Almost every small business has documents that are meant to go on the website: the price list, the menu, the membership form for the club, a patient information sheet, a technical data sheet, directions, an intake form for new clients. The usual route is quickly described: upload the file, add a link, done. And that is exactly where the three problems start that create work later on. The content is hard to read on a phone, it cannot be changed without the original program, and sooner or later an indexed file shows a price that no longer applies. This article reverses the order. First comes the decision whether a document needs to be a download at all. Then the question of what a PDF looks like when it does its job, and what belongs in the link text in front of it so that people know what they are clicking.
Decide first: download or page?
Whether a document belongs on the website as a file can be settled with three checks. Is it going to be printed, because someone pins it to a board, takes it into the workshop or leaves it at the reception desk? Is it going to be signed, because a consent, an order or a membership declaration depends on it? Is it going to be archived, because a dated version has to be verifiable later? If at least one of these three checks applies, a download makes sense. If none applies, the same content belongs on the site as a page. It is then readable on a phone, searchable, editable at any time and considerably easier to keep accessible.
In practice this question is rarely asked, because the route to a PDF is so short. The document already exists in a word processor, the export takes a single click, and the result looks the same on every screen. That very property, the fixed layout, is the core of the problem: a document set on A4 stays set on A4, including on a screen six centimetres wide. Making the decision deliberately saves you from maintaining two truths later on, and from the follow-up questions of people who could not read the document. If you are rethinking your web presence in general, the companion article on when a new website really pays off sets out the groundwork.
Is it printed?
Notices, workshop sheets, forms to take away, a menu for the display case: as soon as paper is the destination, a document with a fixed page format has a real purpose. Printing is then the point, not a side effect.
Is it signed?
Membership declarations, direct debit mandates, treatment contracts, client intake forms: where a signature is needed, you need a document that can be filled in and signed. An online form next to it saves many people the detour.
Is it archived?
Data sheets, schedules of services, annual reports, minutes: where a dated version has to remain traceable, a file is the right form. The version date then belongs visibly inside the document and in the file name.
Most documents sitting in download areas on business websites today do not pass this test. They are read, not printed, not signed and not filed. The overview below lists the most common cases in small firms and the form that fits them. Several rows deliberately name both: a page for reading and a file for printing.
| Document | Why people look for it | Form that fits |
|---|---|---|
| Price list | Orientation before enquiring, usually on a phone | Pricing page as a page, PDF in addition for printing in the office |
| Menu | Choosing before the visit, often on the move | Menu as a page including allergen information, PDF only for the display case |
| Club membership form | Filling in and signing | PDF for printing, plus an online form for the fast route |
| Patient information sheet | Reading before and after the appointment, sometimes taking it home | Page with the full text, tagged PDF as a second form |
| Technical data sheet | Comparing, passing on, filing in a project folder | PDF with version date and number, short summary as a page |
| Directions and parking | Retrieved on the move, shortly before arrival | Page with address, notes and a map link, no download |
| Client intake form | Filling in, signing, bringing along | PDF to complete, plus a protected form for secure submission |
The double form is the normal case
What a page does better than a file
The most important difference is reflow. A web page rearranges its text as soon as space gets tight; a document set to a fixed page size does not. The European standard states a clear requirement for this: content has to be presentable without loss of information and without scrolling in two dimensions, for vertical scrolling at a width equivalent to 320 CSS pixels (EN 301 549, clause 10) and for horizontal scrolling at a height equivalent to 256 CSS pixels (EN 301 549, clause 10). Clause 10 applies this explicitly to documents that are not part of a web page. An A4 PDF usually fails it: it forces zooming and dragging in both directions. How that plays into general usability is covered in the article on the website on a phone.
Then there is weight. A page requested on a phone weighs a median of 2,362 kilobytes (HTTP Archive Web Almanac), measured in July 2025 across a corpus of 17.2 million (HTTP Archive Web Almanac) websites; ten years earlier the figure was 845 kilobytes (HTTP Archive Web Almanac), an increase of 202.8 percent (HTTP Archive Web Almanac). Against that background, a scanned information sheet weighing 20 megabytes is far out of proportion: it equals roughly eight complete page loads. On a mobile network that means waiting time, on a data plan it means consumption, and both hit precisely the moment when someone is close to making an enquiry. Why lean delivery pays off elsewhere too is described in the article on PageSpeed and static delivery.
The third point is findability. PDF is among the file types that search engines process (Google Search Central), and indexed documents appear in result lists just like ordinary pages. Two limitations matter here: images inside PDF files are not indexed (Google Search Central), so a scanned document stays invisible in terms of content. And PDF files are refreshed less frequently than HTML pages, because their content is assumed to be stable (Google Search Central). A corrected price list therefore takes longer until the correction shows up in results.
- Readability: text on a page reflows, can be enlarged and read aloud by a screen reader without an additional program.
- Maintenance: a correction is entered in one place and is visible everywhere at once, with no export, upload and re-linking.
- Findability: pages work with headings, a description and structured data, and are re-crawled more often than files.
- Connection: from a page, links lead on to the contact form, the service page or the next explanation; from a file, there is rarely a way back.
- Multiple languages: a second language version is manageable as a page, whereas each language as a file means a separate document.
- Accessibility: the technical foundations exist on the web; in a document, structure, language and alt text have to be produced separately.
The scanned notice
If PDF, then tagged
A PDF is first of all just a description of which character sits where on a page. Whether that turns into a heading, a paragraph, a table cell or a caption follows from the visual design for people and from so-called tags for software. Without those tags, the document remains an image made of letters: a screen reader may read across columns, a search finds nothing, and converting it to another format produces chaos. Tagged means that the structure travels with the file, and that is precisely the difference between a document that merely looks good and one that can be used.
Accessibility is most effective when it is taken into account while the document is being created.
The technical reference for implementation is PDF/UA, set out in the ISO 14289 series. Part one describes tagged PDF on the basis of the earlier file version; part two, ISO 14289-2, was published in March 2024 (ISO 14289) and refers to PDF 2.0. The requirements for accessible documents themselves are stated in EN 301 549, primarily in clause 10 (Bundesfachstelle Barrierefreiheit). They take the familiar success criteria and transfer them to documents: the page becomes the document, the page title becomes the document title. The six points below cover most of what goes wrong in practice.
Structure instead of looks
Headings, lists and tables are created through styles, not through larger type or tab stops. That structure moves into the document tags on export and carries outline, read-aloud order and navigation.
Set the document title
The standard requires documents to have titles that describe topic or purpose (EN 301 549, clause 10.2.4.2). The title lives in the metadata, not only in the first line, and later shows up in the window and in bookmarks.
Declare the language
The default human language of a document has to be programmatically determinable (EN 301 549, clause 10.3.1.1). Without it, a screen reader may read German text with English pronunciation.
Check the reading order
The order in which content is read aloud has to follow meaning (EN 301 549, clause 10.1.3.2). With multi-column layouts, margin notes and info boxes, the technical order often differs from the visual one.
Provide alt text
Non-text content requires a text alternative (EN 301 549, clause 10.1.1.1). Logos, charts and plans need a description; purely decorative elements are marked as such and skipped.
Contrast and type size
Grey footnotes on a light background and seven-point type fail the same contrast requirements as on the web. If the text is also published as a page, the values used there are a good guide.
The effort stays manageable when the structure is already right in the source document. The Bundesfachstelle Barrierefreiheit points out that basic elements such as headings created with styles and alt text for images are worth having even when not all criteria are met yet (Bundesfachstelle Barrierefreiheit). For most small firms that means: do not aim for the perfect document, change the order of work instead. First set headings and alt text in the word processor, then export, then check. How to write alt text that genuinely helps is covered in the article on images, formats and alt text.
- Headings in the source document formatted as headings, not as bold body text.
- Tables built as real tables with a header row, not simulated with spaces or tab stops.
- Alt text assigned for images, charts and plans, decorative elements marked accordingly.
- Document title and language set in the metadata and verified after export.
- Reading order spot-checked, particularly on multi-column pages and in info boxes.
- Form fields labelled so it is clear what goes into which field.
- File size checked and images inside the document brought down to a sensible resolution.
The legal frame in brief
The German Accessibility Strengthening Act has applied since 28 June 2025 (Bundesfachstelle Barrierefreiheit). Among other things it covers services in electronic commerce, meaning services provided through websites or apps at the individual request of a consumer with a view to concluding a consumer contract (Bundesfachstelle Barrierefreiheit). Micro-enterprises with fewer than ten employees (Bundesfachstelle Barrierefreiheit) and at most 2 million euros (Bundesfachstelle Barrierefreiheit) in annual turnover are exempt when they provide services; the exemption does not extend to products. What follows from that for your own site is set out in detail in the article on the Accessibility Act and business websites.
One passage is particularly instructive for downloads. Clause 10 of EN 301 549 applies explicitly to documents that are not web pages, that are not embedded in web pages and that are provided with a web page without being rendered together with it; the standard calls them, in brackets, downloadable documents (EN 301 549, clause 10). The argument that a PDF sits outside the website and is therefore a world of its own does not hold up technically. In addition, documents protected by signatures, encryption, password protection or watermarks fall under the same requirements as soon as they are presented to users (EN 301 549, clause 10).
Orientation, not legal advice
File name, link text and file size
The second big lever is not in the document but in front of it: in the link. The web content accessibility guidelines require, at the lowest conformance level, that the purpose of each link can be determined from the link text alone or from the link text together with its programmatically determined context (W3C WCAG 2.2, success criterion 2.4.4, level A). Clause 10.2.4.4 applies the same requirement to non-web documents (EN 301 549, clause 10). In reality this is a frequent weak spot: ambiguous link text was found on 15.2 percent (WebAIM Million) of the home pages analysed, with an average of 5.3 (WebAIM Million) occurrences on affected pages, and 46.3 percent (WebAIM Million) of pages contained links with no discernible text at all. Overall, 95.9 percent (WebAIM Million) of home pages showed detected failures against the guidelines.
For downloads that means three additional pieces of information directly in the visible text: which document this is, what format it comes in and how large the file is. Adding the version date answers the most frequent follow-up question at the same time. A meaningless reference turns into an announcement that leaves nobody surprised.
| Frequently seen | Better | What the difference does |
|---|---|---|
| Click here | Price list, PDF, 480 KB, version 03/2026 | The purpose is clear without surrounding text and stays clear in a list of links |
| Download | Membership application form, PDF, 210 KB | Several downloads on one page become a distinguishable set of choices |
| More info | Appointment preparation sheet, PDF, 340 KB | Expectation matches what actually opens |
| File | Data sheet type 4, PDF, 1.2 MB, version 01/2026 | Large files are chosen deliberately instead of loaded by accident |
| Form.pdf | Application for membership 2026, PDF, 180 KB | The name in the link is understandable, including later in a downloads folder |
A building block for the download list
The file name is often overlooked, even though it becomes visible twice: in the address bar and later in the downloads folder. “document_final_v3_NEW.pdf” is an annoyance there, “cleaning-price-list-2026-03.pdf” is not. What works well: lower case, hyphens instead of spaces, no special characters, a descriptive core and the version date as year and month at the end. If you need version numbers, append them in the same place. The file name then doubles as the answer to the question of which version someone is looking at.
Swap versions, remove the old ones
Documents on websites age more quietly than pages, and that is what makes them risky. A PDF gets indexed, opened directly and passed on without anyone ever seeing the surrounding page. Someone who finds an old price list through search sees the figures from back then, not the note on the current pricing page. Since PDF files are refreshed less frequently than HTML pages (Google Search Central), such a version stays in circulation longer than a business would like. And because price statements have to be accurate under competition law, an outdated file is more than a blemish; the connections are covered in the article on price statements and advertising claims.
The remedy is unspectacular and has worked the same way for years: one stable address per document, holding the current version. The version date sits in the document, in the link text and in the file name, not in the address. If the address does change, for instance during a rebuild, a permanent redirect belongs with it so that saved links and search results keep working. How that is planned at a larger scale is shown in the article on a relaunch without losing rankings. If you are weighing up whether to handle such tasks yourself or hand them over, the comparison of builder, agency and AI sets out the trade-offs.
- Define one fixed address per document that holds the current version.
- Replace the file at that same address instead of placing a second file next to it.
- Carry version date and number in the document, the link text and the file name, not in the address.
- Delete replaced files, even when no page links to them any more.
- After the swap, check whether old versions are still discoverable through search.
- Set up a permanent redirect for changed addresses and verify it briefly.
- Put one fixed date in the year in the calendar for reviewing all downloads.
Two truths is one too many
In day-to-day operation, document upkeep therefore belongs in the same place as the other recurring tasks. Which those are, and at what rhythm they make sense, is summarised in the article on website maintenance for small firms.
A form in front? Only for real value
A widespread reflex is to put downloads behind a form in order to collect contact details. For most of the documents discussed here that is the wrong call. Price lists, menus, directions, opening hours and information sheets answer questions people want settled before getting in touch. Building a hurdle at that point loses exactly the people who are close to enquiring, and returns addresses from people who wanted something else.
The trade only makes sense where genuine value is created: a substantial planning aid, a template that saves work, an analysis that is not available elsewhere. The same standard then applies as for any other consent: name plainly what follows, and obtain agreement properly. The legal requirements are covered in depth in the article on newsletter signup with double opt-in; for the download area, the rule of thumb below is enough.
- Freely accessible: prices, menus, information sheets, forms, directions, opening hours, data sheets.
- Worth weighing up: substantial guides, calculation templates, checklists with real work value.
- If a form, keep it short: name and email address are enough; every extra mandatory field costs completions.
- Transparent: before submission, state what happens with the data and whether further contact follows.
Managing downloads in XICflow
In XICflow, documents are part of media management rather than a loose attachment. Every file carries a title, a version date and the size determined by the system; the download block outputs format and size automatically in the link text, so those details do not have to be maintained by hand. When a new version is uploaded, the address stays the same and the date at the link updates with it. On request, the AI assistant produces an HTML version of the same document as a page, so that reading and printing get separate routes without two texts having to be maintained. What else is included is shown in the feature overview; the costs are listed in the overview of plans and credits.
Document with a date
Title, version date and description live with the document itself. The same information is therefore visible everywhere on the site, even when one document is linked from several pages.
Download block
The block outputs label, format and file size together and presents several documents as a clear list instead of hiding them inside body text.
Replace a version
A new version replaces the old one at the same address. Saved links stay valid, and the old version disappears from the site instead of lingering as a second truth.
For converting an existing download area, a fixed order of work has proven itself. It can be worked through in an afternoon and answers the question of which files should stay at all along the way.
- List all existing files with address, size, version date and the page that links to them.
- Apply the three checks to each file: printed, signed, archived.
- Anything that passes none of them becomes a page, and the download is removed or kept as a second form.
- Recreate the remaining documents from the source format, with headings, alt text, title and language.
- Replace scanned files or run text recognition on them and then add structure.
- Standardise file names: lower case, hyphens, a descriptive core, version date at the end.
- Rewrite link texts and add label, format, size and version date.
- Fix the addresses, delete replaced versions and redirect changed addresses permanently.
- Spot-check on a phone: can the document be read without dragging in two directions?
- Put a review date in the calendar and work through the list once a year.
The benefit shows up faster than the effort suggests. Content available as a page is read more often and appears in search results with a description that fits the business. Documents that genuinely need to be files are opened with a clear expectation: knowing that a file is 480 kilobytes and carries the date 03/2026, people click deliberately. And maintenance gets shorter, because a correction happens in one place instead of three. For firms with many information sheets and forms, it is worth looking at the related articles on the medical practice website, the club website and the restaurant website with a menu, where the typical documents appear in their own context.
The question behind the question
Sources and studies