01 - Software development

Websites that load faster than a visitor can notice.

I design and build front ends for companies that treat speed as a requirement, not an extra. Every project ends with a measurable result: Core Web Vitals in the green, a smaller bundle, a shorter time to first interaction.

Fig. 01 - how a page paints, frame by frame illustration, not a measurement Five frames of a page loading between 0.2 and 1.0 seconds. The largest element paints at 0.8 seconds and the page is complete at 1.0 seconds.

02 - Services

Performance work, from a one-off audit to a budget that holds.

  • S.01

    Performance audit and optimisation

    Profiling of the existing site and identification of bottlenecks in rendering, images, fonts and JavaScript. You get a report listing every change, ordered by its impact on the result.

  • S.02

    Sites and stores from scratch

    Static or server-rendered builds with minimal JavaScript. The architecture follows the content and the scale, not the current framework trend.

  • S.03

    Migration off heavy platforms

    Moving a site from a bloated CMS or site builder to a lightweight stack, keeping the content, the URLs and the search rankings intact.

  • S.04

    Monitoring and maintenance

    Continuous measurement from real devices, performance budgets in CI, and an alert whenever a metric crosses the agreed threshold.

03 - What we measure

Core Web Vitals are the three metrics Google uses to judge how well a site performs. They affect search rankings and how many visitors stay long enough to see the page.

Three numbers that decide whether a site feels fast.

LCP Largest Contentful Paint
How long it takes before the visitor sees the main content: the headline, the product image, the first screen. The usual culprits are uncompressed images and render-blocking scripts.
< 2.5 s
INP Interaction to Next Paint
How quickly the page responds to a click or a keystroke. A high INP usually means too much JavaScript running at once.
< 200 ms
CLS Cumulative Layout Shift
Whether content jumps around while loading. A shifting layout causes mis-clicks and makes the site feel unfinished.
< 0.1

This site, measured

Google Lighthouse, mobile Run a live test - opens in a new tab

Performance
100
Accessibility
100
Best practices
100
SEO
100

New: AI visibility

I also prepare the site so AI models can find it.

More and more people look for services through AI assistants rather than a list of search results. To appear in those answers, a site’s content has to be readable by systems that never run JavaScript and read the document, not the design.

  • A.01

    Content in HTML, not in a script

    Server-side rendering and a semantic document structure, so the full content can be read without executing any code.

  • A.02

    Structured data

    Schema.org markup for services, products, reviews and contact details, so the facts about the business are unambiguous.

  • A.03

    Access for AI crawlers

    robots.txt and llms.txt configured deliberately: which crawlers get in, and which parts of the site they may read.

  • A.04

    Content ready to be quoted

    Clear headings, concise answers and FAQ sections a model can quote from directly.

04 - What it is worth

The calculation uses published results from other companies’ sites. It is an order of magnitude, not a forecast for yours - every figure below is linked to its source.

What a faster site is worth, in money.

Speed is easy to agree with and hard to fund, because it rarely arrives with a number attached. So here is the number, using the arithmetic and the studies I would use in an audit.

Your numbers

Revenue today, per month
€108,000
Estimated conversion uplift
5.2% – 10.4%
Additional revenue, per month
€5,641 – €11,281
Additional revenue, per year
€67,687 – €135,373
How this is calculated

The site improves from the LCP you entered to 2.5 s, Google’s threshold for “good”. That is expressed as a relative improvement: 4.2 s down to 2.5 s is a 40% better LCP.

That relative improvement is multiplied by the ratio Vodafone measured in an A/B test published by Google: a 31% better LCP produced 8% more sales, so roughly 0.26% more sales for each 1% of LCP improvement. The upper figure uses that ratio directly; the lower figure halves it, as a deliberate margin - it is not a second study.

The ratio is used rather than the more famous Deloitte figure on purpose. Deloitte measured a 0.1 s improvement, and dividing that into a per-second rate implies +84% conversions for one second, which is not a real effect. A ratio of relative changes carries its own diminishing returns instead: 6 s to 3 s is a 50% improvement, while 3 s to 2.5 s is only 17%.

Effects on average order value are left out, although Deloitte measured those too. Including them would roughly double the result. This is the conservative half of the evidence.

These are correlations from other companies’ sites, not a promise about yours. Speed is one input among many, and no calculator can separate it from the rest. Use the figure to decide whether the work is worth scoping - then measure your own site, which is what the audit is for.

Sources

  1. An A/B test on Vodafone Italy: the variant with a 31% better LCP produced 8% more sales, a 15% better lead-to-visit rate and an 11% better cart-to-visit rate. This ratio is the basis of the calculation above.

    Google, web.dev case studies, 2021 - opens in a new tab

  2. A 0.1 s improvement in mobile load time raised retail conversions by 8.4% and average order value by 9.2%; for travel sites, conversions rose 10.1%. Drawn from over 30 million user sessions across 37 brands.

    Deloitte and 55, commissioned by Google - “Milliseconds Make Millions”, 2020 - opens in a new tab

  3. Nykaa improved LCP by 40% and saw 28% more organic traffic. NIKKEI STYLE improved LCP by 18% and saw 9% more page views per session. Cdiscount reported a 6% revenue uplift over Black Friday.

    Google, Core Web Vitals business impact case studies, 2024 - opens in a new tab

  4. Google states plainly that “Core Web Vitals are used by our ranking systems” - so the effect of speed is not only on the visitors you already have, but on how many arrive.

    Google Search Central documentation, current - opens in a new tab

05 - Process

How a project runs, from first measurement to a budget that holds.

  1. Measure

    Field data and lab data together. We fix the starting point so the effect of the work can be proven.

  2. Prioritise

    A list of changes ranked by impact against effort. What the visitor sees comes first.

  3. Implement

    Changes ship in stages, each measured on its own. No full rewrite unless it is genuinely necessary.

  4. Performance budget

    Limits written into the build process, so the site does not slow down again with the next release.

06 - Questions

Questions I get asked before a project starts.

How long does an audit take?

Usually two to three business days. You get a document with the measurements, the causes and a list of changes in the order they should ship.

Do I have to rebuild the whole site?

No. In most cases a handful of specific changes carry the biggest gain: images, fonts, script loading order. I only propose a rewrite when the existing architecture makes a good result impossible.

Do you work with an existing team?

Yes. I can work as a subcontractor, or as technical support for a marketing team without its own front-end capacity.

How do you charge?

The audit is free, implementation is a fixed price per scope, maintenance runs on a monthly retainer.

07 - Contact

Send me your URL. You get back a report with measurements and a list of fixes.

The audit is free and comes with no obligation. I reply within one business day.

I use these details only to reply to your message.