ApiaryActiveLive
Try: pause · settings · learn · wipe
← Community / Reading Room
ST
craft · 24 min read

Show the Filename After the Download Button

A museum visitor opens a public resource page and clicks “Download.” The browser finishes some activity and the page displays a crisp success mark. Later the…

AI disclosure: Drafted with AI assistance for Apiary. All files, filenames, reader tasks and test traces are fictional. No deployed downloads were audited, and no outside files were retrieved for the examples.

A museum visitor opens a public resource page and clicks “Download.” The browser finishes some activity and the page displays a crisp success mark. Later the visitor finds two files named Guide and one named Guide with an appended number. One opens last year's map. The visitor doesn't know whether the intended current guide was downloaded, whether clicking a second time made a duplicate, or which of the files came from the museum at all. The website successfully communicated that a button was used. It didn't leave enough information to recover the document the person came for.

A useful download experience has several distinct facts. The page can name the approved artifact and describe the choice before activation. It can state the offered filename and edition. The browser or receiving tool may choose an actual local filename and may know whether saving finished. Most web pages shouldn't claim to know that location or final name without evidence. Separating these facts lets a person find the right document manually and lets a maintainer test the real behavior without turning a simple download into a mystery.

This is a guide for ordinary authorized public documents such as community schedules, exhibition guides and reviewed worksheets. It is not a method to bypass download restrictions, inspect users' devices, or recover undisclosed private material. Use fictional files in a local or appropriately authorized preview. Where the interface has no way to observe a save or open event, say exactly what it did observe.

Define the artifact before you design the button

Start with the document's intent. A “museum map” could mean a directional diagram for several floors, an annotated route for a particular tour, or a dated exhibition guide with the same name as an earlier event. Give the artifact a title, meaningful edition or date where relevant, an approved exact underlying identity, and whatever delivery choices readers actually need. A publisher who can't distinguish the worksheet from the installation instructions will struggle to make a useful download row even if both are technically served without error.

A file's display name can differ from its stored source identity. A human title such as “East Wing level map” says what it is. An offered filename like East-Wing-map-edition-4.pdf helps the person recognize it in a local list. A protected internal reference might be useful to the publication system but shouldn't be exposed just to make a receipt more precise. For a public-facing design, align readable document title, format, and edition with the authorized artifact. Keep any internal locator or checksum within its proper scope, and use a documented identity where further verification is needed.

The action label should describe the request the person is making. “Download East Wing map” is an understandable starting point where the control does request a downloadable file. “Open current East Wing map” might be more accurate if the product simply navigates to a public document view, which the browser may render before the person chooses to save it. Don't use a stronger verb because it looks more decisive on a button. The actual experience may involve an approved link, a choice of formats, or a request submitted to a dedicated export workflow. Name that action, and test whether the next state follows it.

Scope size and format information to what is known. A PDF tag beside a link is helpful only if it belongs to the offered version. A size label can be valuable if the publisher actually knows the relevant file size and it changes the user's choice. A broad estimate from an older edition should not be copied into every row as if independently checked. A release receipt can distinguish “format and offered name verified” from “size not tested,” rather than forcing each box to be marked complete when the job never examined it.

Make a distinct accessible action and a legible result

The W3C Accessible Name and Description Computation specification describes how names and descriptions are computed for accessibility APIs. It doesn't validate the actual words chosen for a particular page. Test the action's name in context, not only the decorative download symbol and title that surround it. A row with three adjacent links all reading “Download” may be comprehensible visually to a designer who remembers the row order. A person navigating sequentially or using a supported assistive technology list could need the target and its format clearly distinguished.

Place the most useful descriptive information before or with the action. Someone may need to decide between a PDF and a text version, or between the current guide and a dated appendix. A short readable label and description can carry those distinctions. More detail after activation can be useful too. “Offered file: East-Wing-map-edition-4.pdf” belongs close enough that a reader can recover it when switching to their browser's ordinary download controls. Don't make a temporary message the only place the name exists if the product can also keep it in readable context. Respect the user's task before asking them to memorize transient interface text.

Use a carefully scoped status update if there is one. “Download requested” can be accurate after an activation led to the supported request. “File saved as …” requires stronger evidence about local completion and name than most websites have. “Your file” followed by the offered filename can be ambiguous if the surrounding copy implies the app knows exactly where it landed. Prefer direct language: “Offered filename: East-Wing-map-edition-4.pdf. Your browser may use a different local name.” You don't have to overexplain ordinary saving. You do have to avoid confidence that the available evidence does not support.

Check what happens to focus and announcements when that status changes. A short notice can be useful without moving a person away from the controls they chose. If an after-action component currently steals focus for every request, consider whether it forces an unnecessary extra step. A visible result can remain in the document, and a supported accessible status can report an appropriate state. Test meaningful feedback with the actual named environment instead of asserting a string is usable merely because it was added somewhere in the DOM. A stable accessible name matters before activation; a limited status receipt matters after it. They serve different jobs.

Don't remove the ordinary route to the information once the action ends. Someone may press Download and have their browser prompt for a location. When they return to the page, the visible resource title, version and offered filename can reassure them that the correct thing was requested. Replacing the entire row with a green check and no identity loses the very details a person might want to compare against the local copy. Keep the page useful without making it pretend to be a filesystem manager.

Distinguish a suggested name from a local saved name

RFC 6266 defines HTTP Content-Disposition use and includes guidance on filename information. The filename value is advisory to recipients, not a promise of what a user's local browser will choose. The WHATWG HTML Standard's downloading resources section describes how HTML download requests and possible filenames relate to user agent processing. These documents help maintainers understand how a site can offer a name. They do not establish that every browser, device, delegated application, or saving workflow retains it exactly.

Separate the app's evidence into three fields. “Published file title” answers what content is intended. “Offered filename” names what the publisher supplies or intends to supply where supported. “Actual local name observed” is a fact only when collected appropriately in a specific controlled test or through a legitimately permitted product feature. In ordinary public copy you can often omit the third. A page shouldn't infer it from the offered name after a click, and it shouldn't imply the user owes you the details of their local downloads. A test can use a designated safe local output and compare results without looking at unrelated files.

Duplicate local names can arise for ordinary reasons. A browser may apply a suffix to avoid overwriting an existing file, and a reader may rename the file after saving it. Those choices don't change the underlying approved artifact. Conversely two different editions delivered under an identical suggested name can make it hard to decide which belongs to today's instructions. Use meaningful edition or date in the offered filename when it fits the publisher's release practice. Nevertheless show edition in the actual document or credible context as well. A suffix imposed by a local tool is not an approved new edition indicator.

Use readable file names without promising uniform handling of every character. If a name is localized or includes punctuation, test the documented delivery and appropriate receiving environments with synthetic values. A displayed title can carry full text even if a conservative offered filename uses a practical representation. Avoid asserting a universal normalization or rewriting a valid human title merely because one platform saves a slightly different local string. If the exact file identity matters, an authorized visible edition or reference inside the artifact is stronger evidence than the exact spelling of a chosen local name.

The same reasoning applies to format changes. A browser could open a readable document rather than saving it under the predicted name. Another context could open a system sheet asking what to do with it. Those are valid observations about the tested behavior, not permission to fabricate a receipt. Check the actual action label and expected experience in the supported product. If the intended route is to open, provide choice text accordingly. If the feature is a delivery request under known conditions, state what occurred without extending it to a successfully saved file.

Work a case: two versions have the same friendly title

Imagine the fictional Harbor Rooms museum maintains a current self-guided route and an older reference version. On the site both have been called Room guide for years. The current page now lets someone choose Current route guide, edition 4, or Older guide for last spring, edition 3. The publisher approves distinct offered names and visible edition context. Some visitors previously downloaded a file called Rooms Guide, and their local lists may contain it. The migration is intended to make future retrieval easier, not prove that every old file can be recognized once detached from all approved context.

The first controlled check exercises the site. Does each description identify the correct edition and kind of visit? Does the ordinary action request the intended synthetic file? Does its visible offered name match the planned artifact? A browser tool can inspect the activation behavior in an authorized environment. Verify actual file content, where permitted in the local fixture, with a known unique edition title rather than assuming the name is sufficient. Download button correctness and artifact correctness are separate. The site could offer edition 4 by name and send the old content from the same route.

Observe local duplicate behavior only in your controlled context, not in an actual reader's folder. Put a benign test copy with the same synthetic filename in the designated clean area and follow the normal supported test. If the tool saves a separate named copy, record the observed difference. If it asks a user whether to replace the earlier file, do not accept any destructive choice outside the particular test conditions. The public description need only explain that local names can differ and remind readers to choose by actual edition or content. A receipt “offered filename matches; application added suffix in named test” helps explain the variance without promising it happens everywhere.

Now change a different factor. Suppose the museum introduces a separate exhibit route guide for evening tours, but its plain title looks similar and one staff member copies the wrong offer into the daytime page. The filenames could be genuinely different without signaling why each belongs to its audience. Ask a test reader to choose the right document for the daytime route and identify what makes it safe to recognize after download. If they choose by which name best matches the generic site title, the visible offer hasn't done enough. A phrase like “Daytime self-guided route, edition 4” may resolve the task better than a longer string of technical details.

A release receipt can distinguish the successes. “Both approved synthetic editions requested from their corresponding rows; visible offer names checked; edition detail confirmed by opening designated fixture copies. One supported browser added a local suffix to a duplicate. No production file saved or shared.” That statement is more useful than a blanket “download works” because it separates the results a reader is likely to question later. It also provides the maintainer one reliable replay if a new site template reuses the old generic display title.

Work a different case: the request starts but saving isn't established

Consider a volunteer downloading a fictional community reading list. An action triggers the correct public request, but the person's browser asks whether to save or open it somewhere else. They close the prompt and return to the site. The page has already displayed “Saved to Downloads,” even though it has no observation of the browser's final choice. Someone interpreting the sign as proof may look in the wrong place, or submit several requests to try to recover a file that never existed. The interface doesn't need more confidence. It needs to say which part of the transfer it can actually verify.

In the controlled example, give the browser an authorized test file and avoid interacting with unrelated prompts. Observe whether the page offers the right filename and information at the moment of request. If the available web component only knows that it initiated the supported route, report that. The browser's ordinary UI may provide its own state. Where the product has a documented trustworthy completion event for a specific workflow, test it under its own contract separately. Don't assume that support for exporting a file means the website can inspect an arbitrary local save location. Treat what a person chose and what the website attempted as independent.

Plan a useful response to uncertain status. The page can keep the offered name and relevant version visible and offer the normal action again where authorized. A user-facing hint can say to consult the browser's own download controls, without implying a precise button label shared by every device. If the first request may still be in progress, acknowledge that the website hasn't established whether saving finished. Don't present the possible duplicate as a sure failure or hidden success. A clear view of what was requested can be enough for the visitor to decide whether to wait, retry, or inspect the browser's own reported result.

A different partially observed failure might show an upload-style progress indicator from the site's preparation step. Completion of document preparation is meaningful, but isn't proof of saving. If the server creates a file for a requested export, that job can finish while delivery still needs to be selected. Say “Public copy ready” within scope rather than “Saved.” Check that the actionable next step actually refers to the approved output. A reader who has been told a file exists shouldn't have to deduce its title from a short internal job label and then search a folder. This is a contract between experience states, not an argument for adding more animation.

A test can simulate three benign outcomes with a designated local browser: opening a supported public fixture, stopping before a save decision, and inspecting successful saves only under approved test conditions. Compare the site's statement with each observed state. If the wording is always “saved,” at least one state provides a clean falsification. If it says only “link requested” when it actually has a well-supported success stage, the designer can consider a better scoped message. The appropriate precision depends on reliable event evidence, not on a desire to make the template sound helpful. This scenario differs from the duplicate issue: the file may never have reached local storage in the first place.

Provide retrieval help without taking control of the user's files

A website can offer a small help line: “Look in your browser's own downloads list for the document title or offered filename.” It may link to its supported product help or show how to recognize the edition once opened. It shouldn't ask for a crop of unrelated files when the user only needs an approved title. If somebody thinks they saved a copy under a different name, tell them which content or edition to look for rather than implying your application can list their local files. A simple note can help when the ordinary browser interface has a different name or location on another platform.

Keep alternatives inside the authorized public scope. If the intended PDF is unavailable, a reviewed text page or a clearly described alternative can be useful where actually provided. It's not acceptable to expose a different restricted copy or suggest changing a URL to guess another artifact. If one public route was intentionally retired, explain the verified approved option and its relationship to the older item. A visible working artifact should not silently be replaced by a different resource whose name sounds almost the same. For a date-bound guide, name whether the current replacement or an archived edition is appropriate.

What if a recipient has only the printed content, not the filename? A meaningful title and edition inside the artifact support retrieval and verification alongside the source page. The two records need not duplicate every detail. A public worksheet can display its revision and linked source without exposing private storage location. A person who opens several local copies can compare those credible labels. If the present copy doesn't contain enough information to distinguish it, tell them the actual limitation and point to the approved current resource. Do not make file recovery rely exclusively on a suffix the local tool can change.

A larger support workflow may distinguish current proof from history. The team can retain an approved record of an edition offered on a certain date. It doesn't need a reader's personal saved list to answer whether that edition is now superseded. Conversely a report of “the item in my desktop folder says edition 3” tells the team something about a reader's copy, but not necessarily what the live web page offered today. Keep the stories separate. This lets support assist the stated goal without claiming wider access or inferring more than the visitor chose to share.

Test the requested artifact, not just a download-shaped action

Build a simple fixture matrix. A valid current file should have correct purpose, format, offered filename and content identity. A distinct archived or alternative format should have its own accurate label. An intentionally unavailable public item should not offer a generic button whose target is unverified. A supported no-completion observation should leave the page's state limited. One task can test successive returns with a harmless duplicate while another can test ordinary navigation and visible context. Avoid treating one late screenshot as the evidence for all of them.

Use independent expected artifacts. If the application declares a filename and the test checks it against the same generated value, it can pass even when the value came from the wrong edition. A reviewed expected item can instead specify what label, date, document first lines and delivery reference belong together in the fixture. Test both initial and follow-on state, including an ordinary change of selection if there are two versions. Unsupported browsers and accessibility combinations belong in the untested portion of the receipt, not in a broad “accessible” label attached to one observed result.

Review interruption without altering network behavior outside the allowed environment. A local benign fixture can model a failed request or a blocked unsupported action and show an honest state. This is not a reason to probe someone else's service or repeatedly retry an endpoint. The useful question is whether the experience leaves a person enough truthful information to try the approved route again or take a named alternative. If the page cannot verify a particular stage, don't assign it an automatic error or success when the request ends. The limitations are part of the workflow you promised.

For keyboard or assistive-technology review, focus on both purpose and outcome. Can someone identify the file before acting? Can they return to the relevant controls afterward? Does feedback communicate a request without implying completion, and is the offered name still available? Observe with a named supported combination where feasible. A token announcing “button” in the accessibility tree does not verify the visible filename. A successful local file check does not establish the reader could discover which action to use. Pair these checks deliberately and state which ones passed.

Check format choices as separate public contracts

A downloadable handout can have equivalent formats for different audiences. It might offer a printable PDF and a plain text version of the same approved instructions. The files need not look alike. They should not silently refer to different editions if the page presents them as alternatives for the current version. Before implementing both buttons, decide which facts identify the common content and which differences the publisher intends. A text edition may omit a decorative mark, but a substantive location change in the PDF should have an equivalent approved rendering in the other edition if the project promises that. A brief source identity is more useful to a reader comparing copies than a detailed account of which compiler emitted each file.

Consider a dedicated format choice with three deliberately distinct fields: name of the available format, identity of the publication, and offered filename corresponding to that particular item. Do not infer the intended file from the suffix alone, since a site can accidentally pair misleading metadata with a different document. Choose representative synthetic material that distinguishes both expected format and meaning. In a controlled preview, the printable file might show “Room B, edition 4,” and the permitted text alternative can express the same authorized fact in its own format. A local variant that still says Room A after the review is more than a file naming fault. The reader has received a different answer. Call each observed failure by the part of the contract it breaks.

A file can open in a useful browser preview and still fail a “save to a particular name and retrieve it later” expectation. Check what is actually promised for each delivery option. Some editors may intentionally present a web reading view before offering a download. A person does not need the same interactive steps to read both. The purpose is knowing which route should display content and which should ask a user-controlled save interface. Give those options matching truthful labels. Use a known approved test case rather than claiming the application's DOM or a filename suffix can substitute for observation of the actual receiving tool.

When moving links between pages, keep the approved version pointers tied to the audience. A landing section may offer the latest guide; a course page might need a specific worksheet for an earlier cohort. The edition field prevents the same active-looking link from being copied everywhere. Conflicts should go through the content owner, not be hidden under a rename to “new-format.pdf.” Verify the downloadable file has the anticipated identifying content where the preview allows it. The smallest independent check is often reading one official edition phrase in each artifact after visiting the route. That's good evidence about the tested output and no claim about files outside its scope.

A supported printer version can have a separate name even though it belongs to the same textual edition. Document whether it changes page geometry, carries a separate accessibility review, or is simply another delivery format. Don't present every version as automatically usable for every person. A minimal release table can say “Printable PDF and reading-text file both identify edition 4; text and page geometry assessed separately.” Recognition and usability are complementary. This work earns trust by naming what it verified rather than promising more from the directory labels alone.

Choose what the page should retain after a person returns

The person who starts a download may immediately navigate away, switch applications, or open a system dialog. Later they may see the same web page in a new or restored state. You do not need to reconstruct all actions to provide useful context. Retain ordinary visible information about the publication and offered version based on the page's approved content. If your app has acknowledged a request, it can show a concise scoped notice without promising a completed save. A generic “you already have this file” state may deprive a reader of the next useful action when nothing on the site can establish that fact.

Think through what a user should be able to recover if the transient notification is gone. The public page can still show what material it offers. A complex app may also have a properly authorized list of generated exports. These are different capabilities. A generated-export record might accurately show that a public test copy was prepared on a particular day, but could be wrong if it implies everyone who opened the page later requested that copy. A simple viewing page should not start tracking private local actions just for the sake of a more confident message. Decide which evidence the operation actually needs and what scope is appropriate.

Try a small scenario with one saved fictional guide and one new request. The old guide can remain identified even if a redesign changes the current button position. The new request can show a different offered name if the publisher approved a new edition. Neither action tells the site which was manually opened later. A reader can compare the dates or approved labels in the actual document, not just pick whichever rows the browser lists first. If the tool chose a duplicate suffix, the visible name can be a clue, but the edition inside the file is the more reliable support. Name those two kinds of evidence in the troubleshooting copy so the user doesn't get sent in circles.

Do not keep a misleading badge merely because it appeared in an earlier user journey. If the app reloads and has no retained evidence of the result, an ordinary idle state and visible offered filename can be entirely appropriate. A staff member helping remotely can see the same approved row and describe what the user should expect from that route. This is particularly helpful where users have different browsers or default save locations. It is not an authorization for staff to inspect arbitrary files or solicit irrelevant personal details. The workflow can remain lightweight while being more honest than a detailed but invented completion history.

Review changes to the offered name before publishing them widely

A short filename can appear on public documentation, a support email, and the file itself. A publisher might change the naming convention to help identify language or version. Test what the revised name helps and where it creates unnecessary discontinuity. If the visible artifact remains the same edition, don't call a purely cosmetic name change a new release. If the revision is substantive, don't reuse a deceptive title just to preserve a convenient old saved name. A short change note can describe actual version and offered naming separately. This will help colleagues explain why an earlier public file has different characters but matches the intended content.

Check one migration in a harmless fixture. A site begins offering “Waterfront-walk-summary-edition-2.pdf” instead of “guide-final.pdf,” with the same approved publication wording and an independent copy of the expected artifact. The visitor can identify the edition more easily in the new save list. An old browser might still keep the former file name from a previous save. Do not read that as a defect in the revised offer until its source and actual release are established. Record which file was requested in the test, what filename was presented where observable, and what the artifact says. The next maintainer won't have to guess whether the title, filename or underlying content changed.

If the filename update crosses locales, choose a supported review rather than removing meaningful characters arbitrarily. A typographical preference in one language can be kept in the human title even if the recommended offered name is managed differently. The local destination may have its own restrictions or choices, so validate recognized identity instead of promising character-perfect filenames in every context. Verify that the website's descriptive label doesn't become misleading when the offered name lacks a copied accent or uses a short form. Good public retrieval evidence should still point to the proper document.

Leave a small reusable release receipt

The release record can say “Current trail guide, PDF, edition 4; offered name checked in approved preview. Test fixture delivered the intended file. Actual reader save location is not known to the site. Accessible label and retry copy reviewed in the named environment; other contexts not tested.” Arrange it so a maintainer can tell which item to check after a later update. If a new edition changes only release date, recheck that the visible name, internal artifact and downloaded content all refer to it. If the site changes how saving is initiated, retest wording and feedback rather than blindly reusing the same old confirmation copy.

Measure success as the reader's ability to recover the exact useful thing. A green symbol can remain if its meaning is constrained and the reader keeps enough context. An unverified local name need not block truthful product communication. The right answer might be an ordinary link, accurate label, visible offered filename, and room for a browser to do its own job. Say when you don't know where the actual copy went instead of writing certainty into a message you cannot justify.

Where an export is intentionally replaced after a mistake, a visitor revisiting the public page might benefit from a specific correction line. “Updated map, edition 5; the previous offered file showed the former entrance” is a content decision if that's what actually changed. It does not require removing anyone's old copy or suggesting the website already did so. An older file can remain on a person's device indefinitely. The new version needs publicly distinguishable evidence, not a forced update claim. Verify the revised notice and authorized artifact independently before issuing reassurance.

A small support exercise can ask someone to recover the new guide without giving them a browser history or saved file list. Let them read the authorized page and identify the offer, then show a designated fixture with two appropriately labeled editions. If the available clues identify the intended file even when the names look similar, the revision communication did useful work. If neither document states what it contains, start by fixing the approved identity rather than asking people to choose by file size.

Primary sources

Frequently asked
What is Show the Filename After the Download Button about?
A museum visitor opens a public resource page and clicks “Download.” The browser finishes some activity and the page displays a crisp success mark. Later the…
What should you know about define the artifact before you design the button?
Start with the document's intent. A “museum map” could mean a directional diagram for several floors, an annotated route for a particular tour, or a dated exhibition guide with the same name as an earlier event. Give the artifact a title, meaningful edition or date where relevant, an approved exact underlying…
What should you know about make a distinct accessible action and a legible result?
The W3C Accessible Name and Description Computation specification describes how names and descriptions are computed for accessibility APIs. It doesn't validate the actual words chosen for a particular page. Test the action's name in context, not only the decorative download symbol and title that surround it. A row…
What should you know about distinguish a suggested name from a local saved name?
RFC 6266 defines HTTP Content-Disposition use and includes guidance on filename information. The filename value is advisory to recipients, not a promise of what a user's local browser will choose. The WHATWG HTML Standard's downloading resources section describes how HTML download requests and possible filenames…
What should you know about work a case: two versions have the same friendly title?
Imagine the fictional Harbor Rooms museum maintains a current self-guided route and an older reference version. On the site both have been called Room guide for years. The current page now lets someone choose Current route guide, edition 4, or Older guide for last spring, edition 3. The publisher approves distinct…
References & sources
  1. Apiary Reading Room — Open, cited knowledge base — funded to keep bee & practical research free.
From the Apiary Reading Room. Opinion & editorial — not financial advice. We don't overclaim.
More from the Reading Room