What Was Measured on This Site

The work on this site was built against the Web Content Accessibility Guidelines (WCAG) 2.2 (opens in a new tab), at Level AA. WCAG 2.2 is a W3C Recommendation, published 5 October 2023 with an update published 12 December 2024, and it defines three levels: A, AA and AAA.

That is a statement about what the work was measured against. It is not a claim that this site conforms to WCAG 2.2, at Level AA or at any other level. Conformance is a claim about every success criterion on every page, tested properly, and that is not what happened here. What happened here is listed below, along with what did not.

All figures below were taken from the 26 pages this site currently publishes, including this one, on the review date above.

Contrast was calculated rather than judged by eye, and it was calculated from the colours the pages actually render, each piece of text against the background it truly sits on, rather than from the palette in the abstract. That distinction matters: the first pass worked from the palette, and a palette says which colours are safe together without saying which ones ended up together. WCAG 2.2 asks for a ratio of at least 4.5 to 1 for normal-size text and at least 3 to 1 for interface parts like a focus outline or the border of a form field.

Every combination that actually appears was then measured on the built pages, resting and on hover: 2,447 pieces of text and 650 hover pairs. The lowest text ratio anywhere on the site is 4.57 to 1, and the lowest interface ratio is 3.47 to 1.

Body copy runs at 9.54 to 1. Headings and bold text run at 16.33 to 1. Small print runs at 5.29 to 1 on the ivory background and 4.77 to 1 on the tinted panels. Links inside prose are 4.57 to 1 on ivory and 4.73 to 1 on a tinted panel. The keyboard focus outline ranges from 3.47 to 1 up to 7.50 to 1 depending on what it sits on. Form field borders are 3.53 to 1.

Measuring the pages this way found three combinations that were failing, and all three were corrected before this page was written: links inside prose on a tinted panel, the colour those links turn on hover, and one heading. The heading was the serious one. It had no colour of its own, so it took the site's near-black default and rendered on a near-black band at 1.03 to 1, which is to say it was invisible to everyone. It is 15.83 to 1 now.

Four more had been found and corrected earlier, from the palette rather than the pages: small print, headings inside tinted panels, the focus outline on deep green, and form field borders. Seven in total, and the split is the point. The palette catches colours that are unsafe together. Only the rendered pages catch colours that turned out to be together.

There are 70 images across the 26 pages. Every one carries alternative text. None is missing it and none carries an empty one. Read the limitation below before taking that figure for more than it is worth: all 70 are the brokerage logo and one portrait, repeated across the site. There are exactly two distinct pieces of alternative text on this entire site, and that is now the whole of the picture rather than a temporary state, because the site carries no photographs at all.

Heading order was checked on all 26 pages. Each page has exactly one top-level heading, and no page skips a heading level anywhere in its structure. Someone moving through a page by its headings gets the structure the page actually has.

Both forms on this site were checked control by control. Every visible field has a label attached to it: four controls and four labels on the contact form, seven and seven on the consultation form.

Required fields are marked with the word "(required)", and the optional field on each form, the phone number in both cases, is marked "(optional)". The marking sits inside the label rather than beside it, which means it is part of what a screen reader reads out along with the field name, not only a visual mark. There is no asterisk to decode and no colour carrying the meaning on its own.

A skip link is present on all 26 pages and moves focus past the header straight to the main content. Focus outlines are visible, and their contrast is in the colour figures above.

The site has a few fade-in animations. If your operating system or browser is set to reduce motion, they are switched off rather than shortened.

The site was loaded with its scripts stripped out entirely and checked at a phone-sized window. The navigation is visible and every link in it works. The question-and-answer sections open and close using the browser's own built-in disclosure control, so the answers are readable with no scripts running at all.

Both of those were genuine failures before this review. With scripts turned off there had been no navigation at all below tablet width, and none of the answers could be opened.

Two things, and neither is a small caveat. They are the reason this page does not claim conformance.

This site has no photographs of properties, places or people, other than one portrait. It carries two images in total, a brokerage logo and that portrait, both with alternative text, both repeated across the pages. The writing is the substance here and the pages are built to work that way.

Until 2026-09-16 this section said something different. There had been twenty-five empty boxes across eighteen pages, each holding the shape of a photograph still to be taken. Those were removed and the pages were closed up around them. The limitation did not go away with them. It changed shape.

What remains true, and it is the part worth being plain about: nothing here has been tested against real photographs with real alternative text, and that is the part of image accessibility that actually takes judgement. Deciding what a photograph of a marsh at low tide needs to say to someone who cannot see it, and whether it needs to say anything at all, is work this site has not done, because it has no photographs to do it on. If photographs are added later, that work starts then and this page is updated with what it finds. The limitation is deferred, not resolved.

No screen reader has been used on this site. Neither has voice control, nor switch access, nor a screen magnifier. Not once, on any page.

Everything described further up was calculated from the site's code or checked against the pages it produces. That kind of checking finds real problems and it found several here, but a calculated contrast ratio and an automated structure check are not the same thing as a person using this site with a screen reader, and they do not stand in for it. Plenty of pages pass every automated check and are still difficult or impossible to use. This one has not been put in front of anyone to find out.

That is the honest state of it, and it is the single biggest gap on this page.

Tell me about it and I will look at it. Email is the better route, since it gives you room to describe the problem and to attach a screenshot if that is easier, but the office line is there if you would rather speak to someone.

Two things make a report something that can actually be acted on: the address of the page you were on and a description of what went wrong. If you know what you were using at the time, a browser, a screen reader, a phone, that helps as well, but the page address and the problem are enough to start.

There is no response time promised on this page, because none has been set, and a number here that nobody intends to keep would be worth less than nothing.