Songwriting Analysis: Take a Song Apart with a Stopwatch

What comes out the other end

One three minute song, a phone stopwatch, a notebook, and about an hour. This is the page it produces:

0:00  Intro          4 bars   10s
0:10  Verse 1        8 bars   20s   8 lines
0:30  Pre-chorus     4 bars   10s   4 lines
0:40  Chorus 1       8 bars   20s   6 lines
1:00  Verse 2        8 bars   20s   8 lines
1:20  Pre-chorus     4 bars   10s   4 lines
1:30  Chorus 2       8 bars   20s   6 lines
1:50  Bridge         8 bars   20s   4 lines
2:10  Chorus 3       8 bars   20s   6 lines
2:30  Chorus 4       8 bars   20s   6 lines
2:50  Outro          4 bars   10s   2 lines
3:00  end

Eleven sections, fifty four lines, and a hook that arrives at forty seconds and gets sung ten times before the end. The song runs at 96 beats a minute, so a bar takes two and a half seconds and every duration in that column is just bars times 2.5.

Not one of those numbers is audible. You could play the song two hundred times and still not know a single one of them, because none of them reach you at listening speed.

Loving a song teaches you nothing you can use

The standard advice is to listen closely to the artists you admire. It is comfortable advice and it does almost nothing, because what you take away from a close listen is a feeling, and a feeling does not transfer to a blank page on a Tuesday night.

Here is the mechanical reason. At playback speed a song arrives as one object. Your ear processes texture, voice, and mood, all of which are production. The parts you actually need are durations and counts: how many lines the verse gets before the pressure builds, how long the listener is made to wait, how many times a phrase can come back before it turns to wallpaper. Duration is invisible to the ear in the way the length of a film shot is invisible to the eye. You feel "it kicks in". You do not feel "forty seconds".

So the defended position, and it is not a soft one: stop listening to songs for study. Take one apart with a stopwatch instead. An hour with the pause button gets you more usable craft than a year of admiring playlists, because at the end of the hour you are holding numbers, and numbers are instructions.

This is the reverse of working from a template. If you want the six shapes laid out as things to write into, that is a different job and it is covered in song structure templates. This article goes the other way: taking a finished record apart to find out what it did.

The procedure

Seven steps. You can run it tonight on one song. Use a song you half like rather than one you worship, because reverence makes you sloppy.

1. Play it once and mark where you drifted

One pass, no notes, finger on the stopwatch. The moment your attention slides, write the timestamp. Do not judge it and do not correct for it. That one number tells you where the song ran out of new information, and it will usually sit within a few seconds of a repeat you are about to find on the map.

2. Time every section to the second

Now the work. Pause, scrub, write the start time of every section: intro, verse, pre-chorus, chorus, bridge, outro, tag. Seconds, not vibes. When you have the starts, subtract to get durations. If you know the tempo, divide by the bar length and you get the section in bars, which is the unit songwriters actually trade in. Counting bars from BPM is its own small skill and finding a song's BPM covers the fast ways to do it.

What this reveals: the shape of the thing, on one page, which is the only form in which a song can be compared to another song.

3. Count the lines in every section

Literally count them. Eight, four, six. The counts matter more than the words, because they give you the density: a bridge with eight bars and four lines gives each line twice the musical room of a verse line. That is a decision someone made, and you can steal it.

4. Find the title and count its appearances

Where does the title first land, and how many times is it sung? Write both. Most people badly underestimate the second number. A hook you think appears "a few times" is often sung nine or ten times, and the gap between what a song does and what you remember it doing is the single most useful thing beginners learn here. More on building that line itself in how to write a chorus.

5. Mark every concrete noun

Go through the lyric with a pen and circle everything you could photograph. A door, a coat, a bus. Leave the abstractions alone. Then look at the density and, more importantly, at the distribution: where the objects cluster and where they vanish. Songs that feel vivid are almost never vivid all over. They are vivid in the verses and abstract exactly once, on purpose.

6. Find the one line that is not like the others

Every good lyric has one. A line that breaks tense, or drops the imagery, or addresses someone new, or is suddenly four words shorter than its neighbours. Find it and work out what it is doing. This step separates analysis from transcription, and it is the only step that requires judgement.

7. Write down what the second verse does that the first did not

One sentence. Tense, location, who is speaking, what the listener now knows. If the honest answer is "nothing, it is more of the same", you have learned something about a weak song, which is also worth an hour.

A full pass, on a song written for this article

Analysing a commercial record line by line means reprinting somebody else's words, so here is a song written for the demonstration. It is called "Parking Lot Light". A man sits in his car after a night shift, in the lot outside a block of flats, not driving home. The map at the top of this page is its map.

Step 1. Attention went at 2:30. Look at the map and the reason is sitting there: choruses 3 and 4 are identical, back to back, and the last fifty seconds of the song are those six lines twice and then the hook alone. On the record the two choruses are separated by production, a counter melody that enters the second time. On paper they are the same. That gap between the paper and the record is worth more than most advice you will read about arrangement.

Steps 3 and 4. Fifty four lines total. The hook line is line 1 and line 5 of the chorus, so each chorus sings the title twice; four choruses plus the two line outro puts the title in your ear ten times, first at 0:40.

Step 5. The chorus:

Parking lot light, stay on a while
I'm not done sitting here
Her window's third from the left, second floor
And the heater still ticks when it's off
Parking lot light, stay on a while
I'll drive when the light gives out

Six lines, four things you could photograph, named five times between them: the lamp in lines 1 and 5, the window and the floor in line 3, the heater in line 4. Line 6 points back at the lamp without adding anything new. Line 2 carries none of them, and it is the shortest line in the block.

A detail that only shows up once you are circling nouns: his car is never named. Not once in the song. You get the engine, the heater, the windscreen, the bonnet, and the word "drive", and your brain assembles a car out of the parts. The only car named anywhere in the lyric belongs to somebody else, in the second line of verse 1:

Eleven forty, the gate rolls back
Same nine cars, same slant of rain
I keep the engine off to hear the doors
Somebody laughs by the bins at the end of the row

Four lines of the eight, six photographable things: the gate, the cars, the rain, the engine, the doors, the bins.

Step 6. The bridge:

The heater ticks, the windscreen goes grey
Somebody's dog gets walked past the bonnet
Third floor comes on, then the second goes dark
I have been wrong about this for a year

Three lines carry six things you could photograph. The fourth carries none. It is also the only line in the chorus or the bridge that makes a judgement instead of a report, and it is the last line before the hook returns at 2:10. That is the whole trick of the song, and it is completely invisible at listening speed: three lines of objects, then one flat abstract sentence, placed where the music is already lifting.

The third line of that bridge does something else. "The second goes dark" is the window from the chorus, the one that was third from the left on the second floor. She has gone to bed. The chorus has already been sung twice with that window in it and it meant nothing either time. Now it does. Nobody hears that on the first pass.

Step 7. What verse 2 does that verse 1 did not: two of its eight lines move into the past tense while the other six stay present, it goes indoors for the first time, and it carries the song's only reported speech, three words of it.

She said "don't wait up" like a weather report
and the stairwell light went out on its own

So the second chorus feels bigger than the first without a single note changing, because by the time it arrives you have heard her voice and you know the window matters. That is the finding worth stealing. Not the tempo, not the bar count. The fact that a chorus can be upgraded by the verse in front of it.

The thing analysis teaches that nothing else does: pacing

Here is the number that never appears in a songwriting book and never comes up in a workshop, because you only get it with a stopwatch: how long a listener will wait for the chorus.

In the song above it is forty seconds. That is a specific, checkable decision, and once you have it for one song you can have it for fifty. Run the same measurement across ten records in the genre you actually write in and you are holding a distribution nobody handed you.

What you will find, and check this yourself rather than taking my word for it, is that the answer moves in two directions. It moves by genre: a club record often has its hook inside the first fifteen seconds because it has to survive being skipped, while a folk or singer songwriter piece will make you wait a minute and sometimes never gives you a chorus at all. And it moves by decade: the same genre, sampled from records made twenty years apart, puts the hook noticeably earlier now than it did then, for the obvious reason that the skip button got easier to press. Some current pop songs open on the chorus with no verse in front of it, a structure that was a novelty once and is a normal way to start now.

You do not need a study for this and you should not trust anybody who quotes you an average without showing their sample. Ten songs and a stopwatch, forty minutes, and you own the number for your genre. That number then tells you something brutally practical about your own demo: if your genre's hook lands at 0:25 and yours arrives at 1:05, nobody is getting there.

The trap, and the one rule that gets you out

Analysis is extremely satisfying and produces nothing. It feels like work, it makes a tidy artifact, and you can do it forever without writing a word. Plenty of people have a notebook of forty beautiful song maps and no songs. The pleasure of understanding substitutes for the discomfort of making, and it is a good enough substitute to eat a year.

The rule that stops it: one analysis, then one song, using exactly one thing you found. Not the whole map. One thing, named out loud before you start writing.

From the pass above, my one thing is the bridge: keep it concrete for three lines, then land one flat abstract statement as the last line before the final chorus. That is all. I am not copying the forty second hook or the four bar pre-chorus. One borrowing per analysis, converted into a song within the week, or the analysis was entertainment.

If you want the borrowed technique to have somewhere to land, keep a running list of situations you might write about. Songwriting tips and prompts is a decent place to pull from when the technique is ready before the subject is.

Where the tool honestly fits

A generator will not make you a better writer, and an article on this site claiming otherwise would be insulting the person reading it. What it can do is shorten one specific loop.

You have just reverse engineered a shape. The real question is whether that shape holds when the words are not those words, and until now the only way to answer it was to spend a week writing a song into the structure and then find out. With our song generator you can put the same situation through the structure you just measured and read a full lyric in it in under a minute. It writes a title and the complete lyric in [Verse], [Chorus], [Bridge] tags, from a theme of up to 300 characters plus controls for structure, point of view, hook direction, vocal, style and tone. The six structure options are families rather than arbitrary maps, so my intro, verse, pre-chorus, chorus shape goes in as the classic family, not as a custom bar count.

Then read what comes back as a stranger would and ask the pacing question: does the hook still arrive where the map said it should, and does the second verse still earn the second chorus? Half the time it does not, and that is the useful half. A structure that only works with the original song's words was never a structure, it was that song.

The second use is closing the loop with sound. Write your own lyric, keep the shape you measured, and hand it to Suno to perform. Suno makes the audio, we only ever write words, and hearing your own structure sung back is the fastest way to find out that your verse 2 is thirty seconds of nothing. A song costs 20 credits, which is twenty cents, so testing four versions of a structure costs eighty, and the pricing page has the current numbers. You can also come at it from the prompt side on the generators page if what you want is the Style line rather than the lyric.

None of that is a shortcut and I am not going to pretend it is. It is a faster way to be wrong, which is the only real speedup available in learning to write.

FAQ

How long does this actually take?

The first song, about an hour, most of it fighting the scrub bar. By the fourth or fifth song you will be down to twenty five minutes, because you stop re-listening to confirm what you already wrote down.

Do I need to read music or know theory?

No. Everything in the procedure is counting and a clock. Knowing the tempo lets you convert seconds into bars, which makes your notes readable by other musicians, but the analysis works fine in raw seconds.

Can I do this from a lyrics website instead of listening?

No, and this is the most common way people skip the hard part. A lyrics page gives you the words and destroys every duration, which is where the craft is. It will not tell you that the bridge has twice the room per line or that the hook waits forty seconds. Words without time are not a song.

What if the song has no chorus?

Better. Analyse it anyway. Songs without a chorus have to solve the same problem, which is giving the listener a reason to still be there at 1:30, and they solve it with a refrain line, a rhythmic shift, or a story turn at a fixed place. Time those. The step that changes is step 4: instead of the title, track the recurring line or the moment the arrangement changes.

Should I analyse songs I dislike?

Yes, and hit records you actively resent are the best material. Your taste stops you seeing the mechanism in songs you love, because you keep sliding back into enjoying them. A song you find irritating that eight million people played all summer is doing something, and you will spot it faster precisely because you are not moved by it.

How many songs before it shows up in my writing?

If you follow the one analysis, one song rule, it shows up immediately, because you are deliberately using a found technique in the next thing you write. If you only analyse, it never shows up. The number of songs is not the variable. The conversion rate from analysis to draft is.

Is it a problem to write someone's lyrics out by hand?

Copying a lyric into a private notebook to count its lines is what songwriters have always done, and it stays in your notebook. Publishing another writer's words, or putting a line by line breakdown of their song online, is a different question with a different answer, and not one to settle by guessing.