Design aimed at where people quit

It depends entirely on where your leaks are, and we would rather scope it than quote you folklore. You may have seen the claim that every $1 spent on UX returns $100. It is attributed to Forrester across the internet, but no version of it names a report, a sample or a method, and it appears to be a garbled version of an older software engineering range where $100 was the best case rather than the average. What we can point to is Baymard, which models a 35.26% conversion improvement as available to a large ecommerce site from checkout design alone. That is a modeled ceiling rather than a promise, but it is built on real benchmark data and you can read the working.

Checkout and forms, almost always. Cart abandonment averages just over 70%, and the reasons people give are largely fixable: unexpected extra costs at 39%, forced account creation at 19%, and a checkout that is too long or complicated at 18%. The average US checkout asks for 11.3 form fields when about 8 will do. Cutting fields, showing total cost early and allowing guest checkout are unglamorous changes that tend to move revenue within weeks.

Five per round, in most cases. Nielsen Norman Group’s model puts five users at roughly 85% of usability problems found, and recommends three rounds of five over a single round of fifteen. Large studies mostly re-confirm what the first few sessions already showed. We would rather test three times across a project than run one big study at the end when the budget is spent.

Part of it. WebAIM scanned the top million home pages in February 2026 and found detectable WCAG failures on 95.9% of them, averaging 56.1 errors per page. The top failures are low contrast text at 83.9%, missing alt text at 53.1% and unlabelled form inputs at 51%. Those are design decisions, not deep engineering problems, and they are the same six failure categories that have topped the list for seven straight years. Designing them correctly costs almost nothing. Retrofitting them costs a lot.

Usually pieces, at least to start. A full redesign resets everything you have learned and makes it hard to tell what caused a change in performance. We prefer to fix the highest-value flows first, prove the effect, and let that fund the wider work. If the site is genuinely beyond incremental repair we will say so.

Either. We frequently collaborate with in-house teams, handing over components and specs with accessible states documented. We can also build it, since we do the web development too, which removes the usual gap between what was designed and what shipped.