Most "convert this link" tools look the same in their happy path: there's a field, a button and a spinner. They differ in what happens when something goes wrong. This post looks at the UX decisions behind the public behaviour of Qteros, a free, no-account web tool that turns one public YouTube video into an MP3 (192–320 kbps) or an MP4 (up to 1080p). Everything here comes from the site's public documentation. It's about interface and error design, not about how media is fetched.
Quick legal note before anything else: only download content you own or have the right to use. Good error handling doesn't change copyright.
Decision 1: Validate in the browser, before any request
The input box accepts many URL shapes: youtube.com/watch, youtu.be short links, /shorts/, Live URLs and music.youtube.com. Those shapes all share one useful thing: an 11-character video ID.
So the first step happens on the client: parse the ID out of the pasted string before anything is sent. That has three benefits:
- Instant feedback. A playlist page, a channel page, a search-results URL or a link shortener has no single video ID. The field can say so immediately ("that does not look like a YouTube video link") instead of after a round trip.
- Fewer wasted requests. Every malformed paste that's stopped locally is one less call against a rate-limited relay.
- A clear rule. The site states the limit outright: one video per link. Playlists aren't half-supported, because they aren't supported at all, and the field explains that.
An illustrative sketch of the idea (not the site's code):
// Pull an 11-char ID from common YouTube URL shapes; null if none. function extractVideoId(input) { const m = input.match( /(?:v=|youtu\.be\/|\/shorts\/|\/live\/|\/embed\/)([A-Za-z0-9_-]{11})/ ); return m ? m[1] : null; } Text that isn't a URL gets a different treatment. If the field holds plain words, it switches modes: the MP3/MP4 buttons are replaced by a Search button. Search returns up to 20 result cards, each with a thumbnail, duration, title, channel, its own MP3/MP4 buttons and a Preview that uses the official YouTube embed. Searching never converts anything. A result is only a candidate until you press one of its buttons.
Decision 2: Disable impossible actions in advance
There are two length limits: MP3 up to 150 minutes and MP4 up to 30. Because search results already know the duration, the card can grey out MP4 for a 45-minute video before the user tries it. An error the user never has to see is better than a well-written one.
If someone pastes a long video directly and picks MP4, the conversion still stops, but with a specific code (4.215, explained below), and the status line offers an MP3 button straight away. The error message includes the fix.
Decision 3: Name the stages instead of showing a spinner
A spinner tells you that something is happening. A staged progress bar tells you what is happening. In the UI, the bar moves through checking, extracting and converting, and the public docs number four pipeline stages:
| Stage | What's happening |
|---|---|
| 1 | Authorisation: the relay requests a session key |
| 2 | Setup: the key is exchanged for a one-use conversion address |
| 3 | Conversion: the video is checked and extracted |
| 4 | Progress: the finished file is polled and handed over |
Naming stages helps both sides. Users can see the job is moving. When it fails, the stage number already narrows the cause. A failure at stage 1 or 2 on every video points to the service. A failure at stage 3 on one video points to that video.
Decision 4: Make error codes in the form stage.reason
Failures read like Conversion stopped — code 4.215. The first number is the stage and the second is the reason. Each code links to a plain-language next step:
| Code | Meaning | Next step |
|---|---|---|
4.215 | Too long for the chosen format (MP4 > 30 min) | Use MP3 (up to 150 min) |
4.645 | Not served to the converter's region | Try a different upload |
429 | Rate limited | Wait about a minute, retry |
403 | Upstream refused | Retry once; else another video |
502 | No usable upstream answer | Retry shortly |
net | Browser lost the relay connection | Check connection, keep tab in foreground |
2 (progress) | Signed link expired | Convert again, click Download promptly |
1.0 | Authorisation failed | Reload; if constant, service is down |
The docs sort reasons into three groups: retry now, try a different video, or switch format. That classification matters more than the numbers. Most users don't want to know what happened. They want to know what to do next.
The docs also list things that aren't errors, which saves a lot of support mail: Safari asking where to save a file, or an MP3 with no picture.
Decision 5: Treat 429 as a normal state
Any free, no-signup tool needs rate limiting, and the relay answers 429 when too many conversions arrive in a short time. The UX choice is to say so plainly and give a time: wait about a minute, then press the format button again. That's the whole fix.
Some practical notes for anyone designing something similar:
- Publish the limit. The Qteros limits page lists the rate limit next to the other limits, so a 429 matches something the user may already have read.
- Don't auto-retry quickly in a loop. That turns one 429 into many. Let the user decide when to try again, or back off generously.
- Local validation helps here too. Every bad paste caught in the browser is a request that never counts against the limit.
Decision 6: Short-lived links, stated up front
Finished files are handed over through a signed download link that expires on purpose. That's sensible on the server side, but it can confuse people who wander off and come back. So the UI tells users to press Download as soon as it appears, and the expired-link case has its own code with a one-line fix: convert again.
Similarly, the docs warn desktop users to keep the tab in the foreground, because a suspended background tab can pause the progress polling. Writing that down is cheaper than debugging it in a support thread.
Decision 7: No account at all
There's no sign-up, email or password. That removes a whole category of UX (onboarding, password resets, consent screens) and forces the core flow to stand on its own: paste, choose, save. The same page works in Safari on iPhone, Chrome on Android and any desktop browser.
Takeaways
- Parse and validate on the client before the first request.
- Grey out actions you already know will fail.
- Name pipeline stages, and put the stage in the error code.
- Map every code to a next step, not just a description.
- Treat rate limits and expiring links as documented states, not surprises.
The full public error reference is at qteros.com, and it's a useful example to borrow from even if you're building something unrelated.