How QR codes work

Updated 2026-07-05 · The Ever-QR team

A QR code looks like noise. It is actually one of the most disciplined data formats in daily use — every pixel block has a job, and understanding the jobs explains almost everything practical: why codes survive damage, why long links make dense patterns, and why the white margin is not decorative.

The anatomy, in five parts

• The three big corner squares are *finder patterns*. The scanner hunts for their distinctive dark-light-dark ratio first — that is how it finds the code in a busy photo, at any rotation, in a fraction of a second.
• The smaller square near the fourth corner is an *alignment pattern* (bigger codes have several): it corrects perspective distortion when the code is photographed at an angle or on a curved surface.
• The timing lines — the alternating dots connecting the finder squares — give the scanner the module grid's exact rhythm.
• The quiet zone, the clear margin around everything, tells the scanner where the code *ends*. It is required by the standard, and cropping it is a leading cause of scan failures.
• Everything else is data — your content plus its insurance policy, laid out in a precise zigzag.

Modules, versions, and why link length matters

The small squares are *modules* — the bits. The standard defines 40 sizes ("versions") from 21×21 to 177×177 modules; the generator picks the smallest that fits your data. Practical consequence: a 30-character URL yields a sparse, forgiving pattern; a 200-character tracking-parameter monster yields a dense grid of tiny modules that needs more printed size to stay readable. Shortening your link literally simplifies the physical object.

Error correction: the insurance built into the pattern

Alongside your data, the code stores redundancy using Reed-Solomon codes — the same mathematics that let scratched CDs play. Four levels exist: L (~7% recoverable), M (~15%), Q (~25%) and H (~30%). Choose H for anything printed: it is what lets a scuffed, sun-faded or partially covered code still reconstruct its message — and it is the *entire* mechanism that makes center logos possible: the logo deliberately covers modules, spending the reserve. Which is why a logo without maximum correction is a code living on borrowed math.

What a QR code is not

It is not encrypted, not "smart", and not alive: it is printed storage. It cannot update itself, phone home, or expire — anything that expires is a server someone put between the code and the destination, not the code. And it holds whatever it was given at creation, typos included, forever: hence the verification routine before printing.

The practical takeaways

Short link → simpler pattern. Correction level H → damage and logo insurance. Quiet zone → non-negotiable. Direct destination → nothing to expire. Everything else on this site is these four sentences, applied.

Frequently asked questions