QR Codes: How They Store Data in Squares

A QR code packs up to 4,296 characters into a tiny grid of black and white squares. But how does a camera decode those squares back into a URL or a Wi-Fi password? The answer involves clever binary encoding, built-in error correction, and a layout designed so scanners can read codes at any angle — even when partially damaged.

Anatomy of a QR code

Every QR code contains several structural elements that help scanners locate and orient the code before reading any data:

1. Finder patterns — The three large squares in the corners (top-left, top-right, bottom-left). These tell the scanner “a QR code starts here” and establish its orientation. The ratio of black-white-black widths (1:1:3:1:1) is unique enough to be detected at any angle.
2. Timing patterns — Alternating black and white modules running between the finder patterns. These help the scanner determine the exact grid spacing.
3. Alignment patterns — Smaller squares placed in larger QR codes to correct for perspective distortion (like scanning at an angle).
4. Format information — Two strips near the finder patterns that encode the error correction level and mask pattern.
5. Data modules — Everything else — the actual encoded information, laid out in a specific zigzag pattern.

QR code structure (simplified):

┌─────────┐     ┌─────────┐
│ ███████ │ ─── │ ███████ │   ← Finder patterns
│ █     █ │     │ █     █ │
│ █ ███ █ │     │ █ ███ █ │
│ █     █ │     │ █     █ │
│ ███████ │     │ ███████ │
├─────────┤     └────┬────┘
│ Timing  │──────────│      ← Timing pattern
├─────────┤          │
│         │  Data    │
│ ███████ │  modules │        ← Data fills remaining space
│ █     █ │          │
│ █ ███ █ │   ┌──┐   │
│ █     █ │   │██│   │        ← Alignment pattern
│ ███████ │   └──┘   │
└─────────┴──────────┘

How data is encoded

QR codes support four encoding modes, each optimized for different content types:

The QR encoder automatically selects the most efficient mode for the input data. A URL like https://example.com gets encoded in byte mode (8 bits per character), while a phone number uses numeric mode (just 3.3 bits per digit). This is why QR codes for short numbers are noticeably smaller than codes for long URLs.

Error correction: surviving damage

One of QR's most impressive features is Reed-Solomon error correction. The encoded data includes redundant codewords that let the scanner reconstruct missing or damaged portions — even if part of the QR code is torn, stained, or obscured.

Four error correction levels

Why QR codes are always square

QR codes use a square grid because the finder patterns need to be detectable from any rotation. The three corner squares create a unique geometric relationship that cameras can identify instantly — they determine position, size, and angle in a single scan pass. A rectangular code would need different detection logic for landscape vs. portrait orientation.

The “version” of a QR code determines its size. Version 1 is 21×21 modules, and each version adds 4 modules per side, up to Version 40 at 177×177 modules. Higher versions hold more data but require higher-resolution printing and scanning.

Practical encoding tips

• Shorter URLs = smaller codes — Use URL shorteners if your QR code needs to be printed small.
• Error correction M is the safe default — It balances density and reliability. Only use H if you need a logo overlay.
• Dark foreground on light background — Scanners expect this contrast pattern. Inverted codes work but scan less reliably.
• Maintain a quiet zone — Leave white space (at least 4 modules wide) around the QR code to prevent scanner confusion.

QR codes survived a decade of “that'll never catch on” skepticism because the engineering is genuinely brilliant: self-orienting, self-correcting, and backwards compatible from Version 1 to Version 40.