The idea comes out of the reference, never the other way round. Every caption, overlay and cut should trace back to something we already saw work. Same for the music: find the track, then cut to it.
The whole guide in two columns. The right-hand one matters more.
We work with different editors and creators on every project. This is so you do not have to guess what we are after — and so you can come back with finished content, not just a set of cuts.
Good cuts on their own are not the job. The job is cuts that were chosen, text overlays where they make it stronger, and an editor who can tell what is working in a piece and what could be better.
If the cut already exists and we go hunting for a reference for the caption or the overlay, that overlay is a guess dressed up as a decision. It was not chosen for a reason, because at the moment it was chosen there was no reason available. You can feel it in the content even when you cannot point at what is wrong.
Instagram and TikTok quietly limit posts carrying certain words. Censoring keeps the reach without changing what we say. Do it in on-screen text, captions and subtitles — all three get scanned.
Every term we censor. No need to go further — pleasure, intimacy, wellness and lube are fine as they are. Unsure about a word? Just ask.
A claim can mute or pull a post, usually once the work is finished. Worth checking a track is cleared before you build around it.
Three faces. The sizes below are what goes on screen, at their real size relative to each other.
Apps cover the edges of a 9:16 frame. Keep anything that needs reading in the middle.
Headers, captions and the logo sit inside the safe zone. Footage can fill the whole frame.
Cut-offs move between apps and phones, so tighter is safer.
Six formats. Two of them are quick wins; the other four are what every event owes. Know which one you are making before you open the timeline.
Our editing style has been evolving, and it is still moving. The examples below are a feel for what we have been making recently — not a locked specification. We are working toward something more organised and more specific, particularly on typography, and we are locking it in as we go.
So do not copy these frame for frame. What matters is that the cuts and the type are interesting, that you can read an outlier and tell why it worked, and that you can move with the style as it tightens.
Find the track before you cut anything, and let it decide where the cuts land. Cutting first and dropping a beat underneath afterwards is the same mistake as editing first and hunting for a reference after — the timing was not chosen, it was inherited.
Every setup has motion in it. If a shot sits still for more than a beat it is either a deliberate hold or it is out.
One clean digital pass, one lo-fi. Both live in the same cut — the contrast between them is the look.
Out-of-focus and lo-fi frames are selects, not mistakes. At least one survives into every deliverable.
Do not grade the lo-fi toward the digital to "unify" a sequence. That difference is the point of shooting twice.
All colour on screen comes from what was shot. Type and overlays stay white or cream — never coloured.
Hero text is Helvetica with adjusted measurements, campaign title top, logo bottom. Full spec under Typography.
Reels are shot and cut 9:16 and labelled reels. A 9:16 crop of a carousel is not a reel and reads like one.
Nobody has to go hunting for these right now. This is just so that when one lands on your feed, you can tell in about ten seconds whether it is worth saving.
A video with loads of reactions in the comments, or copy on screen that is genuinely good. That feeling is the trigger — it is not the answer yet.
Do not judge the post on its own. The number means nothing until you know what normal looks like for whoever posted it.
Just look at the view counts on their recent posts. That is the whole check — no maths, no tools.
A real outlier is unmissable — the bar towers over everything around it. If you find yourself doing sums to argue that it counts, you already have your answer. Keep scrolling.
The overlay earns the stop. The caption deepens the idea. They are never the same sentence, and neither of them narrates what is already on screen.
Nancy sounds like a woman who understands culture, intimacy and marketing, and is bored of pretending otherwise. Genuine, witty, confident, unbothered. She can be funny without performing humour, and can sell without sounding desperate to convert. Never crude just because the category is.
Captions and text overlays, written in Nancy's voice. Use this rather than writing from scratch, and rather than any general AI tool that does not know the brand.
Real captions, and the reason each one landed. The pattern is the same every time: the footage is the evidence, the caption is the meaning.
Formats in heavy rotation across the platform, grouped by the emotional engine that makes each one work. Keep the logic, change the filling. Pick the engine that matches the footage first, then find the most Nancy-specific thing to put in the blank.
Four days in New York. Scroll sideways for what is shot each day. Most of it is filmed twice, once properly and once rough on purpose — both get used.