A website is a product: build it like one

The big-bang redesign is the most expensive way to improve a website: months of work, launched on opinions, followed by silence. Treating the site as a product means small, measured changes shipped continuously — each one tested against the metric it is meant to move.

Loading speed affects the experience before a visitor interacts with the page. Performance work can reduce friction, but its effect on conversion depends on the site, audience and task. Measure the result instead of assuming a universal uplift for every millisecond removed.

A visitor decides whether to trust a site in seconds. Typography, spacing and motion quality are not aesthetics — they are the trust signal that decides whether your copy is ever read.

Write a sentence describing who the site serves and what a successful visit looks like. For a consultancy, that might be a prospective customer understanding the relevant service and sending a suitable project enquiry. For a store, it might be finding the correct product and completing a purchase. The design, content and measurement should follow that job rather than compete with it.

Map the decisions between arrival and completion. Does the visitor understand the offer? Can they find evidence relevant to their situation? Is the next step clear and proportionate? A beautiful first screen cannot compensate for missing service scope or an unusable form. Prioritise the whole path, including confirmation and follow-up, instead of treating the homepage as the entire product.

Start with the assets required to understand the page. Compress images, provide appropriate sizes and avoid downloading every video at once. Keep visual dimensions stable so content does not jump as media arrives. A background film should not delay the main message or consume resources needed for an interaction.

For a portfolio, low-resolution source video should be framed at a size that preserves its usefulness rather than enlarged to fill every screen. Use motion to orient attention, not force constant movement while someone reads. Respect reduced-motion preferences and verify that the page remains understandable when animation is reduced or unavailable.

Lab tools help diagnose problems under controlled conditions. Real-user measurements reveal experiences across devices, connections and visits. Use both where data is available. Core Web Vitals describe loading, responsiveness and visual stability; they are not a substitute for measuring whether the visitor successfully completes the intended task.

Connect the measurement to a clear conversion definition, such as a completed contact request rather than a click on the opening button. Inspect abandonment between steps and the quality of the enquiries received. Faster pages can support a better experience, but a specific conversion improvement must be measured on the actual site, not borrowed from someone else's case study.

Collect problems from user feedback, analytics and direct testing. Write each proposed change with a hypothesis and an acceptance check. For example, explaining the service boundary before the form may help visitors decide whether the project is a fit. Improving an error message may help a person complete the request after entering an invalid address.

Prioritise repeated barriers on important paths over decorative polish with no clear purpose. Release small changes that can be verified, and document what changed so future analysis has context. My product and web work combines build quality, conversion-focused design and analytics so the site can keep improving after launch instead of waiting for another full redesign.