Redesign proposal · unofficial · 2026-08-19

The product sells itself.
The page keeps getting in the way.

CherryClipz has one idea that needs no explaining: your video becomes a sheet of lines, and you tick the good ones. The current site says so, then asks you to wait for a live editor to load before you can feel it. Most people meet a black rectangle instead.

So this concept does not describe that idea. It performs it. You scroll, a flat transcript grows timecodes and speakers and thumbnails until it is a spreadsheet, and then you cut a video yourself — before anything asks you for an email.

What I found

I crawled every public route: /, /flagship, /plugin, /subscription/, /my-library/ and /landing/. Everything below is observed on the live build.

1. The front door does not reach the good page

/, /my-library/ and /subscription/ all land on /landing/ — an older, thinner page headed "Every story has its cherries pick yours." The strong page, /flagship, is only reachable if you already know the URL.

Anyone arriving from a link, a search result or a business card sees the weakest version of the site.

2. The hero's first paint is an empty black box

The hero embeds the real editor — a brave idea. On a cold load it renders a full-height dark panel reading "Loading the live editor…". The proof arrives after the moment the visitor decides whether to stay.

3. The hero pins and rewrites its own promise

Scrolling does not move the page at first. The headline swaps three times: "Cherry-pick your clips." → "Any clip in just one click." → "Your footage, fully searchable." Three promises in one spot, about 2,000 px of scroll with no forward progress, and broken keyboard navigation.

4. The navigation lists "Premiere plugin" twice

Seven items, one duplicated, labels wrapping onto two lines at ordinary desktop widths — on the page whose job is to look finished.

5. The two marketing pages contradict each other

A visitor who reads both pages cannot tell what the free tier includes.

6. One story, told twice

/flagship is 7,109 px tall and embeds the live editor three separate times. /plugin then repeats the "Choose." section, the phone-to-Premiere demo and the pricing table in different words.

7. The marketing mocks do not match the live editor

The phone panel drawn on /plugin shows six clips. Five of them match the sheet the hero actually loads. The sixth — "You have to take a leap of faith and do what feels right." at 00:38 → 00:44 — is not in the live data at all. The real sheet has "How's it going over there?" at 00:00:36 and "I hope that I can find the true meaning of my existence…" at 00:00:40.

The static mock elsewhere on the page invents more: "This is the moment everything changed.", "Let me show you exactly how it works.", "I almost gave up right here." None of them appear in the sheet. It is a small thing, but the page is arguing that the tool reads your footage accurately.

8. The vocabulary drifts

The same object is a clip, a quote, a moment, a line, a sentence and a row — sometimes within one section. Three names sit close together (CherryClipz, CherryPick, Black Cherry) and the page never says plainly that the last two are two modes of the same sheet.

The one big move

Show the transformation, then hand over the controls.

The middle of the page is a single sticky scene. It opens on eight sentences of transcript — grey, flat, structureless, exactly what a transcription tool gives you today. As you scroll, timecodes slide in. Speakers and sentiment appear. Thumbnails open. Rows separate. By the end of the scene it is unmistakably a spreadsheet, and nothing has loaded.

Then the page stops talking and gives you twelve real rows to tick. When you have picked two, one line appears: "You cut 0:14 out of 4:12 in six seconds." That number is measured from your own clicks, not invented. It is the strongest sales argument on the page, and the page did not have to make it.

9. The best surface is the least advertised

The phone view is not a cut-down sheet. It is a second product: a vertical reel player with the captions burnt in, a like count, a CherryPick button that pulls a clip into your selection while you watch, and a Reframe control for 9:16. Neither marketing page shows it above the fold, and the plugin page does not mention it at all.

The same is true of the sheet's own toolbar — Silences (show, hide, only), Find filler, Score virality, per-clip Download. These are the reasons an editor would switch, and they are buried behind an embed that takes half a minute to load.

What changed

Why it looks like this

Black for the product, white for the argument

Every piece of product imagery is dark: the editor, the phone, the Premiere panel. On a cream page each one reads as a hole punched in the layout. Here the product scenes are true black and the marketing statements are white, so the interface sits in the design instead of on top of it. Editors also work in dark applications all day.

The system typeface, on purpose

No web fonts. The stack resolves to SF Pro on Apple hardware, which is what this typography was drawn for, and to a good native face everywhere else. It also means nothing blocks the first paint — which matters on a page whose whole argument is that you should not have to wait.

One accent, used sparingly

The cherry appears on the primary button, on ticked rows, on the cut bar and almost nowhere else. Everything else is black, white and grey. Colour means "you did something" rather than "this is a website".

The cut is a bar, not a list

Picked clips render as one proportional strip — each segment sized by its duration, in pick order. It is the smallest honest picture of what you have made, and it doubles as the thing you would export.

The demo data is sampled, not invented

Every row in every demo on this page is read from the sheet the current hero loads (/s/pSG_9ZJz…): nine clips, LIST 00:00:32, speakers A, D, E, F, G and H. The transcript text, the Seconds and Length values, the Q? markers and the Things nouns are the product's own output, including its elisions and its rounding — clip 3 runs 00:00:23 → 00:00:25 but reports 1 second, and that is reproduced rather than tidied.

The screenshots are the product itself, captured from the live build on 19 August 2026 and not redrawn. Two things in the interactive demo are stand-ins, and are marked as such below: its thumbnails, because I have no frames from the source footage, and the Black Cherry scores, because that call needs an API key. The editor also carries Actions (Verbs) and Descriptions (Adjectives) columns that this page leaves out for width — worth putting back if the sheet gets its own page.

The thumbnail column is left out for a different reason. Fabricated frames looked exactly like what they were, and next to real screenshots they cheapened the whole page. Cutting them turned the scroll scene's final beat into something better: instead of "thumbnails open", the payoff is the language columns — Q?, Things — which is the part no other transcription tool has.

Details

• Every tickable row is a real <button> with aria-pressed, so the demo works from the keyboard.
• The reveal states are gated on a .js class, so nothing is permanently invisible without JavaScript.
• All motion stops under prefers-reduced-motion, including the typing and the scroll scene.
• Tabular numerals on every timecode, duration and price.
• Nothing loads late, so there is no layout shift.

Scope

Static HTML, CSS and about 280 lines of vanilla JavaScript. No framework, no build step, no backend. Nothing connects to an account, uploads a file or takes a payment.

• Not affiliated with CherryClipz. Made as a proposal for its author.
• The product name, the brand colour and the sample transcript belong to their owner and are used only to make the proposal legible.
• The pricing numbers come from the live site. Where the two live pages disagree I used the /flagship figure and flagged the conflict above.
• The interactive sheet has no thumbnail column. The real editor does. I tried colour fields standing in for frames, and they read as invented — so the column is gone and the real frames appear only in the screenshots, where they are genuine.
• Black Cherry scores are illustrative. The picks and the personas are real; the percentages are not model output.
• Copy is a first draft, written to fit the existing voice rather than to replace the author's judgement about it.

If you take only three things

1. Point the root URL at the flagship page. One line of config; the largest single win.
2. Put the reel player on the page. It is the most convincing thing you have built and it is currently three scrolls down, behind a loading state.
3. Give people something to tick before the video loads. Keep the live embed — move it below the fold, behind a click.
4. Make the free tier say the same number on every page.