Testing Checklist
A logo is only finished when it has survived these tests. Most can be run with
scripts/svg_audit.py (structure and geometry) and scripts/preview_sheet.py (visual tests); the rest need
judgement. Run them on every concept before presenting and again on the final artwork.
1. Scale
- 16 px favicon: the core idea survives; no mush. If not, design a simplified small-size version (fewer elements, thicker strokes, bigger gaps) — detailed marks may keep a reduced companion.
- 24–32 px (app lists, social avatars): recognisable at a glance.
- Very large (billboard, building): curves are smooth, no lumpy anchors, kerning holds up; flaws invisible at small sizes become glaring.
- Minimum size: shrink until the mark loses its identifying features; the size just above that is the documented minimum (in px for screen and mm for print).
2. Colour and value
- Works in one-colour black on white.
- Works in one-colour white on black — and doesn't look heavier than the black version (irradiation). If it does, provide a slightly thinned reversed file.
- Works on the brand colour, on photos and on patterns (use a graphic device — outline or container — if not).
- Greyscale: colour segments still separate by value; nothing relies on hue alone.
- Colours reproduce in CMYK and have spot/Pantone equivalents if vivid.
3. Form
Squint/blur test: the silhouette alone is distinctive.
Mirror test: flipping reveals no proportion errors you'd adapted to.
Rotation / unintended readings: view rotated 90° and 180°, tiny, and at a distance; show it to someone who hasn't seen the brief. Check for accidental letters, body parts, offensive or sexual readings, political or religious symbols, and resemblance to hazard signs.
Optical corrections done: overshoot, bone effect, horizontal stroke thinning, optical centre.
Geometry clean: no near-miss angles, perfect primitives, consistent radii and stroke widths (
svg_audit.pyflags near-miss angles and tiny details).Balance: not accidentally tilted; weight distributed; proportions close to square for symbols.
Letter test: every customised letter still reads as the intended letter at first glance (ask: "what letter is this?"). A K that reads as an h, or an N that reads as a lightning bolt, needs revising.
Junction check: zoom in on every place strokes meet or overlap — no accidental notches, slivers, lumps, hairline gaps or ink traps.
Peer test: next to 3–4 exemplary library marks at the same size, yours looks equally resolved.
4. Distinctiveness and originality
- Shelf test: placed among competitors (
preview_sheet.py --refs-industry <industry>), it stands out rather than blending in. - Familiarity test: it doesn't remind you (or anyone you ask) of an existing mark. If it feels familiar and it isn't yours, it's someone else's.
- Library check: search the bundled library for the same subject/technique
(
search_library.py --subject <thing>); make sure yours is clearly different. - Trademark check (recommend to the user): search the relevant trademark databases and do a reverse image search before launch. You cannot give legal clearance — say so.
5. Meaning and fit
- The idea can be explained in one sentence.
- Tone matches the brand adjectives (sharp/round, heavy/light, warm/cool, classic/modern).
- It identifies rather than explains; it will still fit if the business expands.
- Cultural meaning of colours and symbols checked for the audience's cultures.
- It works without the tagline and, for symbols, eventually without the name.
6. Media and production
- One-colour printing, embroidery (no hairlines, gaps ≥ ~1 mm at chest-logo size), laser engraving, vinyl cutting, signage, dark mode, animation.
- App icon: optically sized inside the platform's rounded tile; no fine text.
- Social avatars: survives circular crop.
- Master file: vector only, text outlined, strokes expanded, no filters/raster, clean viewBox
(
svg_audit.pyscore ≥ ~90 with no FAIL).
7. Context
- Shown in 5–6 realistic mockups relevant to the business (a café's cup, not a gym bag), in a consistent style. Never judge a logo only in isolation on a white artboard.
- Lockups (horizontal, stacked, symbol-only) all tested at their intended sizes.
Recording results
Summarise test outcomes when presenting: what passed, what was adjusted (e.g. "thinned the reversed version by 3 %; opened the counter gap from 6 to 10 units so it survives 16 px"), and what the user must still do (trademark search, Pantone proofing).