The compliment trap
Beautiful sites that do not earn are the industry's speciality. Everyone involved gets something: the agency gets a portfolio piece, the founder gets compliments, the designer gets a dribbble shot. The business gets a cost.
I have been the paying client five times - five products of my own. That changes how you look at builds. When it is your money, "it looks premium" stops being an outcome.
One number, decided first
Every build I take on starts with the same question, before any design: which number should this move? Qualified enquiries per month. Completed checkouts. Bookings. Newsletter-to-customer conversion. Pick ONE, agree how it is measured, and suddenly every decision has a referee. Should the homepage open with a video? Depends - does it move the number? Should we add a configurator? Same referee.
What this kills, mercifully
The one-number rule kills the features that exist for internal politics: the slider with six director-approved slides, the "About our values" carousel, the animation that demos well in the boardroom and adds two seconds of load on a phone. It also kills endless subjective debates - the number does not care whose taste wins.
What it looks like in practice
The funnel is: visitors → leads → buyers → revenue. Every page has exactly one job inside it. Measurement is set up before launch, not after - first-party analytics, honest attribution, and a monthly look at where people actually drop. Then you iterate on the weakest step instead of redesigning what already works.
If your current site wins compliments and loses money, the diagnosis is one call: bring the number it was supposed to move. If nobody ever chose one - that IS the diagnosis.