Snow sports shops do much of their trade in a few busy months, from the new-season launch in autumn through Christmas and the February half-term. Those are the weeks when a slow page matters most. The pages themselves are built to sell a feeling: riding footage, big action photography and long product cut-outs of boards and skis. They also carry technical tools, such as size finders, flex and camber charts and comparison tables. Each of those has a cost that Google measures.

This guide explains those measurements, then works through the parts of a snow sports site most likely to slow it down.

How Google scores a page in peak season

Google’s Web Vitals programme sets three thresholds. Each is judged at the 75th percentile of real visits, on phones and desktops separately.

MeasureQuestion it answersGoodPoor
LCPWhen does the main image or text appear?up to 2.5 sabove 4 s
INPHow fast does the page react to a click, tap or key press?up to 200 msabove 500 ms
CLSHow much does the layout move unexpectedly?up to 0.1above 0.25

Field data comes from real Chrome visits and is reported in PageSpeed Insights and Search Console. It can only describe pages people have visited, so a new-season collection page has none until the traffic arrives. Test new templates with lab tools before launch, keeping in mind that Lighthouse cannot measure INP and reports Total Blocking Time in its place.

The 2024 Web Almanac performance chapter gives a few relevant figures for the web as a whole, gathered in mid-2024. On desktop, the LCP image was a JPEG on 61% of pages, a PNG on 26%, WebP on 7% and AVIF on 0.3%, with mobile similar. On mobile, 35% of LCP elements could not be found in the initial HTML. And 8% of mobile LCP images were over 1,000 KB. None of this is specific to snow sports, but the PNG and file size figures matter to a trade that relies on cut-out product shots.

What I measured on UK snowboard and ski product pages (October 2026)

On 23 October 2026 I ran five UK snowboard and ski product pages through PageSpeed Insights, using one mobile run each on an emulated mid-range phone over slow 4G. Two returned a 404 when tested and were left out, which leaves three pages: a very small sample.

PagePerformance scoreLab LCPTotal Blocking TimeLab CLS
Retailer A3613.8 s980 ms0.01
Retailer B5714.9 s0 ms0.00
Retailer C4225.3 s1,030 ms0.04

Those three had a median score of 42 and a median lab LCP of 14.9 seconds, and the slowest took 25.3 seconds. Unsized images and forced reflow were flagged on all three, which fits long board cut-outs and script-heavy size finders. Two pages passed 950 ms of blocking time. The third logged none at all yet still needed 14.9 seconds for its largest element, so its delay lay in loading rather than script.

No snow sports URL here had its own field data in the Chrome UX Report. On whole-site (origin) data, one of the three passed Core Web Vitals. A throttled lab phone is much slower than most real visits, and lab LCP is not the figure Google uses. Three pages cannot rank a trade, so treat them as examples.

Action footage and long product shots

Video heroes

Homepages and collection pages often open with riding footage playing behind a headline. For a video, LCP uses the poster image’s load or the first frame, whichever happens first. Chromium also leaves out elements that cover the full viewport. A full-screen video background may therefore not be the LCP element at all, and the headline over it may be measured instead.

That does not make the video free, because it still competes for bandwidth with whatever is measured. Keep it short, compressed and muted, with a small poster in the HTML. Edits hosted on a video platform are better behind a click-to-play preview image than embedded with the platform’s full player on load.

Boards and skis as images

A snowboard or ski is long and thin, so product images are either tall and narrow or wide with the board laid flat. Many shops use cut-outs on a transparent background saved as PNG, which tends to be heavier than the same image in a modern format. WebP and AVIF both support transparency, so the cut-out look survives the switch.

Following Google’s LCP optimisation guide:

  • Crop each file to the board plus a small margin. A tall image placed in a square frame carries pixels nobody sees.
  • Serve widths through srcset that match the gallery slot on a phone.
  • Keep the first product image in the HTML, never lazy-loaded, with fetchpriority="high" on that image alone.
  • Give offscreen gallery images, such as the base and the side profile, fetchpriority="low" or lazy loading.

Category tiles that swap from topsheet to base on hover are common in this trade. Hover is not an INP interaction, so the swap itself does not affect that score. The cost is in bytes: if every tile’s second image downloads with the page, a grid of 48 boards loads 96 images. Fetch the second image when a pointer first enters the tile or the tile receives keyboard focus.

Size finders, spec charts and comparisons (INP)

A size finder asks for weight, height, boot size and ability, then suggests a length. Typing a weight is a series of key presses, and INP counts key presses as well as taps. If the recommendation recalculates and redraws its chart on every key press, each one competes for the main thread. Wait until the buyer pauses or leaves the field, then calculate once.

Other tools on a board page carry the same risk:

  • Length selectors that fetch stock and price again for each length. Keep length data in the page and update the display at once.
  • Camber and flex charts drawn by a charting library on load. A static SVG or image with set dimensions often does the same job.
  • Comparison tables that render every spec of every model in a hidden layer. web.dev lists large DOM size among the causes of slow interactions, so build the table only for the models the buyer picks.

INP has three phases: input delay, processing time and presentation delay. A long script still running from page load shows up as input delay on the first tap, which on a board page is often the length selector.

Snow reports, promo bars and social feeds (CLS)

Snow sports sites add extras that other shops rarely carry: resort snow reports, weather widgets, social feeds of rider photos, and countdowns to a launch or sale. The usual review stars and finance offers sit alongside them. CLS scores the largest burst of unexpected movement over the life of the page, not only during loading.

  • A promo bar inserted at the top after load pushes every element down. Reserve its height in the initial HTML.
  • Social feeds and snow report widgets resize when their content arrives. Give their containers a fixed height.
  • Bold display fonts can render at a different size from the fallback. Choose a fallback with similar measurements.

Only discrete input exempts a shift. A shift within 500 ms of a tap, click or key press is excluded, but scrolling, dragging and pinch-zoom do not count as input here. A banner that slides in while the buyer scrolls through a board gallery therefore counts against the page.

Launch traffic, caching and server response

When an autumn launch email goes out, many visitors can arrive at once. The LCP guide gives a rough budget: about 40% of LCP for server response, about 40% for the resource download, and under 10% each for the two delays. Server response is the part most exposed to a sudden spike.

  • Cache collection and product pages at a CDN where your platform allows it.
  • Campaign links often carry tracking parameters, and web.dev warns that unique URL parameters can defeat CDN caching. Set the cache to ignore parameters that do not change the page.
  • Minimise redirects, such as a launch link that passes through two tracking domains before reaching the page.
  • Run lab tests on the new collection template a few weeks before launch day.

Before the season starts: a checklist

  1. Test the homepage and the main collection template in PageSpeed Insights, mobile first.
  2. Identify the LCP element on the homepage. If it is the headline over a video, check how long the video delays it.
  3. Convert PNG cut-outs to WebP or AVIF and compare file sizes.
  4. Load a category grid on a throttled connection and count how many images arrive before any hover.
  5. Type a weight into the size finder on a mid-range phone and watch for lag.
  6. Scroll a product page and note anything that appears and moves content.
  7. Open a campaign link with tracking parameters and check that the response comes from the cache.

For the search side of season labels, lengths and ex-demo stock, see snowboard and ski SEO. The wider selling questions are in selling snowboards and skis online. A 3D configurator for custom boards brings performance work of its own, described in how a 3D product configurator is built. For help before the season, see Core Web Vitals or Shopify speed optimisation.

Sources

Read October 2026, except the Almanac, published November 2024: