People arrive at digital products with a goal, limited attention and their own language for the problem. Clear experience design respects that reality by organising information and actions around what users are trying to achieve.
It is worth being blunt about how far the industry is from this. WebAIM's February 2026 scan of the top one million home pages found detectable WCAG 2 A/AA failures on 95.9% of them, averaging 56.1 errors per page — a 10% year-on-year increase that reversed six consecutive years of slow improvement. The average home page now carries 1,437 elements, up 22.5% in a single year (WebAIM Million, 2026).
Pages are getting more complicated faster than they are getting clearer. That is a design problem before it is a technical one, and it sits at the centre of our UI/UX design work.
Begin with real tasks
Research should identify the situations, questions and constraints surrounding a journey. Task-focused insight is more useful than broad personas because it connects directly to content, functionality and prioritisation.
A persona tells you that Sarah is 34 and values efficiency. A task tells you that someone needs to confirm whether a part is compatible with equipment they bought four years ago, on a phone, in a warehouse, with the serial number partly rubbed off. The second is something you can design for.
Practical ways to get there:
- Ask about the last time, not the usual time. "Walk me through the most recent occasion you did this" produces specifics; "how do you normally do this" produces a tidied-up summary.
- Collect the language people actually use. If customers say "shipping" and your navigation says "fulfilment logistics", the search box will tell you eventually — but only after you have lost the ones who did not bother.
- Note where the journey leaves the screen. Phone calls, spreadsheets, WhatsApp messages to a colleague. Those detours are usually where the design has failed.
Make the structure explain itself
Navigation and page hierarchy should help people predict where information lives. Use familiar language, reveal complexity progressively and make the next useful action easy to recognise.
Progressive disclosure is doing most of the work in that sentence. Given the growth in page complexity WebAIM recorded, the instinct to put everything on one screen is measurably winning — and it is the wrong instinct. Complexity should be available on request, not presented as a default.
Three tests we apply to any key screen:
- The five-second test. Show it briefly, then ask what the page is for and what you would do next. If people cannot answer, hierarchy is the problem, not copy.
- The one-decision rule. Most screens should ask the user to make one meaningful decision. Screens that ask for three tend to get abandoned at the second.
- The label test. Read the navigation labels aloud to someone outside the organisation. Internal vocabulary is invisible from the inside and opaque from the outside.
Prototype the difficult moments
Test decisions, forms, permissions, empty states and recovery — not only the happy path. These moments expose whether the product is genuinely understandable and where operational support may be needed.
Forms are the clearest example. Baymard Institute's checkout research found the average checkout contains around 23 form elements against an achievable 12–14, and that mandatory account creation and over-long processes each account for a substantial share of abandonment. Their testing suggests that addressing documented usability problems is worth an average conversion uplift of roughly 35% for large ecommerce sites (Baymard Institute).
Roughly seven in ten carts are abandoned, and the figure has barely shifted in over a decade despite substantial investment in payment technology. The technology was rarely the problem.
Empty states and error recovery deserve the same attention and almost never get it. What does the dashboard look like on day one, before any data exists? What happens when a payment fails at 11pm? These moments carry disproportionate weight because they occur when the user is least confident.
Accessibility improves the core experience
Readable contrast, keyboard support, meaningful labels and resilient layouts help more people use the product and make it more robust for everyone. Accessibility belongs in design and acceptance criteria from the start.
The WebAIM data makes the case efficiently. Six failure categories account for 96% of all detected errors, and they have topped the list for seven consecutive years: low contrast text (found on 83.9% of home pages), missing alternative text (over half), unlabelled form inputs (51%), empty links, empty buttons and missing document language (WebAIM Million, 2026).
None of these are obscure. They are the basics, failing at scale, year after year — and every one of them also degrades the experience for users without disabilities. Low contrast text is harder to read in sunlight. Unlabelled inputs break autofill. Missing alt text breaks the page when images fail to load on a poor connection.
One finding is worth dwelling on: WebAIM detected an average of 133 ARIA attributes per home page, more than six times the 2019 figure, and pages using ARIA had more detected errors on average. That does not mean ARIA causes problems. It does undermine the assumption that a modern component library makes accessibility happen automatically. It does not. It has to be specified, built and tested.
Two practices that make the difference in delivery:
- Put accessibility in acceptance criteria, not in a pre-launch audit. Retro-fitting is more expensive and reliably gets cut when the deadline tightens.
- Test with a keyboard and a screen reader during build, not after. Automated tooling catches perhaps a third of real barriers.
For UK organisations, accessibility is also a legal consideration under the Equality Act 2010, and public sector bodies have specific regulatory obligations. But the commercial case usually lands harder: the same fixes that reduce barriers also reduce friction for everyone.
Build a system, then keep listening
Reusable patterns make quality easier to maintain, but a design system is not a substitute for evidence. Continue observing behaviour and feed what you learn back into components, content and journeys.
A design system's value is that it fixes a problem once. That cuts both ways — an accessibility defect in a shared component propagates everywhere the component is used. Component-level quality standards matter more than page-level ones for exactly this reason.
The listening half is what stops a system calcifying. Useful signals to keep visible:
- Site search queries, especially the ones returning nothing
- Drop-off points in multi-step journeys
- The top five reasons people contact support
- Rage clicks and repeated form resubmissions
- Core Web Vitals in the field, since performance is part of usability
That last point is not incidental. Case studies published through Google's web.dev programme have repeatedly linked performance improvements to commercial outcomes — Rakuten 24 reported a 33% conversion rate increase and 53% higher revenue per visitor after Core Web Vitals work, and Vodafone reported roughly 8% more sales after a 31% improvement in Largest Contentful Paint. A slow interface is an unclear interface, whatever the layout looks like.
What this looks like in practice
For The Salutation Kent, a boutique hotel and restaurant, the design problem was separating three distinct visitor intentions — accommodation, dining and location — without fragmenting the sense of a single property. For Stitched Equestrian, it was making garment customisation legible on the product page, so that a genuinely complex configuration task felt like a series of small, obvious choices.
Where to go next
- To work on journeys, prototypes and accessibility standards, see UI/UX design.
- If the interface work needs to be built as well as designed, see web development or mobile development.
- For deciding what to design before deciding how, read from idea to scalable digital product.
- For keeping design quality maintainable as the organisation grows, read building a brand system for digital growth.
Sources
- WebAIM, The WebAIM Million: 2026 report on the accessibility of the top 1,000,000 home pages — https://webaim.org/projects/million/
- Baymard Institute, Cart Abandonment Rate Statistics and checkout usability research — https://baymard.com/lists/cart-abandonment-rate
- Google web.dev Core Web Vitals case studies (Rakuten 24, Vodafone, redBus) — https://web.dev/case-studies
Useful digital work is a continuous loop: understand, design, build, learn and improve.