SnowDriven Research · Teardown
What a reveal-day page has to do.
Porsche’s Mission S page, measured on reveal day. It makes one claim per screen and keeps to a strict visual rule. On a phone it costs a long wait: 8.5 s before the main image shows, against 1.3 s on desktop, and most of that wait is JavaScript. For a launch page: design the first visit, and budget the scripts before the pictures.
- Measured
- 15 Oct 2026
- Page
- Mission S, US site
- Tool
- Lighthouse 13.5.0
- Runs
- 3 per device, median
- Data
- Lab, not field
The short answer
- Does well
- One claim per screen, a three-word thesis and a strict palette. The images are well compressed, and the videos wait until they’re needed.
- Costs
- On Lighthouse’s simulated phone: Performance 31 out of 100, and 8.5 s before the main image shows. Most of the wait is JavaScript, not pictures.
- Takeaway
- A launch page is judged on a phone, on a first visit. Set the JavaScript budget before the design is final, and design the consent dialog as part of the page.
Key facts, 15 Oct 2026
- Performance, phone
- 31/100
- Runs: 27, 31 and 45
- Performance, desktop
- 90/100
- Runs: 87, 90 and 92
- LCP, phone
- 8.5 s
- Desktop 1.3 s. Good is 2.5 s or less.
- Downloaded, phone
- 1,775 KiB
- 139 requests
- Scripts
- 63%
- 1,124 KiB of the 1,775 KiB
- Accessibility
- 97/100
- Phone and desktop
Our own Lighthouse measurements, median of three runs per device, 15 October 2026. Lab data, not field data.
The page practises what the car preaches
The design read: what the page does, in the order a visitor meets it.
One claim per screen
Each section makes one statement and stops. The headlines end in full stops: “A SPORTS CAR.”, “FORM FOLLOWS FUNCTION.”, “FOCUS ON WHAT MATTERS.” There is little on screen to compete with the line you’re reading.
A three-word thesis, said twice
Authentic, Consistent, Timeless. The words open the page as three pillars, one per slide, and come back later as a single line. A visitor who only skims still leaves with them.
Every detail tied to heritage
Features come with a history. The fan wheel at the rear is tied to the 917 race car, and the dial-style wheels to the 928. Each detail gets a reason to care, not just a description.
Copy with rhythm
The body copy is short and spoken. The gear lever gets three words: “Grab. Shift. Feel.” The sentences carry the beat of the headlines into the text.
One colour, used sparingly
The palette is black, white and grey, with red kept to a few accents, such as the red zone on the tachometer. When the accent is rare, it counts.
No calls to action in the body
There isn’t a link or a button in the main content. The page sells a feeling, not a form. That suits a reveal. Most business launch pages need a next step, so this is the one habit we wouldn’t copy as it is.
Page rhythm
- Statement
- Pillars
- Details
What it costs: the measurements
The same page on a simulated phone and on desktop, 15 October 2026. Lighthouse’s median run of three on each.
Largest Contentful Paint (LCP)
Total Blocking Time (TBT)
The data behind the chart
| Metric | Phone | Desktop | Good |
|---|---|---|---|
| Largest Contentful Paint | 8.5 s | 1.3 s | 2.5 s or less |
| Total Blocking Time | 749 ms | 25 ms | 200 ms or less |
The images aren’t the problem
The images are well optimised: 270 KiB across 21 images, every one with alt text. The two background videos are set to preload nothing, so they don’t load until they’re needed.
The phone wait is mostly JavaScript
Scripts were 1,124 KiB of the 1,775 KiB the phone downloaded, about 63%. The main thread worked for 6.8 s, 3.1 s of it running scripts, and about 416 KiB of script went unused. A tag manager accounted for about 177 KiB and the consent platform about 54 KiB.
The hero image waited to render, not to download
The largest element on screen is the poster image of the hero video. In Lighthouse’s breakdown of the recorded trace, it took 538 ms to download, then waited 1,556 ms to render. A smaller image alone wouldn’t remove that wait.
The consent dialog is part of the first visit
On a first visit the consent dialog covers the hero, and it caused the measured layout shift: CLS 0.183 on the median phone run, 0.001 on the run where the dialog’s timing differed. The dialog says it honours the Global Privacy Control (GPC) signal. That deserves praise, and copying.
The 38 contrast flags are a fade, not the palette
Accessibility scored 97. Lighthouse flagged 38 elements for colour contrast, and every one is a word of a scroll-in text effect, caught at 20% opacity before scrolling. The page has one h1, and its language is set.
Built with Astro, like this site
The page is built with Astro, the framework this site uses. The difference is what ships on top.
Five lessons for your launch page
What we’d take from it, for a launch page of any size.
- 01
One claim per screen.
Give each screen one sentence to land, and end it. If a section needs two claims, make it two sections. Read your headlines on their own: they should tell the whole story.
- 02
Let your product’s principle set the page’s rules.
Name the idea your product stands for and turn it into rules for the page: what goes in, what stays out, which colour is allowed. A product about focus gets a page with one accent and nothing extra.
- 03
Image first, video second.
Show a compressed poster image straight away and load the video only when it’s needed, as this page does. The visitor sees the product while the video waits.
- 04
Budget the JavaScript before the pictures.
Here the images were 270 KiB and the scripts 1,124 KiB. Set a script budget before the design is signed off, count every tag and widget against it, and test on a throttled phone, not the office laptop.
- 05
Design the first visit, consent dialog included.
A new visitor meets the consent dialog before the product. Decide where it sits, keep it off the hero, reserve its space so nothing shifts, and honour Global Privacy Control, as this page does.
How we measured
Lab figures from one evening, with the method stated.
Lab, not field
Lighthouse 13.5.0, run in a lab. These aren’t figures from real visitors: Chrome’s field data for the page wasn’t available to us.
Where and when
From a US cloud server, through its outbound proxy, on 15 October 2026, from about 21:50 to 22:00 UTC: reveal evening. The page: the US Mission S page.
Three runs, the median
Three runs on each device; we report the median run by Performance score. Phone Performance across the runs: 27, 31 and 45. Desktop: 87, 90 and 92.
Simulated throttling
Phone: Lighthouse’s default, which simulates a 150 ms round trip, about 1.6 Mbps and a 4× slower processor. Desktop: Lighthouse’s desktop preset.
A first visit
No cookies, so the consent dialog is part of what we measured. Headings, images, videos and links were read from the rendered page with a headless browser the same evening.
It may have changed
This is the page as it stood on reveal evening. It may have changed since, and a run today could give different figures.
This isn’t a scorecard of Porsche’s team. A reveal page carries brand, consent, analytics and video obligations a small business page doesn’t.
Porsche and Mission S are trademarks of their owner. SnowDriven isn’t affiliated with or endorsed by Porsche. See trademarks in our terms.
About this page
Cite this page
SnowDriven. “What a reveal-day page has to do.” SnowDriven Research. https://snowdriven.com/research/teardowns/reveal-day-launch-page/ (last updated 15 October 2026).
One click or tap selects the whole citation, ready to copy.
About these figures
The figures are our own lab measurements of a page we don’t run, taken on the day shown with the method stated above. Lab results vary from run to run, and the page can change at any time, so measure again before you rely on a figure. Information only, provided as is (terms). Reuse: free to quote, with credit to SnowDriven and a link (terms of use).
Start a project
Planning a launch page? We build fast ones.
We can set the JavaScript budget before the design, build the page to it, and measure it on a throttled phone before launch. Tell us what you’re launching; we’ll come back with questions or a time to talk.
Pittsburgh, PA · Since 2015 · First step: a 30-minute call