Craft·June 2026

Colour Grading for Social: The Quiet Lever Behind Watch Time

There's a reel I watched maybe fifteen times one night last year. A fitness creator I'd never heard of, decent hook, ordinary edit. It was 1 a.m. and something about it was bugging me and I couldn't name it. I do this more than I'd like to admit — a video gets under my skin and I sit there replaying it like it owes me an explanation.

Around the tenth replay I muted it, and then I saw it. The colour. His skin looked the same in every single shot — gym, car, kitchen, street. The shadows had a consistent soft warmth. Nothing screamed. Nothing was "cinematic" in the LUT-pack sense. It was just quietly, boringly coherent, and my eye had nowhere to trip.

That's the whole trick. The reel wasn't better shot than a hundred others in its niche. It was better finished. And finishing, on social, mostly means colour.

Most creators think colour grading is about making things look nice. It does that, sure. But "nice" is the smallest thing a grade does, the way "tastes okay" is the smallest thing a good meal does. The real job is retention and recognition — how long people stay, and whether they know it's you before a single word registers. That's the argument of this essay, so let me try to earn it.

I used to treat the grade as decoration

Quick admission, because I wasn't always on this side of the argument.

In the early days of With Media I undercharged and said yes to everything, which meant every project was a race, and when a project is a race the grade is the first thing sacrificed. It's the last step in the pipeline; the deadline lands on it with its full weight. I'd finish the cut, feel the clock, slap on whatever preset was lying around — usually something teal-and-orange, because that's what the internet had convinced me "cinematic" meant — nudge the exposure, export, done.

I thought that was grading. Pick a vibe, apply the vibe.

What snapped me out of it was watching one of our own videos back a few weeks after delivery. The client's face went from slightly sunburnt to slightly seasick across a single scene, because two cameras had disagreed about what his skin looked like and I'd never made them agree. Nobody had complained. The client hadn't noticed, or was too polite to say. But I couldn't unsee it, and once you can't unsee it you start noticing that the videos you admire never have that problem.

So yes — I run a studio that charges for this craft, obviously I'm biased. But I was on the other side long enough to know the difference between the two mindsets. One treats colour as a filter. The other treats it as a stage of the work with its own decisions, its own hours, its own reasons. Everything below is about the second one.

Flat footage is a feature, not a bug

Start where most confusion starts: the moment someone sees their raw footage for the first time.

flat log footage vs the image you meant.

If you've ever shot on a proper camera — or a phone in log or "pro" mode — you know the mild heartbreak of the playback. Everything looks grey, washed out, low-contrast, like the world through a dirty window. Clients see this on set and you can watch their face fall. This is the expensive camera?

Here's what's actually happening. The camera is deliberately recording a flat image because flat holds the most information. Highlights not yet burned to white, shadows not yet crushed to black, colours not yet committed to anything. It's capturing possibility, not a picture.

The analogy I use with every client is raw dough. Nobody looks at dough and says the baker doesn't know what he's doing. Dough is supposed to look like that — flour and water holding every bread it could still become. The oven is where it turns into the thing you actually eat. Log footage is dough. The grade is the oven. Posting ungraded footage is serving dough, and the strange part is how many people do exactly that and then wonder why their content feels unappetising in a way they can't articulate.

The flip side matters just as much. Footage straight off a normal phone camera isn't dough — it's a microwave meal. The phone has already "graded" it with computational photography: shadows lifted, saturation boosted, everything sharpened, skin smoothed. It looks immediately fine, which is why phone footage is so seductive. But it looks fine in exactly the way two billion other people's footage looks fine. The phone made all the creative decisions, and it made the same ones for everyone.

Either way, the image hasn't been decided yet. Someone has to decide. The only question is whether it's you, a preset, or an algorithm in Cupertino.

The grade is where footage stops being a recording of what happened and becomes the image you meant.

What a real grade actually does

By "a real grade" I mean someone going through the video shot by shot with intent — not a look dropped on the timeline. In our pipeline it breaks into four jobs, jargon kept out:

Consistent skin tones. The number one professional-versus-amateur tell, by a distance. More on this below because it deserves its own section.
A deliberate palette. Choosing two or three colours the video lives in, and gently pushing everything else towards or away from them.
Depth and contrast. Deciding where the eye goes. A graded frame has a subject and a background; an ungraded one is just a rectangle of stuff.
Mood matching. The temperature of the image agreeing with the temperature of the content — warm for food, cool and clean for tech, rich and moody for story.

That list looks tidy. The work isn't. It's forty shots filmed across three days under four kinds of light, all needing to feel like one continuous world. The actual labour of grading isn't inventing a look — it's enforcing one.

Skin is where amateurs get caught

One practical thing to take from this essay: when a video feels "off" and you can't say why, look at the faces.

Skin is what colourists call a memory colour. Your audience has no idea what your wall is supposed to look like, or your desk, or the sky out your window. The grade can lie about all of those and nobody will know. But every human alive has spent their whole life looking at faces. We are frighteningly well-calibrated for skin. A few degrees too orange and the person reads as spray-tanned. Too magenta and they look flushed, faintly embarrassed. Too green — and cheap fluorescent lighting loves green — and they look unwell. Nobody watching can name the error. Everyone watching feels it.

Why does skin go wrong so often? Almost always, mixed lighting. A creator sits near a window (blue daylight) with a warm lamp on one side and an LED panel doing something else entirely. The camera picks one white balance and betrays the other two sources. Then a LUT built for somebody else's camera on somebody else's set gets dropped on top and shoves the skin wherever it shoves everything. Then shot two was filmed at 4 p.m. instead of 11 a.m., and now the face changes colour on the cut.

One shot like that is survivable. A video assembled from many of them is a face that flickers between five versions of itself, and the viewer's brain quietly logs it: something here is not professional. They don't think it in words. They just trust the video a little less, and on social, trusting a little less means swiping.

So when I review a grade at the studio, skin is the first thing I check and the last thing I sign off. Does the face look like a healthy human in every shot, and do all the shots agree with each other? Only after skin is locked does anyone get to be creative with the rest of the frame. That discipline is most of what separates the editors who get colour from the ones who just own a lot of LUTs. I've hired somewhere around a hundred editors building this studio, and I can count on my fingers the ones who instinctively protected skin before chasing a look. It's rarer than good taste, because it isn't taste at all — it's care.

Where I look when a grade is off

Since I'm being nerdy anyway, here's my actual checklist when a cut comes to me and the colour feels wrong. It's short, and it's ordered.

Skin first, always. Then the blacks. Shadows tell you almost everything about whether an image feels expensive. Milky, lifted blacks read as webcam. Crushed-to-pure-black kills detail and feels harsh, like a meme compressed twelve times. The expensive feel lives in shadows that are dark but still breathing — you can just barely sense the texture inside them.

Then shot-to-shot continuity. I scrub the timeline fast, almost too fast to watch, because at speed the eye stops following content and starts noticing jumps in brightness and colour between cuts. A well-graded video scrubs smooth, like one long take. A badly graded one strobes.

Last, the saturation balance between subject and background. The amateur move is cranking saturation globally because it looks juicy in a thumbnail. But if the background is as loud as the face, the eye has to work to stay on the person, and a working eye is a tiring eye. We let the subject keep more colour than the world around them. Nobody notices — the viewer just finds the video strangely easy to keep watching.

None of this needs a colourist's certificate. Faces consistent, shadows alive, cuts smooth, subject louder than background. If you check nothing else on your own videos, check those four.

You'll never see "grading" in your analytics

This part took me longest to believe, so I don't expect you to believe it immediately either.

a grade shows up as retention, not a metric.

Open any analytics dashboard — Instagram, YouTube, anything. There is no graph called "grade quality." There never will be. Which is exactly why colour is so easy to dismiss: it never shows up under its own name. It shows up wearing other names. Retention. Watch time. Shares.

Think about what retention physically is. It's a thumb, hovering, deciding every half-second whether to keep resting or to flick. That decision isn't rational. Nobody consciously thinks "the skin tones in this video are inconsistent, I shall depart." It happens underneath language. A coherent, intentional image is restful — the eye settles into it the way you settle into a well-made chair. An incoherent one produces low-grade visual static, a dozen tiny wrongnesses per second, and the thumb resolves the discomfort the only way it knows how.

I've watched this play out across well over a thousand videos through our studio, and I'll be straight about the limits of what I know: I can't hand you a controlled study, and anyone quoting an exact percentage for "grading lift" is selling something. What I can tell you is the pattern we see over and over — same creator, same format, same kind of hook, and the properly finished pieces hold people deeper and get shared more. Shares especially. Sharing is a taste signal; when someone sends a video to a friend they're saying look at this, and people are far more willing to attach their taste to something that looks finished.

Grading never appears in your analytics under its own name. It shows up as retention, and retention shows up as reach.

Could those videos be holding because of a hundred other things too? Of course — the edit, the script, the person. Colour is one lever among many, and I'd never claim it's the biggest. I'm claiming it's the most neglected relative to what it costs, because unlike a better script or better on-camera presence, it's entirely fixable in post, by someone who isn't you, while you sleep.

The warm grade that made a phone look like a streaming show

Let me make this concrete with work you can go look at.

A lot of Dr. Shilpa Arora's content is shot in her kitchen — Morning Routine, What I Eat In A Day, the Ghee Series, all of it. Kitchens are brutal for colour. Mixed light everywhere: a window on one side, warm bulbs overhead, steel surfaces bouncing everything back at you. Ungraded kitchen footage tends to look like CCTV of someone cooking.

The decision we made early was a signature warm grade for her food content. Warm, but specifically — anchored in the colours already living in that kitchen. Brass, wood, turmeric, the gold of ghee melting in a pan. The grade doesn't invent that warmth; it finds it in the footage and turns it up while quietly turning down everything that fights it — the blue-grey of window light on steel, the green cast off the tiles. Her skin stays in that healthy, sunlit zone through every shot, because a health expert who looks unwell is a contradiction the viewer feels instantly even if they'd never phrase it that way.

The effect, and I say this having watched the raw files and the finished pieces side by side more times than anyone should: footage that was often shot on a phone reads like a streaming cooking show. Same kitchen, same phone, same person. The grade is doing the translation. When people ask what camera she uses — and they do — the honest answer is that they're not responding to a camera. They're responding to intent applied after the fact.

And the warmth isn't decoration, it's mood matching. Food content is supposed to make you feel hungry, comfortable, taken care of. Warmth does that below thought; the same footage graded cool and clinical would be accurate and completely wrong. That's the general rule. Cool and precise for tech. Rich and shadowed for story. Warm for anything you're supposed to want to eat or feel at home inside.

Rangeela, or grading until it stops being footage

The other end of our range is a piece called "Rangeela" — embedded right below this essay, and I'd rather you watch it than trust my description.

The brief we gave ourselves was simple to say and hard to do: it should feel like film, not footage. Footage is what a camera saw. Film is what someone meant. And the difference between the two, more than the lens, more than the movement, is the colour.

So Rangeela was shot knowing the grade was coming — which is its own lesson. The best grades are planned before anyone presses record, because a grade can only push colours that exist in the frame. In the grade itself, the highlights roll off soft instead of clipping the way digital video does, the shadows carry colour instead of falling to neutral black, and the palette is committed — the whole piece lives inside a handful of colours that were chosen, not collected. Skin stays honest through all of it; everything else is allowed to be heightened as long as the people remain people.

Watch it once normally. Then again, asking one question: does this feel like something that happened, or something that was made? That gap — happened versus made — is the entire distance a grade can carry you, and it works at every scale. A thirty-second reel sits somewhere on that spectrum too. Most creators don't realise they're on it at all. They're publishing "happened" and competing against people who publish "made."

Colour is recognition before a word is read

One more job the grade does, and long-term it might be the most valuable: it makes you recognisable at a glance.

Try this with any big creator you follow. Open their grid and squint until you can't read a single caption. You'll still know it's them. Before a word, before a face resolves, the colour is consistent — the same warmth or coolness, the same density of shadow, the same handful of hues month after month. That's not an accident of their camera. That's a decision, held.

This is branding in the truest sense, and it can't be faked with a logo. A logo is something you look at. A palette is something you're inside of. When your look is consistent, every video makes the next one more recognisable — the work compounds. Someone scrolling at speed catches half a second of your frame and something in them says oh, it's them before any conscious recognition happens. On a feed where every video begins as a stranger begging for attention, beginning as a friend is an absurd advantage.

Building that signature isn't complicated, but it demands the one thing most creators won't give it: repetition without boredom. You pick a look that suits what you actually make — not the look you admired on some filmmaker whose content is nothing like yours — and you hold it for months. Through trends. Through the itch to try the new thing. A grade becomes a signature the way handwriting does: not by being designed once, but by being repeated until it's involuntary. Shilpa's warmth is recognisable now because it has been hers across hundreds of videos, not because it was clever on day one.

The objections, honestly

A few pushbacks I'd raise myself, answered as straight as I can.

"Isn't this just filters with extra steps?" No — the difference is decisions-per-video. A filter is one global decision. A grade is dozens of small ones anchored to this footage — this face, this room, this mixed light. The filter doesn't know your skin went green in shot seven. That said, an editor who applies one LUT and calls it a grade is selling you a filter with extra steps, and plenty do.

"Will my audience actually notice?" Consciously, almost never. Nobody has ever commented "great skin tone consistency." But you're grading for the thumb, not the conscious mind, and the thumb notices everything. You already feel the difference between a Netflix frame and a Zoom call without once thinking about why.

"Can a grade save my footage?" Here's where I'll argue against my own shop: no. If the lighting is chaos, if the exposure is blown, the grade can only choose the least bad version of your mistakes. Dough analogy again — the oven can't fix a dough you didn't knead. And if you're making pure talking-head information content with decent, even lighting, honestly, your audio matters more than your colour. Fix that first. Grading is a multiplier on visual craft that already exists; multiplying zero is still zero.

"I can't afford a colourist." Then run the four-point check yourself — faces consistent, shadows alive, cuts smooth, subject louder than background. That alone puts you ahead of most of your feed. This essay isn't "hire us or suffer." It's "stop treating the last ten percent of the work as optional," whoever does it.

The quiet part

I keep coming back to that 1 a.m. reel. What unsettled me wasn't the reel itself. It was realising how much invisible work I was responding to — that somebody had sat there matching skin across a gym and a car and a kitchen, and I, a person who does this for a living, needed ten replays and a mute button to catch it.

That's the nature of the grade. It's care that nobody claps for. The hook gets quoted, the edit gets noticed, the grade just gets felt — as trust, as watch time, as the vague sense that this person knows what they're doing. Most levers on social are loud. This one is quiet, which is exactly why it's still lying there, unpulled, in almost every niche I look at.

Pull it properly and nothing dramatic happens. Your footage just starts looking like something people would pay to watch. And slowly, without being able to say why, they watch longer.

“Rangeela” — graded to feel like film, not footage.