, the wildly popular BitTorrent client.

They ended up selling it only months later but that actually didn’t matter.

The code was never the point.

Strigeus, µTorrent's creator, went on to build Spotify's original streaming engine, the peer-to-peer core that made instant playbacks possible, using the same family of technology that powered piracy itself.

So what exactly did he build?

Instant playback is a trick of attention.

(Much like Instagram’s ‘instant upload’ UX magic we wrote about here)

Latency is dominated by the first few 100 milliseconds, so Spotify made the opening of a song as fast as possible, and let the rest arrive however was cheapest.

So what happens when you hit play? 4 things:

→ Cache first. 55.4% of the audio you heard came straight from your device's local cache. Zero network, totally instant.
That's music you already played, still sitting on your hard drive. Spotify never deleted it after playback, so hitting a song twice isn't streaming at all, it's just opening a file.

→ A server head start. If the track wasn't cached, Spotify streamed roughly the first 15 seconds from its own servers over an already-open connection.
The client holds a live connection to Spotify the entire time you're logged in, so there's no dial-up ‘time to get connected’ moment when you hit play. The request goes down a pipe that's already active, and it only asks for the opening seconds, not the whole file.

→ Peers for the bulk. While that intro played, your client pulled the rest of the song from other users who already had it, 35.8% of all data (those strangers' laptops).
You only need the first 15 seconds urgently. The remaining 3 minutes have minutes to show up, so Spotify handed that job to strangers to save themselves the bandwidth bill.

→ Pre-fetch the next track. Before the current one ends, Spotify loads the next song, so "hit next" is gapless too.
In the closing seconds of the current song, Spotify quietly pulls the opening of whatever's up next into your cache, so by the time you press skip it's already ready to play.

Throw these together and you have a ~265ms median start, with <1% of plays stuttering, and only 8.8% of bytes served by Spotify itself.

And when 265 milliseconds still wasn't instant enough (the average human brain can detect down to ~200ms), Ek admitted the last fix was psychological:

The where: 3x high-signal resources to learn more

One of the denser reads, but well worth it.

2 Spotify engineers show their working: how half your music came from other people's laptops and you never noticed.

Shows exactly where the 265ms median actually comes from, how the cache → server → peer split measured, and why only 8.8% of bytes ever touched Spotify's servers

Reid Hoffman gets Ek to confess the 200ms miss, and how a moving loading bar covered for it.

→ Why they obsessed over latency when no one asked

→ The throbber trick, in Ek's own words

→ Building trust when the product has to feel instant

35 minutes on why perception is a product surface.

Recorded at Spotify HQ, Ek walks the whole arc from racing pirates to running the audio internet. The definitive origin deep-dive, the business behind the milliseconds.

3 hours wasn’t enough.

Our flagship course on how to use free internet data to make better strategic decisions. Contains 5 years of strategy expertise, proven methods, and actionable tactics to accelerate your career with modern-day strategy skills.

We have a growing audience of 150,000+ strategists from top companies like Google, Meta, Atlassian, Stripe, and Netflix. Apply to feature your business in front of Strategy Breakdowns readers.

One of the most common questions we get asked is: “What tools do you use to run Strategy Breakdowns?” So, we’ve open-sourced our tech stack to give you an inside-look at exactly what tools we’re using to power each corner of this operation.