How Much Data Fits in a QR Code?
Every QR code has a version, numbered 1 to 40, that sets how many modules it contains. Version 1 is a 21 × 21 grid; version 40 is a 177 × 177 grid. More modules mean more data, but also smaller modules at any given print size, which is why a code stuffed with data becomes harder to scan. The error-correction level works in the opposite direction: higher correction reserves more of the code for redundancy, so less room remains for your payload. Capacity is therefore a three-way trade-off between how much you encode, how much damage you want to survive, and how large you are willing to print.
What the four error-correction levels buy you
Level L recovers about 7% of damaged codewords, M about 15%, Q about 25%, and H about 30%. Those percentages describe the share of the code that can be lost or obscured and still decode. Level M is the default for ordinary printed codes. Level H is the right choice when the code carries a logo, sits somewhere it may be scuffed, or will be printed on a textured surface. The price of H is capacity: a version-1 code holds 17 byte-mode characters at L but only 7 at H. At version 40 the same trade-off runs from 2,953 characters at L down to 1,273 at H. If your payload only just fits at H, consider shortening it rather than accepting a denser code.
Why shorter payloads scan better
A short URL produces a low-version code with large, forgiving modules. A long URL with tracking parameters, or a full contact record, pushes the version up and shrinks every module. That is why the single most effective optimisation for scannability is shortening the payload: use a short redirect you control instead of a long destination URL, strip unnecessary tracking parameters, and keep vCards to the fields people actually need. The redirect has a second benefit — the destination can change without reprinting the code.
Practical rules
- Keep marketing codes at version 10 or below when you can; above that, increase the print size.
- Use level M for clean indoor print, level Q or H for logos, outdoor use, or rough surfaces.
- Count multi-byte characters carefully: emoji and some non-Latin scripts consume several bytes each.
- If the payload exceeds version 40, split the content across codes or link to a page instead of encoding everything.
- Always test-scan the real payload at the real print size — capacity fitting is necessary, not sufficient.
Capacity tells you what is possible. Print size, contrast, and testing decide what actually works.