The six criteria
Each company receives a score out of ten on each criterion. The same six scores are used on every page where the company appears.
Live product quality: 30 percent
We use the product itself: sign up where possible, move through the main tasks, trigger errors, resize the screen and try the keyboard. A case study shows a project on its best day. A live product shows every other day too. This criterion carries the most weight because it is the one a reader can check for themselves.
Research evidence: 20 percent
We look for evidence that design decisions came from watching real users, not from internal opinion. Published methods, described recruitment and findings that changed the product all count. Research based only on stakeholder interviews counts for less.
Accessibility, tested: 15 percent
We check what shipped, not what a company’s website promises: keyboard access, labels on form fields, contrast, and how screen readers handle the product. A public commitment to WCAG earns nothing on its own.
Post-handoff coherence: 15 percent
We check whether the design holds up after the agency leaves: whether a documented component system exists, and whether the product still looks and works like one product a few releases later.
UX structural depth: 10 percent
We check whether the product holds together as a whole: clear navigation, consistent patterns across areas, and sensible behaviour at the edges, not just one polished screen. Much of this already shows in the live product, which is why it carries less weight on its own.
Independent signals: 10 percent
Awards, analyst reports and verified client reviews. This carries the least weight because it measures visibility as much as quality, and it disadvantages companies that work in sectors such as government, where few awards exist.
These weights apply on every page of the site. They were last revised in October 2026, when live product quality and research evidence were given more weight and structural depth less.
How the overall score works
The overall score shown on each card is not typed in. It is calculated from the six criterion scores every time a page loads, so it always matches the numbers printed beneath it. Change one criterion, and every page showing that company updates.
Scores are shown to one decimal place. Many companies tie at that level. Showing more decimals would suggest more precision than the assessment has, so we don’t.
How companies are ordered
On the home page, companies are ordered by their overall score. On the four capability pages (product redesign, UX consulting, accessibility audit and usability testing), the order also reflects how well each company’s practice fits that specific service. So a company with a higher overall score can appear lower on a capability page where that service is a smaller part of its work.
The six scores are always printed on every card. A reader who prefers a different order can compare the columns directly, which is why we publish them.
What the scores leave out
Some things can’t be judged from outside a company:
- whether it is easy to work with
- whether it delivers on schedule
- whether its price is good value for your particular project
- which people it will actually assign to you
The last one matters most, so ask for names before you sign.
Where the facts come from
Team size, founding year, location, rates and minimums come from each company’s own website, public records or verified directories such as Clutch, DesignRush and GoodFirms. The sources are listed on each card. Where a company does not publish its rates or minimum, the card shows a dash rather than an estimate.
Reviews and corrections
Each page is reviewed quarterly and shows its own review date, which changes whenever the content changes. Anyone can request a correction through the contact page, including the companies listed. A correction to a criterion score updates every page where that company appears. Companies can also ask to be added through Submit a company.