Why QR codes stop working: five causes, in triage order
Five things account for almost every "our QR code stopped working" report: the destination died, the print lost its quiet zone or shrank, contrast or glare defeated the camera, the payload was malformed from the first print, or the code was physically damaged or covered. The first step is always the same — scan with two different phones and read the decoded text. If both phones decode the same string, the code is fine and the destination is broken. If neither can decode, the problem is on paper.
Triage before you reprint anything
Reprinting is expensive and, in three of the five cases, pointless: the new code will fail in exactly the same way. Spend two minutes on diagnosis first.
| What you observe | Most likely cause | Fix |
|---|---|---|
| Camera shows the link, browser shows an error page | Destination failure | Restore the URL or redirect; the print is fine |
| Camera cannot find the code at all | Quiet zone, size or contrast | Reprint with margin and contrast corrected |
| Phone joins nothing after a WiFi scan | Payload escaping or security type | Regenerate the payload, verify the readout |
| Works close up, fails at arm's length | Modules too small | Bigger print or shorter payload |
| Worked last month, now nothing | Damage, obstruction or laminate | Inspect the physical copy |
1. The destination behind the code is gone
This is the most common failure and the least visible. A static code pointing at a page that was moved, deleted or renamed still scans perfectly — the phone just lands on a 404. A dynamic code adds more ways to break: the redirect provider's domain lapsed, the subscription ended, the short link was retired, or the vendor changed their API and the redirect stopped resolving.
Related variants that look like "the code broke":
- An expired TLS certificate. The link resolves, but the browser shows a full-page security warning. Most people turn around at that point.
- A redirect chain that ends in a loop after a site migration.
- A destination that now requires a login or has geo-restricted the visitor.
- A domain flagged as malicious. If a shortener's domain gets blocklisted, every code using it dies at once, and users see a warning instead of your page.
This is the structural argument for static codes on permanent assets, covered in static versus dynamic: a fixed URL only breaks when the thing it points at breaks, which is at least a failure you control.
2. The layout ate the quiet zone or the scale
A code that scans beautifully in the generator can fail the moment it enters a design file. The three usual injuries:
- Cropping to the symbol edge. Someone trims the white border to make the code align with a text column, and the finder patterns now touch artwork. Detection at an angle collapses.
- Scaling down to fit. A code rendered at 600 px and placed at 18 mm wide ends up with modules around 0.3 mm — below the practical print floor.
- Resampling. Rotating or scaling a raster QR in a layout tool introduces interpolation. Modules get soft edges, and soft edges are what decoders measure.
The fix is to place the vector export and keep the four-module margin as part of the placed asset. Details and the module-size maths are in print size and quiet zone.
3. Contrast, colour and reflections
A scanner needs the difference between dark and light modules, in the lighting where the code actually lives. Three failures dominate:
- Brand colours that are too close in value. Pink on orange and mid-grey on white look distinct on a monitor and merge in a photograph.
- Inverted codes. Light modules on a dark field work on some phones and not others, because finder-pattern detection expects dark squares on a light background.
- Specular reflection. Gloss laminate, glass cabinets and phone screens produce a bright patch that wipes out a corner of the symbol. Codes behind glass frequently fail at exactly the angle people naturally hold their phone.
Matte stock, a dark foreground on a light background, and a contrast ratio well above 3:1 solve all three. The generator flags anything below 3:1 before you download; treat that warning as a blocker for print, not a suggestion.
4. The payload was wrong from the first print
These codes scan fine and still do nothing useful, which makes them the hardest to spot. The scanner decodes a string that the phone cannot act on:
- Unescaped separators in a WiFi payload. A semicolon in the SSID ends the field early, so the phone tries to join a network that does not exist. See the guest WiFi guide.
- A missing double semicolon at the end of the payload, which some parsers treat as an incomplete record.
- vCard line endings converted to LF, collapsing the record into one unparseable line — the subject of the vCard guide.
- A truncated URL because the payload hit the capacity ceiling and the encoder silently dropped the tail instead of erroring.
- Characters encoded as Latin-1 instead of UTF-8, so an accented SSID or a Chinese name decodes to mojibake.
Every one of these is visible in the payload readout before printing. That single habit — read the encoded string, not just the picture — prevents most of this category.
5. The physical copy degraded or was covered
Paper and vinyl do not age gracefully. Floor decals get walked on, warehouse labels get scuffed, window stickers collect dust on the outside and condensation on the inside. A code with level M error correction tolerates roughly 15% damage; a scuff across a finder pattern is often worse than that percentage suggests, because the decoder needs those three corners intact.
Then there is the simplest cause of all: something was placed over the code. A new poster on top of the old one, a product label across the corner, a sticker advertising a promotion. When a code "suddenly" stops working in a fixed location, walk over and look at it.
What changed without you touching the print
Sometimes the paper is identical and the world moved. Operating system updates change how camera apps treat detected links, browsers tighten their handling of plain HTTP, and blocklists change. A code printed three years ago pointing at an HTTP URL may now trigger a warning interstitial that nobody wants to click through.
None of that is repairable after printing. It is, however, a reason to encode an HTTPS URL from the start, and a reason to review the destinations of long-lived printed codes once a year — the same way you would review a phone number printed on a sign.
Preventing the next failure
- Keep the exact payload in a text file next to the artwork, so a reprint needs no guesswork.
- Prefer static codes for anything that must outlive a vendor.
- Print one spare copy and store it flat; it is the fastest way to prove whether a problem is the code or the environment.
- Put the destination somewhere you control — your own domain — rather than a third-party shortener.
- Test the printed proof with two phones before the run, and again after the first hundred come off the press.
Frequently asked questions
Can a QR code expire?
The symbol itself cannot — it is ink on paper and has no clock. A dynamic code expires in practice when the redirect behind it stops being paid for or the provider disappears. That is why the phrase "expiring QR code" almost always refers to the service, not the code.
Why does it scan on my phone but not on my colleague's?
Camera quality, autofocus behaviour, the scanning app and the lighting all differ. Older phones with slower autofocus struggle most with small modules and low contrast. If one device fails consistently, treat the code as borderline rather than blaming the phone.
Can a damaged code be repaired?
Only within its error-correction budget, and only if the three finder patterns and the timing lines survive. Cleaning a scuffed label sometimes helps; a torn corner never does. Once a finder pattern is gone, no amount of error correction will find the symbol.
Generate a replacement code and check the payload readout before you print — the cheapest insurance against every failure in this list.
Written by the QRQuick team. Last updated November 19, 2026.