Accessibility

What we are aiming for

We aim to meet WCAG 2.1 Level AA. WCAG stands for the Web Content Accessibility Guidelines. It is an international standard for making websites usable by disabled people, and Level AA is the level most organisations aim for.

This page says what we have done, what we know is not right yet, and how to tell us when something stops you.

What this statement covers

This statement covers insights.bysands.com: this page, the front page of the site, and the survey pages you reach from an invitation link.

It does not cover bysands.com, our main company website — that is a separate site, and we have not assessed it here. It also does not cover admin.bysands.com, the private portal our own staff use to build surveys.

How accessible this site is

We believe insights.bysands.com is partially conformant with WCAG 2.1 Level AA. “Partially conformant” means we meet most of the standard, but there are things we know we do not meet, and other things we have not yet checked properly. Both are listed below.

We are not claiming full conformance. We would rather say that plainly than overstate it.

Where we stand legally — and why we are doing this anyway

Bysands Limited is a microenterprise: fewer than ten people, with turnover and balance sheet both under €2 million.

The European Accessibility Act (Directive (EU) 2019/882) exempts microenterprises that provide services from its accessibility requirements. On that basis the Act does not oblige us to meet them, and this page is not a compliance statement.

We are publishing it anyway, and we still aim for WCAG 2.1 Level AA.

Being exempt from a rule is not a reason to build something a person cannot use. Our surveys are sent to people by their employer or their organisation — most did not choose to receive one, and cannot simply go elsewhere if the page does not work for them. That seems to us the wrong situation in which to lean on an exemption.

Two honest caveats. The exemption depends on our size, so it would fall away if we grow. And it covers that Act specifically — it is not a statement about every obligation that might apply to us.

Before you start a survey

Some things about how our surveys work are worth knowing before you begin, particularly if you use assistive technology or expect to need more than one sitting.

• You cannot save and come back later. Your answers are saved each time you move to the next page — not as you type — and this browser will not pick up a part-finished survey afterwards. Please plan to finish in one sitting.
• If you open your link again after starting, you begin a new, separate response. The part-finished one is not resumed and is not removed.
• There is no review step before submitting. Pressing Submit sends your answers, and you cannot change them afterwards.
• We set no time limit of our own, and nothing will move you on. A survey can have a closing date, though — if it has one, we show it before you start. If a survey closes while you are part-way through, your remaining answers cannot be sent, and we will tell you that rather than blaming your connection.
• If you are shown a personal report at the end, it lasts only as long as that browser tab. Reloading keeps it; closing the tab loses it, and reopening your link afterwards will not bring it back. Download the PDF if you want to keep it. This is deliberate — the link may be shared, so holding on to your report for longer would mean the next person to open that link on the same device could read it.
• The survey needs JavaScript. If your browser has it turned off, the questions will not work.
• Surveys do not ask for your name or email. How your answers are reported depends on the survey, and each one tells you before you start.

What we have done

Using it with a keyboard

• On 24 August 2026 someone went through the front page, a whole survey and this page using only a keyboard, with macOS Full Keyboard Access turned on. Everything was reachable, and nothing trapped focus.
• Every control shows a clear outline when you reach it with a keyboard — buttons, the points of a rating scale, and choice options. Added 23 August 2026; before that the outline was missing, which made the survey hard to complete by keyboard even though every control worked.
• If you press Next with a required question unanswered, we take you to that question, rather than leaving you on the button wondering why nothing happened.
• When you move between pages, and when you finish, we move you to the new heading so you are not left at the bottom of a page that has changed underneath you.

Getting around the page

• The front page, this page and the survey pages each begin with a “Skip to content” link. It is hidden until you move keyboard focus to it, it is the first thing you reach when you press Tab, and activating it moves you into the content. An automated test checks the Tab order on the front page at desktop size. That the link stays hidden until it is focused, and that activating it moves you into the content, were confirmed by hand on 24 August 2026 and have no automated test yet.
• Those pages use standard HTML landmarks — a header, a main content area and a footer — so a screen reader’s landmark navigation works.
• The two fallback pages are deliberately plainer, and we would rather say so than let the two claims above sound broader than they are. The “page not found” page has the skip link but no header or footer; the page shown if something breaks on our side has only its main content, a way to try again, and two ways to reach us. Both are stripped back on purpose — they have to render when something else has already failed, so they carry as little as possible.
• Pages declare English as their language.

The survey questions themselves

• Underneath, the questions are the browser’s own form controls. Choice questions are real radio buttons or checkboxes, grouped inside a labelled group with the question text as the group’s label. Short and long answer questions are real text fields with a label attached. On rating-scale and choice questions the visible control is drawn by us, with the real control behind it — so the focus outline is ours to provide, and now is.
• Required questions are marked with the word “Required” — not a colour or an asterisk alone.
• If you leave a required question unanswered, the reason appears in words beside it, linked to the question so a screen reader reads it with the question rather than as loose text.
• Guidance under a question, and any limit on how many answers you may choose, are linked to the question too, so a screen reader can announce them rather than leaving them as text only a sighted person sees. This was wrong until 23 August 2026 and the fix has not yet been confirmed by ear — see the list below.
• There are no pop-up dialogs anywhere in a survey.

Colour, motion and media

• Some text colour contrast is checked automatically rather than by eye. An automated test measures a named list of our text colours against a named list of background colours and requires the WCAG AA minimum. It covers a maintained list, not every combination we ship, and it does not cover semi-transparent colours or text over an image. Borders and outlines are now measured — see the list below for the one that falls short.
• A survey can be shown in one of three colour schemes, and all three are measured. Every text-on-background pairing in every scheme is checked on each run as a complete set, worked out from the list of schemes and the list of roles rather than from a list somebody maintains by hand — so unlike the check above, no pairing here can be missed by being forgotten. Whoever runs a survey picks from those three schemes and cannot enter a colour of their own, precisely because an arbitrary colour cannot be checked in advance.
• If you have asked your device to reduce motion, the progress bar does not animate.
• There is no audio or video anywhere on this site, so captions, transcripts and audio description do not arise. A survey QUESTION page carries at most two images: the Bysands logo, and — where the organisation running the survey has supplied one — that organisation’s own logo. Both carry a text alternative naming the organisation. We accept only PNG, JPEG and WebP files for a supplied logo. The personal report you may see at the end is not bound this way; the next point covers it.
• If the organisation has set one up, you may be shown a personal report when you finish, and that report can contain charts and pictures. Every chart is published as a figure with a written description of what it shows, and the same numbers again as a table a screen reader can read out — so the chart is never the only way to reach its data. Any picture carries whatever text alternative the organisation wrote for it; where they left it blank we treat the picture as decoration and a screen reader skips it. Downloading that report opens a new browser tab, and both the download button and the link say so before you activate them. A link in a report’s text also opens a new tab: a screen reader says so; nothing on screen does. See below for what we have not checked about this.
• Nothing autoplays, loops, or moves as you scroll, and nothing navigates you away on a timer.
• We do not disable pinch-zoom or cap the zoom level on any page. A second automated check confirms our main pages still fit a 320-pixel-wide window without sideways scrolling, which is also what zooming in produces — including the end-of-survey page when it is showing a personal report, whose charts are laid out for a printed page and are re-flowed to a single column on a narrow screen. Nobody haslooked at those pages at that size — see below.

What is not right yet

We would rather publish an honest list than a short one.

Problems we know about

1. Every survey page has the same browser tab title, “Survey”. If you have several tabs open, the title will not tell you which is which.
2. In Windows High Contrast mode, text answer boxes may show no focus outline. The controls we fixed in August use an outline that mode repaints automatically; the text fields still use an older technique that it can remove.
3. One button border is very slightly too faint. We have now measured borders and field outlines. Field borders pass. The border on outlined buttons falls just below the required level while you hover over it. It is recorded and on our list; nothing depends on seeing that border to use the page.
4. Helper text under a question may not be read out. Found by our first screen-reader test and fixed the same day; it is listed here because the fix has not yet itself been confirmed by ear. The symptom was concrete: on a question where you may choose up to two answers, someone using a screen reader was never told the limit existed.
5. A very long word or web address can make a personal report scroll sideways on a narrow screen, if it is in a comment or in the title of an Actions section. Text, narrative, picture captions and the cover wrap it; those two do not yet.

About the surveys themselves

• Survey questions are written by the organisation that commissioned the survey, not by us. We do not review their wording, so we cannot promise a question is clearly or simply worded.
• Every page declares English. If a survey is written in another language, a screen reader will still read it with an English voice.
• A question can be published with no wording at all, which would leave a form field with no name for a screen reader. Nothing currently prevents that.
• The same applies to a report’s pictures. The description attached to a picture in a personal report is written by the organisation, not by us, and we do not review it. A picture left with no description is treated as decoration and skipped, which is right when it is decoration and wrong when it is not — and nothing currently tells the two apart.
• And to a report’s text. Its headings, lists and links are written by the organisation, not by us. We make sure a link can only go to a web or email address, but whether its words say where it goes, and whether a heading describes what follows, is theirs.

Things we have not tested

• The personal report and the PDF it produces have been listened to once — with VoiceOver on an iPhone on 1 October 2026, by one member of our team — and everything in them was announced and could be used. That was before a report’s text could hold headings, lists and links, which no screen-reader pass has heard yet. That is one pass on one platform: in other screen readers and other PDF readers, how a chart’s description and table come across is what the code does and what our automated checks confirm, not something we have heard. The PDF carries its headings and each chart’s description as real structure rather than a flat picture of a page, which our checks confirm by reading the file back. We would particularly like to hear from you if you use one.
• Using a screen reader, a chart’s numbers come up twice — once in the figure’s description and again in the table behind it. That is deliberate: one description has to serve the printed PDF, where the table cannot follow. It means more to listen to, not less information, but we would rather say so than let you discover it. The table is there for screen readers only — if you can see the chart but find it hard to read, you have the chart and its caption and nothing else, and we would like to hear from you if that is you.
• We have tested four times with a screen reader — VoiceOver on a Mac on 23 and 24 August 2026, and VoiceOver on an iPhone on 1 October 2026, the last two against this live site. All four were by one member of our team. The survey could be completed from start to finish each time, and on the iPhone that included the personal report and its PDF. We have not tested with TalkBack on Android, or on a tablet, and because a survey invitation is usually opened on a phone, that is now the gap we are least comfortable with. We have not tested with NVDA or JAWS (the screen readers most people use on Windows), and not with voice control, switch access or magnification. If you use any of these we would particularly like to hear from you.
• We have not had an independent accessibility audit.
• We have not checked zoom and reflow by eye. An automated check confirms the home page, this page, the survey itself, and the pages you see when a link does not work all fit a 320-pixel-wide window without sideways scrolling — which also covers zooming in. That tests reflow automatically, but not by eye: text squeezed inside a box, or two controls overlapping, would still fit the window and still be wrong.
• Our browser coverage is narrow. Automated checks run in desktop Chrome and in an emulated iPhone. We do not test Firefox, real physical devices, or tablet-sized screens.
• Automated tools only find part of the problem. Many of the requirements cannot be checked by a machine at all — whether a description is meaningful, whether reading order makes sense, whether keyboard focus moves in a logical order. The helper text problem above is a real example: our automated checks passed and the problem was still there.

How we checked

This statement is based on self-assessment by our own development team. It has not been checked by anyone outside Bysands.

Four manual screen-reader passes have been done (VoiceOver on a Mac on 23 and 24 August 2026, and on an iPhone on 1 October 2026), following a written procedure we keep so it can be repeated and compared. Everything else is automated: accessibility checks using axe-core, driven by an automated browser, at two screen sizes and in two browser engines, plus the colour measurements described above.

Two limits worth knowing. First, those automated checks are run by our developers before a release — they do not run automatically when we publish a change. Second, of the seventeen checks, four run every time — the four pages you can reach directly; the other thirteen need a deployed test environment, because they need a real survey to answer. Before a release our process now fails rather than quietly skipping those if that environment is missing, but nothing forces a developer to run the checks at all.

Tell us about a problem

If something on this site stops you, please email [email protected] with “Accessibility” in the subject line.

It helps if you can tell us:

• the address of the page, or just “the survey link from [the organisation that sent it]”
• what you were trying to do, and what happened
• what you use to browse — a screen reader, keyboard only, magnification, voice control — if you are happy to say

What happens next:

• We will acknowledge your message within three working days.
• We will tell you what we can do and, where we can, when.
• If you cannot use the survey at all, say so. We will find another way for you to take part. You should not miss out because our site does not work for you.

If you are not happy with our reply

If we do not respond, or you are not satisfied with our response, you can ask for it to be escalated to Stephen Clements, Director, at [email protected].

About this statement

• Prepared on: 23 August 2026
• Last reviewed on: 6 November 2026
• Prepared by: self-assessment, by the Bysands development team. No external audit has been carried out.

We will review this page at least once a year, and whenever we make a significant change to insights.bysands.com. When we fix something on the list above, we will remove it from the list.