Music Theory for Songwriting: The 5% You Actually Need

Play C, then G, then A minor, then F. Four shapes, four bars, and you are sitting on the harmonic engine behind a very large slice of popular music. Now play the same four in a different order: A minor, F, C, G. Your hands did almost nothing. The room changed completely. The first loop sounds like a lit stage. The second sounds like someone standing outside it in the cold, looking in.

Everything a songwriter actually needs from music theory lives in that gap and in about four other places. Call it five percent of the subject. The remaining ninety-five percent, Neapolitan sixths, tritone substitution, the chart of which mode belongs over which chord, is genuinely interesting, and it is also the most effective procrastination technology ever built for people who want to write songs. You can spend two years in there and come out with a confident grasp of modal interchange and no finished songs.

So here is the five percent, taught by ear. Names arrive at the end of each idea as labels, not at the front as entry requirements. Nothing below needs you to read music. If you can find four chords on a guitar or a keyboard, or even hum them, you can test every claim here in about ten minutes.

One boundary first. How an arrangement sits, which frequencies fight, why a mix turns to mud, is a different subject and has its own piece: music theory for producers. This article is only about what the listener feels, and about the handful of decisions that change it.

The four chords, and why the order does more work than the chords

In the key of C, the four chords that carry most of what you hear on the radio are C, F, G and A minor. In numbers, they are the 1, the 4, the 5 and the 6. The numbers matter because they travel: in the key of G they become G, C, D and E minor, and every claim below still holds.

Why those four and not four others? Because of what they share with home.

• C is home. C major is built from C, E and G.
• A minor is C's shadow. It is built from A, C and E, so it shares two of its three notes with home. Play C then A minor and almost nothing moves, yet the light drops. That is the cheapest emotional move in music.
• F is the step outward. F, A and C. It shares the note C with home, so it feels like leaving the house without leaving the street. It creates space, not tension.
• G is the one with the pull. G, B and D. That B sits one half step below C, the closest possible approach. A listener's ear treats a note a half step below home as a note that is falling toward home. Add the seventh, making G7 out of G, B, D and F, and now you have both B and F at once: B leans up to C, F leans down to E, and C and E are two thirds of the home chord. That is why the 5 chord sounds like a sentence about to end.

Now the part most articles skip. Those four chords, in a loop, produce a different song depending on where the loop starts and, above all, on which chord it ends on. The last chord in the bar before the loop repeats decides what the loop wants to do next.

Take the same four and rotate them:

Loop A:  | C  | G  | Am | F  |     starts home, ends on the soft one
Loop B:  | Am | F  | C  | G  |     starts in the shadow, ends on the pull
Loop C:  | C  | Am | F  | G  |     home, shadow, step out, pull

Loop A ends on F. F has no urgency in it, so the loop can spin forever without demanding anything. That is why it holds up under a chorus that repeats four times: it never insists on stopping. Loop B starts on A minor and the C, when it arrives in bar three, lands like relief rather than like home. The same four chords now sound like a song about wanting something. Loop C ends on G, and G will not let you stop. Play it and try to end the song on that G. Your ear stays on its feet waiting.

I am not going to list famous songs under each one. Half of those lists on the internet are wrong, and the useful version of this knowledge is not trivia, it is the ten minutes you spend rotating your own loop and picking the rotation that matches what the lyric is doing.

The practical rule: write your loop, then play all four rotations before you commit. Most people never do this. It costs nothing and it is the single highest-yield theory move available to a songwriter.

Swap one chord and watch the temperature drop

Keep Loop A and change exactly one thing. Not the order this time, and not even a whole chord. One note inside one chord.

Before:  | C  | G  | Am | F  |
After:   | C  | G  | Am | Fm |

F major is F, A and C. F minor is F, A flat and C. One note moved down by a half step. Everything else in the loop is untouched.

What you hear is the loop going cold in its last bar. And the return to C at the top is suddenly stronger than it was, because that A flat sits one half step above G, and G is the top note of the home chord. The ear hears a small fall into home.

The label is a borrowed chord: the 4 chord borrowed from the parallel minor key, written iv. You will hear it at the end of choruses in every genre that wants a bruise in it. You do not need the label to use it. You need to know that changing one note inside one of your four chords is an available move, and that it usually does more than adding a fifth chord would.

While you are here, try one more swap on the same loop: replace A minor with A major. A major is A, C sharp and E. The C sharp is not in the key of C at all, and the ear reads it as a door opening onto somewhere else, usually toward D minor. Use it once in a song. Use it twice and it stops meaning anything.

The relative minor: the lights go out, not one note changes

Here is the trick that costs the least and buys the most.

The key of C major uses seven notes: C, D, E, F, G, A, B. The key of A minor, in its natural form, uses exactly the same seven notes. Same notes, different centre of gravity. A sits a minor third below C, three half steps down, and if you make your phrases begin and end around A instead of around C, the identical material turns minor.

This is why Loop B above sounds sad without a single new chord. Nothing changed except which chord the ear decided was home. You did not modulate. You did not borrow anything. You rotated a loop.

Use it structurally: verse built around A minor, chorus built around C. The chords can literally be the same four in both sections. Write the verse melody so it keeps landing on A and E, write the chorus melody so it keeps landing on C, E and G, and the chorus will sound like a light turning on. The chords underneath will not have told anyone why.

If you have no idea which key a song of yours is already in, that is a solvable ten-minute problem and there is a method for it: how to find the key of a song.

Why the pre-chorus exists, in chord terms

Songwriters usually learn the pre-chorus as a lyrical idea: the section that raises the stakes before the chorus. That is true and it is not the mechanism.

The mechanism is this. Your verse loop contains home. C keeps coming back every four bars, so the listener is never hungry. A chorus that arrives on C after a verse that has been playing C all along is not an arrival, it is more of the same. The pre-chorus fixes that by taking home away.

Verse       | C  | G  | Am | F  |                      (looped, C present every four bars)
Pre-chorus  | Dm | Dm | F  | F  | Dm | Dm | G  | G  |   (the C chord does not appear at all)
Chorus      | C  | G  | Am | F  |                      (C lands on the downbeat)

D minor is D, F and A, all inside the key. For eight bars the chord of C is simply absent, and the section ends on G, the chord whose whole personality is wanting to go to C. When the chorus arrives, the C is not just the next chord. It is the answer to a question the pre-chorus spent eight bars asking.

Two failure cases worth knowing, because both are common in first drafts:

1. The pre-chorus that still contains the tonic. If C appears anywhere in those eight bars, the section stops functioning as a pre-chorus and becomes a second verse with different words. Delete the C.
2. The pre-chorus that ends on the tonic. Ending it on C resolves the tension one bar early, and the chorus opens on a listener who has already relaxed. End on G, or on F going to G, or hold G for a whole bar with the band dropping out.

This is also why the pre-chorus melody almost always climbs. The harmony is withholding, so the voice pushes. If you want the lyric side of the same problem, the chorus itself has its own set of rules: how to write a chorus.

The melody note that is not in the chord, which is where the feeling lives

This is the section most songwriters have never had explained to them, and it is worth more than the previous three combined.

Take Loop A and sing one long note per bar, choosing a note that is inside the chord underneath:

Chord:   C     G     Am    F
Sing:    E     D     C     A

Every one of those four notes belongs to the chord under it. E is in C. D is in G. C is in A minor. A is in F. The result is clean, tuneful and completely inert. It sounds like a jingle for a bank. There is nothing wrong with it and nothing in it.

Now move each note by one step of the scale:

Chord:   C     G     Am    F
Sing:    D     C     B     G

None of those four notes is in the chord underneath it. Check them: C major is C, E, G, so D is not there. G major is G, B, D, so C is not there. A minor is A, C, E, so B is not there. F major is F, A, C, so G is not there. All four are still inside the key of C, so nothing sounds wrong. What it sounds like is somebody feeling something.

Listen to what each one does, because they are not the same:

• D over C sits a whole step above the root. It hangs. It does not hurt and it does not resolve, it just refuses to settle. Pop choruses live on this note.
• C over G sits a half step above the B in the chord, and that half step is a rub. The ear wants that C to fall to B. Let it fall and you have written a suspension. Hold it through the whole bar and refuse to resolve, and the line sounds stubborn.
• B over A minor is a whole step above the root of a minor chord. It floats. It is the note that makes a sad chord sound thoughtful rather than defeated.
• G over F is a whole step above the root again, but over a chord that was already open, so it widens rather than tenses.

A practical version of this that you can use tonight: stop landing your melody on the root of the chord. The bass is already playing the root. When the voice doubles it, the chord and the melody collapse into one object. Land on the third instead, or on a note next door to a chord tone, and the same chord suddenly has two things happening in it.

And the dangerous version, which is also the useful one: hold a note through a chord change. Sing an E and let it sit while the chords move C, then A minor, then F. E is inside the first two chords and outside the third, so the note never breaks, but its meaning shifts under your feet three times: it is the third of C, then the fifth of A minor, then a stranger over F, a major seventh that turns the bar into Fmaj7 and leaves the line hanging in the air. One sung note, three different feelings, no effort. Find the note your own chords share, and the one chord it does not belong to, and you will never write a boring sustained line again.

The one modulation trick worth knowing, and the one to skip

Everybody knows the key change that lifts the last chorus up a whole step. In our key that is C, G, A minor, F becoming D, A, B minor, G. It works, in the narrow sense that the listener notices. It is also a blunt instrument, it puts every note of the melody two half steps higher at the exact moment the singer is most tired, and it announces that the song had nothing else left. Learn it, keep it in the drawer, use it roughly once a decade.

The one actually worth knowing is the relative shift from earlier, used as a section change instead of a rotation. Verse centred on A minor, chorus centred on C, back to A minor for verse two. No new notes, no pivot chord, no awkward gear change, and the listener hears the song open and close like a door. It costs nothing, so it survives arrangement, transposition, and a singer who wants the whole thing down a tone.

And the honest footnote: the borrowed F minor from earlier is not a modulation at all, it is one chord wearing the clothes of the parallel minor for a bar. It buys you most of what a modulation buys and asks for none of the commitment. If you only take one harmonic move from this article into your next song, take that one.

What theory our generator cannot do for you, said plainly

Our tools write words. The AI song generator returns a title and a full lyric with section tags in it, and nothing in it is a chord, a key or a melody. It cannot pick your four chords. It cannot tell you to end the loop on G. It will not hear that your chorus note is doubling the bass.

That is not a limitation we are planning to route around. Harmony and melody are the part of songwriting that has to sit in your own ear, and any tool claiming to hand you those is selling you somebody else's decisions.

What it genuinely contributes to the work in this article is shape. Three specific things:

1. Section structure that leaves room for the harmonic turn. The generator's structure setting includes options for a classic build, a version with no bridge, a double chorus, a story shape, a loop and a ballad. Pick the one whose section map matches the harmony you already have. If your pre-chorus needs eight bars of held tension, do not ask for the loop structure.
2. A hook direction that matches the chord move. Among the hook directions is contrast, and contrast is the lyrical twin of the relative-minor flip: the verse says one thing in the dark, the chorus turns the same thing over in the light. Choosing them to agree is worth more than either alone.
3. The tags. The output comes back in [Verse], [Chorus] and [Bridge] labels, which is exactly what Suno reads to decide where a section begins. If you want the harmonic turn in a specific place, you put a [Pre-Chorus] tag there yourself before you paste, and Suno treats it as a section boundary. There is a whole grammar to those labels: Suno lyric metatags that work.

Cost, since it comes up: a song costs 20 credits, and credits run 100 to the dollar, so twenty cents per generated lyric. Current plans are on the pricing page. Twenty cents is roughly the right price for a drafting partner and roughly the wrong price for a teacher, which is the correct way around.

The workflow that actually uses both halves: work out your loop and your rotation at the instrument first, decide where the harmony turns, then use the lyric generator to throw twenty candidate versions of the line that has to land on that turn. Read them, take the one word you would not have reached, and write the rest yourself.

What to learn next, if five percent runs out

It will run out, eventually, and the order matters more than the syllabus.

First, learn the numbers rather than more chords. Train yourself to think 1, 4, 5, 6 instead of C, F, G, A minor. The system used in session work for this is called the Nashville Number System, and it exists because a room full of players needs to change key in four seconds when a singer says the song is too high. Learning it takes an afternoon and it makes every chord trick you know portable to every key.

Second, ear training, in tiny daily amounts. Not theory, recognition. Being able to hear that a chord went to the 6 rather than the 4 is a different muscle from knowing what the 6 is. Free interval and chord drills exist on the web, musictheory.net being the long-standing one, and fifteen minutes a day for a month will do more than a book will.

Third, transcribe. Pick three songs you wish you had written and work out the chords by ear, badly at first. Write out where each section starts and which chord each one ends on. That single habit teaches more functional harmony than any course, because you are learning it attached to music you already care about. There is a method for pulling a song apart this way in our piece on how to analyse a song.

Fourth, and only if a genre demands it, modes and extended harmony. If you are writing jazz, gospel, bossa, film cues or anything with a seventh chord on every beat, you will need more than the five percent and you will know exactly when. Until then, learning Lydian is a hobby, not a tool.

What to skip, actively: notation, unless you need to hand parts to players who read; counterpoint, unless you are writing for voices in parts; and any course that begins with the circle of fifths before it has played you a single progression. Books are a real route too, but pick them by what they teach rather than by their reputation: any book that gets you to a chord under your hands in the first chapter beats one that spends forty pages on the harmonic series.

FAQ

Do I need to read music to write songs?

No, and the number of successful songwriters who cannot read a score is large enough that the question answers itself. Reading is a skill for communicating with players who read. Writing is a skill for deciding what happens next. They overlap less than music schools imply. Learn notation when someone is waiting for a chart.

Is it true that four chords are all you need?

Four chords are all you need to get a listener from the start of a song to the end of it. They are not all you need to be interesting for three minutes, which is what the melody, the lyric and the arrangement are for. The people who mock four-chord songs are usually pointing at the chords when the actual problem with the song is that the melody sits on chord tones and the lyric has no specifics in it.

Will learning theory make my songs sound formulaic?

The risk is real and it is not the theory, it is the sequencing. If you learn a device and then hunt for a song to put it in, you write exercises. If you write a song, notice it is limp in bar twelve, and reach for a device to fix bar twelve, you write songs. Keep theory downstream of the idea and it stays a toolbox.

What key should I write in?

The one your voice sits well in, and you find it by singing, not by choosing. Write wherever the chords fall under your fingers, then move the whole song until the chorus sits just below the top of your comfortable range, so the chorus feels like effort but not like strain. Guitar players default to G, D, E minor, A minor and C for a reason that is purely physical, and there is nothing wrong with that.

How do I know if my chord progression is any good?

Play it, then sing complete nonsense over it for two minutes without stopping. If a melody starts appearing on its own, the progression is doing its job. If you keep drifting into a melody you already know, the progression is too familiar and needs one chord swapped or one rotation turned. If nothing comes at all, it is usually not the chords, it is the tempo.

Can I ask an AI for a chord progression?

You can, and there are tools for it, including chord progression generators. What you get back is a statistically average loop, which is fine as a starting block and useless as a destination, because average is the one thing a song cannot be. Our own generator does not do chords at all and will not pretend to. Use a chord tool to get unstuck at midnight, then change one chord in it yourself so at least one decision in the song is yours.

My chorus feels flat even with the right chords. What is wrong?

Nine times out of ten, one of three things. The chorus melody is in the same octave as the verse melody, so nothing lifts. The chorus melody lands on the root of every chord, so it doubles the bass and has no friction. Or the section before the chorus never removed the home chord, so there is nothing for the chorus to resolve. Check them in that order, they take about five minutes each.