How to write a worship song a congregation can actually sing

There are hundreds of articles on how to write a worship song. Almost none of them contains a number.

They will tell you to write from a place of honesty, to keep it simple, to serve the church. All true, all unfalsifiable, and none of it tells you whether the top note of your chorus is going to empty the room. Meanwhile the people who have counted things, the ones who measured what untrained adults can actually sing, and what keys churches actually download, and which features make a melody survive being passed from mouth to mouth, publish in worship-leading journals and academic music education, where songwriters never read them.

So this is that research, translated into writing advice. Every figure has a source. Where the honest answer is that nobody has measured something, I say so, because a page like this is only worth reading if you can tell the evidence from the opinion.

One thing before we start, since it bears on why I am the one writing this. I have led my own songs in church, and I write them to be singable there. Behold The Lamb comes back to me in performance royalties reported from congregational use, and it is my largest earner on that measure, which is the only evidence on this question I would put any weight on, because it records rooms actually singing rather than anybody's opinion of whether they could.

I also aim my songs at the six days between Sundays: headphones on a commute, the moment somebody needs reminding how much the Father loves them when nobody is watching. A handful of people have told me they enjoy them more in the car than in a service. That is a handful of people, not a finding, and it would take a great many more before it meant anything. There is an article coming about writing for those six days. This one is about the room.

Speaking to one another with psalms, hymns, and songs from the Spirit. Sing and make music from your heart to the Lord.

Ephesians 5:19, NIVUK

Speaking to one another. Whatever else a congregational song is, it is a thing people do together, out loud, without rehearsal. That constraint is not a limitation on the art. It is the art.

---

The short answer

For the reader who wants the rules before the reasoning:

Everything below is the evidence, the songs that demonstrate it, and then eleven rules at the end you can apply to a song you are working on today.

---

What range can a congregation actually sing?

Far less than most songwriters assume. The only proper measurement found that a range comfortable for virtually everybody is just a major sixth, and the practical consensus among worship-music authorities is A3 to D5, an octave and a fourth with the top note on D.

Start with that measurement, because it is more sobering than any rule of thumb.

In 1979 Terry Lee Kuhn and three colleagues measured the vocal ranges of 597 undergraduate non-music majors at the University of Maryland, the University of Oregon and Shepherd College, and published the results in the Journal of Research in Music Education. Untrained adults, in other words, the population in your pews, only younger and healthier. Paul Ryan later set out their percentile bands in Reformed Worship:

Read the top line again. The range that suits virtually everybody is a sixth. Ryan's own reaction: "the range that will comfortably suit nearly everyone is incredibly narrow, just a sixth," and "this range for 75 to 85 percent of young adults is much lower than I had believed."

Nobody writes songs inside a sixth. But that table explains why the practical guidance lands where it does, and it explains something else: the people who cannot reach your top note are not being difficult. They are the majority.

The rule everybody independently arrives at

What is striking about the craft literature is how closely people who have never cited each other agree. Eight of them, in their own words:

A3 to D5. That is the number to write on the wall above your desk. An octave and a fourth, with the top note on D.

Be precise about what that consensus is, though, because it is a cluster rather than a chorus. Four of those eight, Lamm, Kauflin, Chamberlain and Woodard, name A to D exactly. The others differ at the edges: Ryan and Cole put the floor at B♭ rather than A, Willmington argues for an octave from C to C, and Enfiedjian splits the ceiling by voice. Ryan, working from his own data, actually pushes the ceiling down, advising writers to keep the melody below C5. So the honest summary is that eight people who mostly have not read each other all landed inside the same narrow neighbourhood, and the middle of that neighbourhood is A3 to D5.

Range versus tessitura, which is the distinction that actually matters

Two songs can have identical spans and behave completely differently in a room, and the reason is tessitura: where the melody spends its time, as opposed to the extremes it touches once.

A melody that touches D5 for a single quaver at the top of a bridge and otherwise lives around middle C is a comfortable song. A melody that never goes above D5 but sits at C5 for twenty bars is a punishing one, because a congregation is not sprinting; it is running the whole distance. Chamberlain's instruction to put the strongest notes "across that E to C mid-way fulcrum where both men and women have more of a chance" is a tessitura instruction, not a range one.

And there is a squeeze that makes this worse, which every practitioner mentions and, I have to be honest, nobody has measured. Men and women sing the same melody an octave apart. So a melody comfortable for the women in the room may be sitting at the top of the men's range, or the reverse. The usable window is narrower than either group's individual range, and it is the overlap you are writing for. That is craft consensus, not data. I found no study of it.

---

Why do popular worship songs break the range rule?

Because a wide span is the norm rather than the exception in the current repertoire. In the top 50 songs on PraiseCharts, the most common range is an octave and a fourth, and 77% exceed an octave. So span alone cannot be what separates the songs that work from the songs that do not.

Daniel Thornton, Head of Ministry and Worship at Alphacrucis University College, analysed the top 50 songs on PraiseCharts by download and interaction across the 365 days to March 2025. What he found:

• The most common range is a perfect 11th, an octave and a fourth
• 51% of the songs span a perfect 11th or larger
• 77% exceed an octave
• The narrowest were a major sixth (Trust In God, Graves Into Gardens); the widest a perfect 15th, two full octaves (You've Already Won)

Three of the fifty had no range data listed, so those proportions rest on forty-seven songs rather than fifty.

His conclusion is blunter than anything in the craft literature. On the songs themselves: "There is no way that a 'singable key' exists for your whole congregation on those songs." And, closing the piece: "So, are congregationally 'singable keys' a thing? I'm yet to be convinced."

This is the most useful finding in the whole subject, and it reframes everything. An octave and a fourth is the median, not the transgression. Which means span alone cannot be what separates the songs that work from the songs that do not, because on span, the successful repertoire and the failures look the same.

So what is the difference? Here the evidence gets specific and genuinely useful.

The cleanest case study in modern worship: So Will I

So Will I (100 Billion X) was written by Joel Houston, Michael Fatkin and Benjamin Hastings, and appeared on Hillsong UNITED's Wonder in 2017. At the 2018 GMA Dove Awards it was nominated in two categories, Worship Recorded Song of the Year and Worship Song of the Year, and won neither; those went to Resurrection Power and Do It Again. Thornton, analysing CCLI Top 25 appearances between 2014 and 2023, still uses it as his example that industry recognition does not predict adoption: "if causation was clear, 'So Will I (100 Billion X)' should have reached the Top 25 list, which it did not."

One correction while we are here, since this article's whole premise is that you should be able to tell evidence from assertion. Thornton's paper describes the song as award-winning. It was nominated twice and won nothing. The point survives the correction, because a song at the top of the industry's own shortlist still did not reach the chart of what churches sing.

Now look at the numbers. Recorded in A. Melodic range F♯4 to A5, a minor tenth.

A minor tenth is narrower than Oceans (A3 to D5, a perfect 11th), and narrower than Cornerstone (C4 to G5), Goodness of God (A♭3 to E♭5) and Reckless Love (D♭4 to A♭5), which all span a perfect twelfth. On span, So Will I is one of the tamer songs on the modern list.

But look at where it sits. The bottom of the melody is F♯, above every floor any of them names, so the lower voices in the room have nowhere comfortable to be. And the top is A5, a perfect fifth above the D that four of them name as the ceiling, and higher still above Willmington's C and Ryan's C. The whole melody is parked in a register that belongs to a lead vocalist.

A wide melody in the right place beats a narrow melody in the wrong place, every time. So Will I is the narrow one, and it is the one that did not reach the chart. If you take a single diagnostic away from this article, take that: when somebody tells you your song is hard to sing, the first question is not "how wide is it," it is "where does it live."

And the song everybody blames for the wrong reason

Oceans (Where Feet May Fail) is the standard cautionary tale, and I came into this research believing the standard explanation. The research corrected me.

Written by Matt Crocker, Joel Houston and Salomon Ligthelm; released 23 August 2013; recorded in D major at 64 bpm. Melodic range A3 to D5.

That is a perfect 11th, Thornton's exact median. And its top note is D, precisely the ceiling Lamm, Kauflin, Chamberlain and Woodard all name. On paper, in its original key, Oceans obeys the rule. Jonathan Aigner, one of its sharpest critics, complains in the same sentence that "the melody is almost directionless, the rhythm syncopated, the range low and cumbersome." Low, note, not high.

So the difficulty is real but it is not the span. The sourced explanations are: the bridge, which climbs and where the notated melody may not be what the recording actually sings; the length and the dynamic build, which need an arrangement most rooms cannot deliver; and the fact that it works as a record before it works as a congregational song.

There is a postscript. In October 2016 Brian Houston, then Hillsong's global senior pastor, told the Catalyst conference in Atlanta: "If you come all the way to Australia and you hope to hear 'Shout to the Lord,' your chances are slight. We don't even sing 'Oceans' much anymore."

---

What key should you write in?

The key a church will actually play, which in practice means C, G or D. Those three take 68% of chart downloads. D♭ is the single most common key for a modern worship song to be recorded in, and it takes none of them at all.

Thornton's same PraiseCharts study compared the key each song was recorded in against the key churches actually downloaded it in. The two lists barely overlap.

And the single fact that should change how you write: not one song was downloaded in D♭. The most common recorded key in that sample has a zero per cent share of what churches take away to play. A bare majority, 28 of the 50, were downloaded in the key they were recorded in; nearly half were transposed before a church would use them.

D♭ and B are artist keys. They are where a particular singer's voice sounds best on a record. C, G and D are congregation keys, and they are also guitar keys, which is not a coincidence.

Three honest caveats, because this figure gets over-quoted. Both halves measure a top-50 sample on PraiseCharts, the recorded-key figures as much as the download ones, and D♭'s 16% is eight songs. The 68% is Thornton's own printed figure; the 38% beside it is arithmetic I did from his three recorded-key percentages, and because third place is a two-way tie it is a slightly awkward comparison. And downloads are not what was played on Sunday; they are a proxy for the room, not the room.

On the widely repeated belief that recorded keys have drifted upward over the decades: I looked for the study and it does not exist. Nobody has measured melodic peak pitch over time in this repertoire. What has been measured is the gap, not the drift.

For your own writing the implication is simple. Write in the key the congregation will sing it in, then let the record be transposed for whoever sings it. Most writers do this backwards: they find the key that flatters their own voice, and the church inherits it.

---

What makes a melody easy to pick up?

Stepwise motion, mostly pentatonic notes, short phrases, heavy repetition, and low surprise. Two independent bodies of research land on that same list, one studying three centuries of congregational song and the other studying which folk melodies survive being passed between singers.

The worship-specific study

Daniel Read's 2017 master's thesis at the University of Kentucky, Why We Sing Along: Measurable Traits of Successful Congregational Songs, coded 206 songs across 300 years, being 48 hymns, 60 gospel songs, 42 praise choruses and 56 contemporary worship songs, for harmony, melodic range and intervals, rhythmic complexity, repetition, borrowed pitches and tempo.

He found five traits shared by the songs that succeed. In his words, they "consist largely of pentatonic melody notes; have a moderately fast tempo range; contain a straightforward rhythm; have a limited tessitura; and contain a significant amount of repetition."

1. Largely pentatonic melodies, scale degrees 1, 2, 3, 5 and 6. The notes you can sing without thinking
2. Moderately fast tempo. Read pins this to around 100 bpm by way of Paul Fraisse's tapping studies, in which most people left to tap freely settle near a 600 millisecond beat. Worth knowing that this is spontaneous tapping tempo rather than preferred tempo, which the same literature puts nearer 500 milliseconds, or 120 bpm. The 100 bpm figure is Read's reading of Fraisse, not a separate finding
3. Straightforward rhythm, limited syncopation
4. Limited tessitura, melodies gravitating towards a median pitch rather than roaming
5. Significant repetition

He also identifies a pattern he calls step declination: "large melodic intervals are more likely to ascend, and small melodic intervals are more likely to descend," so a leap tends to be followed by stepwise motion the other way.

The study from outside the field, which is stronger evidence

Berit Janssen, John Ashley Burgoyne and Henkjan Honing analysed 9,639 melodic phrases from 147 folk-tune families in the Meertens collections, published in Frontiers in Psychology in 2017. They asked which features predict whether a phrase survives unchanged as a tune is passed orally from singer to singer across generations.

That is about as close as research gets to measuring what a crowd retains without sheet music. The predictors, strongest first:

Be careful with these: the fixed effects account for only 5% of the variance, and the authors say plainly that most of it is unexplained. They are real, directionally clear, and small.

But notice the convergence. Stepwise motion, gap-fill after leaps, repetition, low surprise. A study of Dutch folk-song transmission and a study of three centuries of congregational song, using entirely different methods on entirely different material, arrive at the same four features. When that happens, you are probably looking at something true.

The other thing worth having: phrase length is the biggest lever you have, and it is the one songwriters think about least. A congregation running out of breath stops singing, and it does not start again mid-phrase.

Is syncopation the enemy?

Read measured this properly. His coding rule was that a song with more than two instances of syncopation counted as syncopated. By era:

• Hymns (1700 to 1825): syncopation essentially absent
• Gospel songs (1825 to 1950): lilt and swing appear
• Praise choruses (1950 to 1996): onset of syncopation
• Contemporary worship (1996 to 2016): universal, with Read writing that in this era "every song contains syncopation to some degree"

From absent to universal in three centuries. Read treats it as working against participation, and warns specifically about "excessive and hurried syncopation."

But the best answer to this comes from Paul Ryan, who reframes it: syncopation is fine "as long as it is predictable." That is the more useful rule, because it tells you what to do rather than what to avoid. A syncopated figure a congregation has heard three times is not a difficulty; it is a hook. A syncopated figure that changes every verse is a trap. Repeat your rhythms as deliberately as you repeat your words.

The problem that is not in the song at all

There is one more mechanism, and it may matter more than anything above.

Shannan Baker at Baylor analysed years of YouTube performances by the primary writers on CCLI's US Top 100 and documented what she calls a typology of ad-libbing. What follows from her work, and it is obvious once said: the recorded vocal and the congregational melody are different artefacts. The leader's runs, held notes and countermelodies are not the song; they are a performance of it. And "if the lyrics are on the screen, the people will sing them, which will, by nature, make the words no longer an ad-lib."

Worship leaders describe the consequence more bluntly. Worship Team Coach: "When the worship leader jumps to that high countermelody on the big part of the song, what do you […] do? You stop singing." They call it "vocally abandoning the congregation." Kenny Lamm lists the same thing among his reasons people are not singing: "when we stray from the melody to ad-lib, people try to follow us and end up frustrated and quit singing."

For a writer this cuts two ways. Your demo will teach people the wrong song if you perform it rather than sing it. And if the only version anyone hears is the one with the runs, you have written a record that happens to have a congregation standing near it.

---

What the lyrics of successful worship songs actually look like

People argue about worship lyrics constantly and almost always without counting anything. Several people have counted.

The best single study

Ian Hussey of Malyon Theological College compared 27 popular hymns against 27 contemporary songs from the CCLI Top 100, published in Ecclesial Practices in 2019:

So the measured picture of a modern worship lyric: about 140 words, a reading age of roughly 12, and noticeably more "I" than "we." Hussey's own conclusion is more generous than the usual commentary. The songs are "not significantly different in content" from the hymns they replaced, but carry less meaning, and he discerns "a greater sense of individual personal intimacy expressed in the contemporary songs compared to the hymns."

His word counts are worth a glance too. "Worthy" appears 0 times in the 27 hymns and 13 times in the 27 contemporary songs. "Name" goes from 12 to 55.

That last number is the one I have spent the most time on. Great Titles was written on the train from Paris to London, working out what I should actually call God, and it ends up at Daddy. Fifty-five uses of "name" across twenty-seven songs suggests I am not the only person circling the question.

The claim that gets made most, and the count that refutes it

"Modern worship songs are all about me." Thornton actually counted it, across the 25 most-used songs, as a ratio of references to the singer against references to God. In 20 of the 25, God is referenced more often than the worshipper. Four run the other way, being Amazing Grace (My Chains Are Gone), Desert Song, Hosanna and Oceans, and one is level.

There is a real pronoun shift, and it shows in Hussey's numbers. But the strong version of the complaint does not survive being counted, and Thornton is not defending the industry.

The largest lyric count, which complicates the picture

Christine Longhurst analysed 509 songs from Songs of Fellowship 5 in Direction in 2015. On voice she found: individual 34%, collective 44%, combined 19%. On primary focus: God 43.2%, the individual worshipper 29.7%, the corporate body 20.6%. And 57% contained content about God's story, while 43% did not.

Her 44% collective runs against Hussey's first-person-singular finding, and the honest explanation is that they measured different populations, a British songbook against the CCLI chart, and different units, song-level voice against word-level frequency. A songbook is not a chart. Both results are real.

---

Structure, chords and length

The shape of these songs has been counted too.

From Longhurst's 509 songs: 83.5% use a traditional verse and chorus format, 16% are cyclical, and half have exactly two verses.

From Thornton's 25 most-used songs:

• All 25 contain I, IV and V; 19 include vi; 13 include ii
• 16 of 25 use four chords or fewer
• 21 of 25 are in 4/4
• And the surprise: on the final chord, 10 end on I, 8 on IV and 5 on V. Thirteen of the most-sung worship songs in the world do not resolve to the tonic

Two notes on that last line. Thornton's three figures account for 23 of the 25 songs and his piece does not say what the other two do, so the gap is in the source rather than in the arithmetic. And his description of the keys as "white-note major keys, commonly G, D and B" is his own phrase, quoted as he wrote it, although B major carries five sharps and is not a white-note key by any usual definition.

On length, Thornton's doctoral corpus gives an average recorded duration of 6 minutes 18 seconds, against an estimated five minutes in actual church use. Which tells you the recorded version is not the congregational version, and never was.

One gap worth naming: nobody has counted how many current top songs have a bridge, and nobody has tested whether bridges help or hurt participation. Read establishes that they arrive with the contemporary era. Beyond that it is all assertion, including the Oceans diagnosis above, which is craft observation rather than measurement.

---

Four songs, and what they teach

How Great Is Our God, the tightest of the durable songs

Chris Tomlin, Jesse Reeves and Ed Cash, 2004, recorded in D♭. The notated melody spans E4 to F5 in the print key of C, a minor ninth.

That is the tightest range among the durable modern songs, and it has been in continuous global use for over twenty years. Tomlin's account of writing it is not a story about craft at all: he was sitting on his sofa in Austin, Texas, and "I just started singing, 'How great is our God?' I don't know where the melody came from." Then the confession every songwriter will recognise, that he had the chorus and the verses would not come: "I could not ever find the verses for this song."

The bridge arrived from Ed Cash at their first writing session. Cash had his back turned, reached behind him for a guitar off the wall and said "Man, when you sing that, this is what I think about." Cash's verdict on the song at the time: "It's pretty good. I think it's pretty good."

10,000 Reasons, four chords and the truth

Matt Redman and Jonas Myrin, copyright 2011, recorded in G, though print editions circulate in C, so you will see both. Range a major tenth in either key. Chorus first, verses second, so the chorus sets the posture and the verses supply the reasons.

Redman: "this song, 10,000 Reasons, was one of the quickest songs I've ever been part of writing." It was half past one in the morning in a village chapel and he was trying to go home when Myrin said "just listen to this one little chorus melody." And then the line to pin above your desk: "it doesn't have a pre-chorus, what I say is it's got four chords and the truth."

On writing for a room, he is unusually explicit: "when I write I'm trying to help the saints sing to Him, but I have one eye on it being an invitational moment for someone who's not a believer."

Worth knowing, though: even this song, the narrow-range four-chord exemplar, is the one worship leaders most often say they have to transpose. Musicademy reckons "most congregations still need to bring the key down a minor third to keep it in range."

In Christ Alone, the one that should not have worked

Keith Getty and Stuart Townend, begun 2000 and completed in January 2001. Four verses, no chorus, in long metre double. Range about an octave and a fourth.

Long metre double is a difficult form. The hymnologist Austin Lovelace wrote that "it is virtually impossible to get a really good congregational LMD tune." Getty got one. James Cheesman notes the melody "exhibits the golden mean, rising to its peak in the third phrase before settling back down in the final phrase," which is step declination and the earworm contour and the golden mean all describing the same shape from three directions.

Getty was explicit about the intent: "the idea was to write a hymn that declared the whole life of Christ in a song… a simple, creedal hymn." Townend on receiving the tune: "the first melody on it really hit me. It was tuneful and memorable, and yet had gravitas and real emotion."

And then the line that should be read by everybody who believes worship songs arrive from heaven fully formed. Getty on their process: "for us, it's usually a long, drawn-out, painful, anarchistic, tense, argumentative, difficult, refining process." The Hymnology Archive notes he was being half-facetious when he said it, which I think makes it more useful rather than less. It is the joke you only make about something you have actually done.

Note also that this is the wordiest song in Thornton's 25, at 224 words against an average of 123. Every rule has an exception and this song is several of them at once.

Way Maker, written in Lagos, sung everywhere

Osinachi Kalu Okoro Egbu, who records as Sinach, released Way Maker as a single on 30 December 2015. Recorded in B, marked slow at 66 bpm, spanning D♯3 to C♯5. The structure is the thing: the verses sit low, and the second chorus lifts an octave above the first, which is where the room finally opens up.

In June 2020 it reached number one on CCLI's Top 100, the chart of what the world's churches actually sing. A song written by a Nigerian woman at Christ Embassy in Lagos, sung in churches that had never heard of her.

She wrote it in a season of upheaval, feeling "like a fish out of the water because I was going to a kind of area where I haven't been before," and God met her in Abraham's story: "He said, 'Oh see, I'm your Way Maker.'"

And the sentence that is really the whole subject of this article: "the joy of a writer is that when you write a song, the whole world will sing it, because the song is really not about you."

I will not pretend to neutrality about this one. I was born in Nigeria, and the most-sung worship song on earth came out of Lagos. That did not happen because it followed the rules better than anyone else's song. It happened because it was true, and its structure let a room join in.

---

The eleven golden rules

Everything above, compressed into things you can do to a song this afternoon.

1. Write between A3 and D5, and treat D5 as a place you visit, not a place you live. Four authorities name it exactly and four more land within a note of it. If your top note is above D, you are writing for a soloist. Touching E♭ once at the peak of a bridge is fine; sitting there is not.

2. Check tessitura before you check range. Where does the melody spend its bars? Sing your own chorus eight times without stopping, at the tempo, standing up. Whatever is aching afterwards is what your congregation will feel in verse two. So Will I has a narrower span than Oceans and is, on my reading of the lead sheet, the harder of the two to sing, because of where it is parked rather than how wide it is. Nobody has measured that comparison; what is measured is that one of them never reached the Top 25.

3. Write in the key the church will play, not the key that flatters your voice. C, G and D take 68% of chart downloads. D♭ is the most-recorded key in modern worship and takes zero. Pick the congregation's key first, then transpose the record for whoever is singing it, rather than the other way round.

4. Shorten your phrases. The biggest measured effect on whether a melody survives being passed person to person. Say the line out loud, at tempo, in one breath, while thinking about something else. If you cannot, your congregation cannot.

5. Move by step; when you leap, come back down by step. Stepwise motion survives transmission. The leap-then-fill shape appears in the folk research, in Read's step declination, and in the melody of In Christ Alone. Related but not identical: the earworm research found that melodies people cannot shake tend to combine a very common overall contour with unusually large leaps between the melody's turning points. A familiar outline, taken at surprising strides.

6. Stay mostly pentatonic. Scale degrees 1, 2, 3, 5 and 6, the notes people find without thinking. Save your borrowed pitch for the one moment that earns it.

7. Repeat your rhythms as deliberately as your words. Syncopation is not the problem; unpredictability is. A syncopated figure heard three times is a hook. A different one every verse is a trap. Keep the syllable count consistent between verses so the tune fits the words the second time round.

8. Four chords is not a poverty of imagination. Sixteen of the twenty-five most-sung modern worship songs use four or fewer. All twenty-five use I, IV and V. Redman's whole account of 10,000 Reasons is "four chords and the truth."

9. Aim at about 140 words and a reading age of twelve. That is where the measured repertoire sits. It is not dumbing down, since In Christ Alone carries a full creed in 224 words. It is choosing concrete words over impressive ones.

10. Decide, deliberately, who is singing, and to whom. Thornton counted how his 25 songs refer to God: eleven use only second-person pronouns, eight only third-person, and six use both. On pronouns for the singer the measured repertoire leans about three to one towards "I" over "we", whatever a British songbook shows. Any of these works. What does not work is drifting between them by accident inside one song. And a counted reassurance: in 20 of those 25, God is mentioned more often than the worshipper.

11. Demo it the way a congregation will sing it. No runs, no held notes above the melody, no countermelody on the last chorus. Your ad-libs teach people that the song is beyond them, and they respond by going quiet. Record the tune. Put the ornament on a different version if you must have it.

And one rule underneath all of them, which is not measurable and matters more. Getty called their process "long, drawn-out, painful, anarchistic, tense, argumentative, difficult, refining." Tomlin wrestled with his verses and said so. Redman's came in one of the quickest sessions of his life, at half past one in the morning, when he was trying to go home. There is no method here that substitutes for having something true to say. The rules above will stop a good song being unsingable. They will not make an empty song worth singing.

YHWH is the one of mine I would put in front of that claim. It came out of a comedian's joke about colonial missionaries, sent me looking for the name of God, and left me finding it in my own breathing. No rule in this article produced that. The rules only decided whether anybody else could sing it afterwards.

Sing to the LORD a new song; sing to the LORD, all the earth.

Psalm 96:1, NIVUK

All the earth. That is the brief. Not the people who can reach a high A.

---

What nobody has measured

The credibility of everything above depends on being straight about where it stops. These are real gaps, not gaps in my reading:

• Nobody has measured an actual congregation's vocal range. The undergraduate study is the anchor, and there is adult range data elsewhere, including Moore, Staum and Brotons on the over-65s and Moore comparing children with adults, but not one study of a mixed-age room in a church, as far as I can find.
• No test of tessitura against participation. Universally asserted, including by me, never measured.
• No study links key choice to how much people sing. Every "lower the key and they'll sing" claim in worship publishing, including rule 3 above, is craft experience rather than evidence.
• No longitudinal study of recorded keys or top notes over time. The upward-drift belief is unmeasured.
• No worship-specific research on phrase length or breath. The best evidence comes from Dutch folk songs.
• No study of how many hearings a congregation needs to learn a song. Practitioner estimates run from three to five.
• No count of how many current songs have a bridge, and no test of whether bridges help or hurt.
• No measurement of the male and female octave squeeze. Asserted by everyone, tested by nobody.

If you are a researcher looking for a question, several of those are sitting there unanswered, and the answers would change how a great many songs get written.

---

One more thing about the chart everybody quotes

If you are going to use the CCLI Top 100 as evidence of what churches sing, and this article does, you should know what it is now.

On 23 September 2024 CCLI began publishing a weekly Top 100 based on SongSelect download data. It is updated weekly, unweighted by church size, and not tied to royalties. The older twice-yearly chart still exists and still works the old way: six months of self-reported song usage from licence holders, weighted so that larger churches count for more, and tied to royalty payouts. It was not removed from SongSelect when the weekly chart launched, but it was taken out of the navigation, so it now takes a few extra clicks to reach; the six-month data is also published at worshipfuel.com.

Two different charts, two different things being measured. Helen Miller, comparing the two UK lists for May 2025, found they shared 81 songs out of 100, which is high, and also means a fifth of each list is absent from the other. A download is a worship leader considering a song. It is not a room singing it.

Worth knowing too: because the weekly chart is unweighted and a new song needs a fresh download, it favours smaller churches that download regularly, and newer material that nobody has the files for yet. Read it accordingly.

---

Why can't my congregation sing the songs from the radio?

Usually tessitura rather than range: the melody sits in a register suited to a professional lead vocalist, so there is nowhere comfortable for ordinary voices to be even when the span looks reasonable. Ad-libs are the other common cause, because when a leader departs from the melody, congregations stop singing rather than follow.

Both of those are fixable at the arrangement stage, and neither requires you to reject the song. Transpose it down until the busiest part of the melody sits around middle C, and have whoever leads it sing the tune as written for at least the first two times a room hears it.

How long should a worship song be?

The average recorded length of the most-used modern worship songs is 6 minutes 18 seconds, and in church use they run closer to five. The recorded arrangement is not the congregational arrangement, and it never was.

If you are writing one, the practical figure is the second one. Everything on the record that exists to make it a satisfying listen, the long instrumental build, the third repeat of the bridge, the ambient outro, is arrangement rather than song, and a congregation does not need any of it.

How many chords does a worship song need?

Sixteen of the twenty-five most-used modern worship songs use four chords or fewer, and all twenty-five contain I, IV and V. Four is not a compromise, it is the median.

Nineteen of the twenty-five add vi and thirteen add ii, so there is room for more if the song wants it. What the data does not show is any song reaching that chart on harmonic interest alone.

Does syncopation stop people singing?

Not by itself. Syncopation appears in every contemporary worship song in Read's sample, and that era is also the one whose songs fill the charts. The variable is predictability rather than presence.

A rhythmic figure repeated consistently across verses is learnable in one hearing and becomes part of the hook. One that changes each time through is what actually loses a room, because the congregation is sight-reading rhythm they have no way to anticipate.

How many times does a congregation need to hear a new song?

Nobody has measured this, which is worth saying plainly given how confidently it gets asserted. Practitioner estimates run from three to five hearings.

The common structure behind that estimate is to play a new song on three consecutive weeks before dropping it into normal rotation, so that the room has heard it enough to attempt it before it has to compete with songs everybody already knows.

---

Where to go next

• The difference between gospel and worship music, what worship actually means before you write one.
• The evolution of Christian worship music, how congregational song got here, from AD 112.
• Free and legal Christian music, including a full stem set you can study, free.
• The top 10 Nigerian gospel songs, ranked on chart data rather than taste.
• Everything I have recorded, free, including Behold The Lamb, which churches do sing.

---

Scripture quotations marked NIVUK taken from the Holy Bible, New International Version Anglicised, Copyright © 1979, 1984, 2011 Biblica. Used by permission of Hodder & Stoughton Publishers, an Hachette UK company. All rights reserved. 'NIV' is a registered trademark of Biblica. UK trademark number 1448790.

Sources

The measurements

• Kuhn, T. L., Wachhaus, G., Moore, R. S., & Pantle, J. E. (1979). "Undergraduate Nonmusic Major Vocal Ranges," Journal of Research in Music Education 27(2), 68 to 75, DOI; percentile bands as reported by Paul Ryan, "Highs and Lows of Singing," Reformed Worship
• Read, Daniel (2017). "Why We Sing Along: Measurable Traits of Successful Congregational Songs", MA thesis, University of Kentucky
• Janssen, B., Burgoyne, J. A., & Honing, H. (2017). "Predicting Variation of Folk Songs: A Corpus Analysis Study on the Memorability of Melodies", Frontiers in Psychology 8:621
• Jakubowski, K., Finkel, S., Stewart, L., & Müllensiefen, D. (2017). "Dissecting an Earworm: Melodic Features and Song Popularity Predict Involuntary Musical Imagery," Psychology of Aesthetics, Creativity, and the Arts 11(2), 122 to 135, DOI
• Hussey, Ian (2019). "The Songs We Sing: A Textual Analysis of Popular Congregational Songs of the 20th and 21st Century", Ecclesial Practices 6(2), 217 to 234
• Longhurst, Christine (2015). "The Words We Sing: An Exploration of Textual Content in Contemporary Worship Music", Direction 44(2), 158 to 172
• Baker, Shannan K. (2023). "A Typology of Ad-Libbing: Performing Authenticity in Contemporary Worship", Religions 14(3):337
• Fraisse, Paul (1982). "Rhythm and Tempo," in Deutsch, D. (ed.), The Psychology of Music, 149 to 180, for the spontaneous tapping figure Read builds his tempo trait on

Daniel Thornton's chart analyses

• "What on Earth Are We Singing? A 'Key' Perspective", Worship Leader, 2 May 2025, for recorded versus downloaded keys and the range distribution of the PraiseCharts top 50
• "What on Earth Are We Singing? Part 2" and Part 4, for pronoun and word counts
• "What on Earth Are We Singing? Part 5", for harmony, metre and final chords
• "What on Earth Have We Been Singing the Past 30 Years?" and "What on Earth Are We Singing… In Context?", Worship Leader, the latter for recorded song length
• Thornton, D. (2025). "Hillsong's Swansong? On the Decline of Hillsong Within the Contemporary Congregational Song Genre", Religions 16(4):427, on recognised songs that churches did not adopt

The range consensus

• Kenny Lamm, Renewing Worship, congregational keys, the cheat sheet, and "Reason #3: We Are Singing in Keys too High"
• Bob Kauflin, "Finding the Right Key to Sing In," Worship Matters
• Andy Chamberlain, Musicademy, on the right key for a small church and on worship songs and the local church; Kenny Lamm's guest post there on congregationally friendly keys
• Luke Woodard, "The Wrong Key," Discover Worship
• Ed Willmington, Worship Resources International
• Mark Cole on finding the right key · Alex Enfiedjian, Worship Ministry Training · Paul Ryan, "What Makes a Good Congregational Song"
• Worship Team Coach on ad-libbing · Renewing Worship, "9 Reasons People Aren't Singing"

The songs

• In Christ Alone: Hymnology Archive · Keith Getty at The Gospel Coalition · Stuart Townend's own account · James Cheesman on the melody
• How Great Is Our God: CCM Magazine's twentieth-anniversary piece · PraiseCharts · Musicnotes
• 10,000 Reasons: Matt Redman interviewed by the Evangelical Alliance · PraiseCharts
• Way Maker: the story behind the song · Sinach in Christianity Today · Baylor's Nigerian Christian Songs project · Musicnotes
• Oceans: Musicnotes · Jonathan Aigner's critique · Brian Houston on no longer singing it
• So Will I (100 Billion X): Musicnotes · the 49th GMA Dove Awards results, for what it was nominated for and what won

The chart

• Worship Leader Research, "The CCLI Top 100 Changed …And Nobody Is Talking About It"
• Miller, Helen D. (2026). "Singing in the Immanent Frame", Religions 17(2):201, on the two parallel CCLI charts and their overlap