Four concepts, each from different thinking. Every one is shown at 16 px in a real tab strip and at size, in dark and light. That row is the test — most of the argument for a logo is settled there and nowhere else, and two of the four below changed shape because of what it showed.
Before anything else: the current mark is a placeholder, and the current Share button label uses a pattern this work is deliberately not allowed to repeat — a lettered wordmark evoking the show's own on-screen styling. That is a live decision the owner is reconsidering, on the one control whose entire job is to push text outward into other people's chats. It is not reproduced on this page, on purpose: showing it here would work against the reason it is flagged. Nothing below uses it, and the recommendation at the end includes removing it.
Replace — and the reason is measured, not aesthetic. The three obvious ways to redraw the stopwatch as a real icon all collide with a symbol every phone user already knows. Rendered at 16 and 48 px:
A ring with a hand is the power button. A disc with a horizontal bar is the no-entry sign — and that one is worse than it looks, because the app's timer bar turns brick red at the critical mark, so a red-adjacent stopped-sign is the last thing this product should own. The third (a disc holding three lines) survives, but it is concept A wearing a circle, not a different idea.
Set against that, the current mark's equity is close to nil in cost terms: the site is pre-launch and gated to testers, so nobody outside has learned it. The replacement keeps its one real virtue — a mark that costs no bytes and follows the page — and spends the idea budget on the premise instead: three lines, one of them true.
The first tab in each strip is the site as it stands today —
<link rel="icon" href="data:,">, a literally empty icon, on all three pages.
Why the favicon gets a filled tile and the in-page mark does not. A browser tab strip is
#dee1e6 in light Chrome and #202124 in dark, and the site cannot know which.
A bare stroke mark has to survive both; a tile only has to survive itself. So the favicon is a solid
brand tile with white marks — Wordle's answer to the same problem — while the mark on the page stays
two ink tones and follows prefers-color-scheme exactly as the stopwatch does now.
Authored once at full size, then shrunk. 330 px is roughly a WhatsApp in-chat preview; 160 px is a Telegram compact row. Two deliberate calls:
No launch date on the card. WhatsApp and Telegram cache an og:image hard and by
URL, so a date baked into the picture becomes a stale asset on 11 September with no way to bust it
short of renaming the file. The date belongs in the og:description, which is cheap to
edit. The pitch line breaks at the sentence, not at the box. The string never changes, so the
rag is a decision, not an outcome — left to the box width it puts "said." alone on line two.
An og:image has to be a PNG or JPEG — WhatsApp, Telegram and Twitter will not render
an SVG. The cards above are HTML, so somebody has to rasterise the chosen one at exactly
1200×630 before the meta tag can be added. The favicon does not have this problem: a single
.svg covers every modern browser, with a 32×32 .ico only for old Safari.