Site Speed as a Trust Signal: Brompton, Vodafone, Zalando

I have watched a lot of e-commerce rebuilds, and the thing that quietly keeps a mid-market store from converting is almost never what the marketing team wants to talk about. It is the half second the Add to Basket spinner takes. The three seconds the account page takes to render after a password reset. The ten seconds the checkout holds the card between tap and confirmation. None of this reaches a dashboard a CMO reads, and none of it is on the agenda in the quarterly review.

For a long time, agencies sold site speed as a ranking factor, which it is. That framing turns it into a line in an SEO report, buyers nod, and the file gets closed. On a considered purchase, though, something more useful is going on. The speed of the site is one of the few pieces of evidence a visitor has that the brand can be trusted with their money. Nothing on the page literally says we will fulfil this order. The page speed says it on the brand’s behalf, before the visitor has consciously registered any of it.

So I went and read what three companies that take this seriously said in public. A British folding-bike maker that priced its own checkout in seconds. A European telco that ran the A/B test at a scale most of us cannot. A continental fashion retailer that measured revenue moving against a hundred milliseconds. They sell different things. They reached the same conclusion.

Here are Brompton Bicycle, Vodafone Italy, and Zalando, and what each put a figure behind.

Brompton Priced the Trust Problem in Seconds

Brompton is a British maker of premium folding bicycles with a direct-to-consumer channel and a basket that is comfortably in four figures. Before the replatform, their checkout took a full ten seconds to resolve. For a shopper with their card in hand and the house credit limit in the back of their mind, ten seconds is long enough to form a very specific thought: is this one of those websites that is about to charge me twice?

Chris Matthews, Brompton’s Global DTC Channel Director, put the framing on the record in a 2025 Econsultancy interview: “Nothing gives you a lack of trust like a slow website.” The team rebuilt the storefront with speed as a first-order requirement rather than a tuning pass later. Post-replatform, Brompton reports page loads in the one- to two-second range and treats site speed as a core success metric for the premium brand.

• Checkout time before the replatform: around 10 seconds (Econsultancy, 2025).
• Page load times after the replatform: 1 to 2 seconds.
• Metric Brompton explicitly names as critical for a high-ticket, premium brand: site speed.

The transferable lesson is not the number of seconds. It is that Brompton stopped treating speed as a performance report and started treating it as a trust test the shopper runs on the brand without being asked. On a premium purchase, failing that test quietly is worse than failing it loudly, because the shopper does not file a complaint. They close the tab.

Vodafone Proved It Under A/B Conditions Most of Us Cannot

Vodafone Italy ran the kind of controlled test most e-commerce operations never get to run: identical landing page, identical traffic sources, identical offer, split fifty-fifty, with the only difference being that one version had been optimised for Core Web Vitals and the other had not. The optimisations were server-side rendering of critical HTML, moved from client-side, and a run of image work (resizing, SVG optimisation, lazy-loading, PNG compression). About 100,000 clicks and 34,000 visits per variant, per day.

Davide Grossi, Head of Digital Marketing (Business) at Vodafone, said after the test, in the web.dev case study published 17 March 2021: “Thanks to the collaboration with Google and the introduction of LCP as the main KPI for page performance, it was possible to significantly improve the customer experience of our e-commerce.” That is a careful, corporate sentence. The underlying numbers were not careful.

• Largest Contentful Paint improvement on the optimised variant: 31%.
• Total sales uplift: 8%.
• Lead-to-visit rate uplift: 15%.
• Cart-to-visit rate uplift: 11%.
• Source: web.dev case study, 17 March 2021.

The piece of this that an SEO report will skip past is the shape of the three uplifts. Sales moved by 8%. The intermediate steps, lead-to-visit and cart-to-visit, moved by 15% and 11%. The further up the funnel you look, the larger the effect, because that is where the trust test is actually being run. By the time a visitor has reached the cart, they have already decided the site is credible. The speed won them that credibility earlier, on the pages where they were still deciding whether to engage at all.

Zalando Showed the Hundred Milliseconds Mattered

Zalando sits at the opposite end of the catalogue from Brompton. Fashion, broad SKU count, lower average basket, far more sessions per visitor. In 2018 the engineering team wanted to know, under A/B conditions, how small a speed delta actually mattered to revenue. The answer was a hundred milliseconds.

Christoph Luetke Schelhowe and colleagues, writing on the Zalando Engineering blog, 11 June 2018, reported that a 100ms loading-time improvement led to a 0.7% uplift in revenue per session, and that over five months the team had improved overall site speed by 25%, past their original 20% goal. Their summary was deliberately plain: “Shorter loading times go hand in hand with higher revenue per session.”

• Speed delta tested: 100 ms.
• Revenue uplift per session at that delta: 0.7%.
• Overall site-speed improvement across five months: 25% against a 20% target.
• Methodology: A/B test, correlations across the user journey, segmented by device type.

The transferable lesson is not the 0.7% figure, which varies with catalogue economics and session volume. It is the shape of the finding: the relationship between speed and revenue was, in the authors’ phrasing, “very clear and consistent” across the user journey. There was no threshold at which it stopped mattering. The curve kept going as long as the team kept pushing. At Zalando’s scale, 0.7% per 100 milliseconds compounds into real money. At a smaller store, 0.7% is often the difference between a profitable quarter and a flat one, and the arithmetic lands even harder.

What the Three Share, Across Three Very Different Stores

Three brands, three catalogues, three price points, three different ways of measuring. The shared answer is the one that is easy to miss: speed does not improve conversion by letting more shoppers check out faster. It improves conversion by not giving them a reason to doubt that the business behind the page works. The Vodafone data shows this cleanly. The further up the funnel from the sale, the larger the uplift from the same speed change. The visitor was leaving earlier than the operator thought, because they had already failed a trust test the operator did not know was being run.

None of the three brands framed site speed as an SEO problem. Brompton framed it as a premium-brand problem. Vodafone framed it as a KPI for the customer experience. Zalando framed it as a revenue input. All three published figures. None of them hedged with “it depends”, which an agency speed audit usually does when it has nothing specific to recommend.

What This Means Inside a Real Store

Our build section. The reason to read this piece rather than a strategy blog is that we are the people who get handed the ticket after the strategy call, and we want to be straight about where the work actually sits on a WordPress and WooCommerce stack.

A real storefront has three load paths that matter for the trust test, and they are three different problems with three different fixes:

• First paint on category and product pages. This is a hosting and front-end problem. Server location, PHP version, object cache, a sensible CDN in front of the images, a theme that is not shipping 400 KB of JavaScript on the first paint. On a well-provisioned WordPress stack this moves from a four-second LCP into the Vodafone range inside a sprint. The hard part is not any single fix. It is holding the line against the next five plugins the marketing team wants to install.
• The Add to Basket round trip. This is a WooCommerce AJAX call that touches the cart session, which in turn touches the database. On a store with 20,000 SKUs and session objects accumulating in wp_options, this call quietly degrades over months. The fix is not glamorous: a dedicated session table, a review of what the cart carries between pages, a hard look at the plugins that attach to woocommerce_add_to_cart. We have moved this round trip from well over a second into the 200 ms range on real client estates, and it is almost always the single biggest conversion-rate change in a performance pass.
• The checkout confirmation step. This is where Brompton’s ten seconds lived, and it is a payment-gateway timing problem rather than a web-stack one. The round trip between the gateway and the store, the webhooks that confirm it, the order creation that follows, all contend for the user’s patience on the one screen where patience is already thinnest. The fix is operational: pick a gateway that returns inside two seconds under load, move the heavy work off the request thread, and never put a third-party script on the confirmation page that waits on an external domain to resolve.

Roughly what each one costs to ship, as a planning number rather than a quote. A hosting and front-end pass is two to four weeks of focused work on a stack of our size, and it belongs to engineering. The cart round-trip work is one to two weeks, and it belongs to the WooCommerce team because every intervention has to be measured against order flow. The checkout work is a joint piece with the payment gateway and often depends on their roadmap more than ours. None of these is a one-off project. All three need to live inside the acceptance criteria of every subsequent release, because speed decays silently and a tuning pass once a year will not hold the line.

Where I would start, if you are the person reading this and deciding whether to put speed on next quarter’s plan: ask the agency that currently owns your storefront how long the Add to Basket round trip takes on your top five product pages, under production load, with the plugin stack you actually run. The answer tells you more about the state of the operation than any SEO audit will. If the answer is “we would need to look”, that is itself an answer, and we would be glad to look with you.