Huissteden.

Huissteden sells sewing machines and everything that goes with them: presser feet, bobbins, needles, thread. Thousands of items, each linked to the machines they fit. The brief in their own words was a feature-rich frontend and that complex product data, within an SMB budget, and still get through the Core Web Vitals.

We rebuilt the storefront on Hyvä Theme and Hyvä Checkout. Third-party scripts were removed or now load on demand, the cookie bar renders server-side, images are served automatically as right-sized WebP, and Magewire adds products to the cart without a page refresh. RUMvision measures the frontend on real visitors, New Relic and Tideways the backend.

That work won Best Performance Improvement at the Hyvä Hive Awards 2025 in London, for the largest speed jump from Luma to Hyvä that year.

Google Lighthouse mobile, before and after, on five templates: home page 66 to 100, category machines 60 to 84, category yarn and interfacing 62 to 95, product page overlock thread 59 to 98 and a Juki product page 37 to 97. An average improvement of 74.6 per cent. RUMvision, on real visitor data: LCP 100, CLS 100, INP 99, FCP 100, TTFB 95.

That jump was not made on an empty demo store. The fuller the catalogue and the more functionality a page carries, the heavier it gets and the harder speed becomes. Huissteden kept the machine finder, the linked parts and the whole assortment, and still got through. All the figures in this chapter are from 2025 and belong to that build, not to the 2026 design.

The design from the first build did its job. It is just that a few years have passed: we have learned more, more is possible than back then, and Huissteden likes being out in front of that. As an existing client the store chose Vendic AI Design on the platform that was already there, so the award-winning foundation underneath kept running.

That did not call for a redesign from scratch. The logo stayed, the Huissteden red stayed, the familiar categories stayed and the URLs of the products and service pages did not change. The machine finder runs on the same Magento module as before: in the old design you also picked the brand first and the type second. Anyone who knew the store landed in the same places after go-live.

The brand colours stayed, their roles changed. Red used to fill every button on the page and competed with itself. Now there is one rule: the buy button is the only filled button, in a calm navy base colour, which leaves red free for the place the eye needs to go. Typography went from two type families to one, Figtree for headings and body.

Those rules were first captured in a brand book, the step between corporate identity and theme, and then built: sixteen template areas against seven approved design pages, with two internal QA rounds and an acceptance test by Huissteden on a preview environment. On 8 September 2026 the new theme went to production, after which checkout was verified straight away with a test order.

Speed was the precondition throughout. On 9 September 2026 we measured five comparable pages again with Google Lighthouse mobile: 94 on the home page, 87 and 88 on the two category pages and 93 on both product pages. That is our own 2026 measurement and not the same setup as the 2025 one, so not a direct before and after. It does show that the new design did not cost the store its speed.

Every go-live produces a list. The question is not whether there is one, but who finds it and how fast it is empty. At Huissteden it was seventeen points, spread across sixteen template areas: a customer service page whose widgets were still on the old theme, image widths that landed differently in Safari, images in the blog articles, the country list of the EU store, and a newsletter and review block that sat on the page twice. Error monitoring held one alert that belonged to the release and did not touch the store.

That duplicate block did not come out of the preview round, and it could not have: the CMS blocks of the live store only sit on the page in their real composition once the new theme is the active theme in production. All seventeen points were resolved that same day, six hours and fifty minutes of work, last sign-off at 17:03. Since then, CMS pages with widgets from the old theme go on the aftercare list up front, and we do not fill up the go-live day.

After that day the store moves into the maintenance rhythm of Vendic Pulse. We assess security patches within 48 hours of publication and apply them when they are safe for this store, we test updates before they go live, we monitor errors 24/7, and two Magento and PHP upgrades a year are planned in. What sits in scope each month is set out in the maintenance plan.

There is no figure here yet on the effect on conversion or revenue. The new theme has been running since 8 September and a statement about that needs a measurement window of several weeks on comparable traffic. The figures in this case are therefore about the build and the platform, not about the effect.