Good Logos vs Bad Logos: A Practical Design Audit with Before-and-After Fixes

What this comparison is actually about

A “good” logo and a “bad” logo are not moral categories. They are audit outcomes. The design concept under discussion is a vector mark—symbol, wordmark, or lockup—judged by whether it stays clear, distinctive, and buildable when it is redrawn as paths, reduced to one colour, and scaled from a favicon to a sign. This article does not name a single official brand file, grant reuse rights, or rank living companies. It describes the visual and construction faults that make one version of a mark fail in use, and the practical redraws that usually repair them.

The useful distinction is version versus version: a weak original against a tighter redraw of the same idea. Popular round-ups that “fix the worst logos” work the same way. They keep the name and the core metaphor, then correct geometry, spacing, contrast, and craft. That is a redesign comparison, not a verdict on the organisation behind the mark.

The audit criteria that actually separate the two

Judge the drawing, not the story the brief told. Observable features are what you can see on the artboard: silhouette, stroke, alignment, type, colour count, and how the paths are built. Documented claims are what a brand or designer has stated about intent. Editorial interpretation is the leap from those facts to “this will age well” or “this feels cheap.” Keep those three layers apart, or a stylish concept gets scored as a finished identity.

Check	Weak version, as drawn	Stronger version, as drawn
Silhouette	Competing mini-shapes; the outline is a scribble at arm’s length	One readable mass; you can recognise it from the outer contour alone
Scale	Hairlines, tiny counters, and texture vanish below about a centimetre	Holds as a solid at small sizes and still has presence at large sizes
Colour dependence	Meaning lives in a gradient, a photo fill, or a five-colour stack	The idea survives in a single flat colour, then in black on white
Type	Uneven sidebearings, colliding letters, novelty faces with fragile thins	Even rhythm, optically even gaps, strokes that match the symbol
Geometry	Circles that are not circles, mismatched radii, “centred” forms that look low or left	Optical alignment, consistent corner language, deliberate weight
Construction	Auto-traced bitmap, stacked effects, open overlaps, stray anchors	Closed paths, few nodes, clean joins, a file that exports without repair
Distinctiveness	Generic badge, swoosh, or clip-art mash that could belong to any neighbour	A specific shape or letter treatment you would not confuse with a peer

A mark can fail one check and still work in a narrow use. It becomes a bad logo in practice when several checks fail together: it is hard to print, hard to embroider, hard to put on a dark ground, and hard to remember. A good logo is the inverse. It is a small set of shapes that still look like themselves after every production constraint has been applied.

Vector construction: where “looks fine on screen” falls apart

Direction for this audit is vector logo design, which means the file is supposed to be geometry, not a picture of geometry. A PNG, a screenshot, or a filename that ends in a vector-style label does not prove the artwork is paths. Open it. If the “logo” is a placed image inside a drawing, it is still a raster. If expanding appearance turns a crisp mark into a cloud of overlapping strokes, the identity was styled, not built.

• Count anchors on a curve that should be a simple arc. A good circle or radius uses a handful of points; a bad trace uses dozens of tiny segments that shimmer when scaled.
• Inspect joins and overlaps. Visible mitre spikes, hairline gaps between a letter and a symbol, and shapes that only cover each other because of a drop shadow are construction faults.
• Flatten to one colour. If the idea disappears, the colour was doing the drawing. Rebuild the silhouette first.
• Check stroke versus fill. A logo that is only strokes will change weight when someone scales it as an outline; convert the final artwork to filled shapes at a stated master size.
• Look at compound paths. Counters in letters such as O, A, and B should be true holes, not white shapes sitting on top of black ones. White-on-black “holes” vanish on a coloured background.
• Test export. A clean mark produces a tight SVG or EPS with no embedded bitmap, no leftover guides, and no clipping masks required to hide mistakes.

Availability of a file is not permission to reuse it. A downloadable vector of someone else’s mark is still their mark. The audit below is for work you own, work you have been hired to redraw, or comparison studies that stay on the page as critique.

Before-and-after fixes that actually change the drawing

The pattern in designer “worst logo” redraws is consistent: keep the recognisable idea, throw away the decoration that was impersonating craft. The after state is usually simpler, heavier in the right places, and optically even. The following repairs map to faults you can point at on the artboard.

1. Too many ideas in one mark

Before: an animal, an initial, a globe, a swoosh, and a tagline stacked into a crest. After: one primary device. If the animal is the memory hook, the globe goes. If the letterform is stronger, the animal becomes a supporting graphic for packaging, not part of the master lockup. Editorial note: “busy” here means competing focal points you can count, not a subjective dislike of ornament.

2. Fragile detail at small sizes

Before: engraved lines, tiny stars, outlined type with a thin inner fill, or a face whose eyes are two-pixel dots. After: features merged into larger planes. Eyes become a single dark mass or disappear. Fine lines become cuts in a bold shape, or they are reserved for a secondary, large-size version. The test is mechanical: shrink the mark until it is roughly the height of a browser tab icon. If a part turns to mud, that part is not part of the master.

3. Type that was never spaced

Before: default tracking, a capital with a wide sidebearing next to a tight pair, a script whose connecting strokes break, or a free display face whose thins vanish in print. After: custom sidebearings, optical kerning you can see between every pair, and stroke weights that match the symbol. For a wordmark, the letters are the logo. Uneven gaps read as a mistake even when the concept is sound.

4. Mathematical centre that looks wrong

Before: a triangle or “A” sitting on the geometric midpoint, so it appears to fall; a round icon optically low in a square; a lockup whose symbol and wordmark share a bounding-box centre and still look unrelated. After: shift toward optical centre. Heavy bases move up. Circles in squares sit slightly high. Align by the visual mass of the strokes, then check at the sizes the mark will actually run.

5. Effects standing in for form

Before: bevels, outer glows, photographic fills, chrome, and drop shadows used to fake depth because the silhouette is weak. After: depth implied by overlap, a change in weight, or a single well-drawn counter. Gradients can exist in a brand system; they should not be the only thing holding the mark together. A one-colour redraw that still looks like the same organisation is the proof.

6. Inconsistent corner and stroke language

Before: sharp corners on the shield, round terminals on the type, three different stroke weights in the icon, and a radius that changes from corner to corner. After: one corner rule and one weight system. If the type is rounded, the symbol’s outer corners follow. If the icon is cut from a bold slab, the wordmark should not be a hairline script unless that contrast is clearly intentional and tested at size.

7. Hidden pictures that nobody finds

Before: a second image buried in negative space that only appears after a caption explains it, or a letter that becomes an object only when rotated. After: the primary reading is immediate; the secondary reading is a bonus that does not break the first. Cleverness that needs a walkthrough is a poster idea, not a logo. If the negative-space device is the point, redraw until a new viewer names it without a prompt.

8. Colour as camouflage

Before: low-contrast pairs, rainbow stacking, or a mark that only works on a photograph. After: a dominant colour plus one support, checked as black on white and white on black. Adjacent hues of similar brightness will merge in print and on cheap screens. If two shapes must stay separate, their values—not only their hue names—have to differ.

9. Auto-trace leftovers

Before: a scan of a sketch or a low-resolution bitmap traced in one click, leaving stair-step edges, duplicate paths, and “circles” made of dozens of straight facets. After: a manual redraw over the sketch. Rebuild with the fewest anchors that still describe the curve, snap obvious symmetries, and replace faceted circles with true ones. This is the vector-specific fix that most “worst logo” makeovers skip in the narrative and do first on the artboard.

A working sequence for a before-and-after pass

Run the same sequence on every version you are comparing so the conclusion is about the drawing, not about which mockup had the nicer photograph.

• Place the current mark in a square frame at a large size and at a very small size, in colour and in one-colour black.
• Describe in one sentence what the silhouette is. If you need a second sentence, the mark is carrying two jobs.
• List every decorative layer: outlines, shadows, textures, extra icons, slogans inside the symbol. Remove them from a working copy.
• Rebuild remaining shapes as fills with consistent corners and as few anchors as the curves need.
• Space the type optically; match stem weight to the icon.
• Test on light, dark, and mid-value grounds, and in a single-colour print-style setting.
• Only then restore one extra: a second colour, a simplified large-size pattern, or a lockup with a tagline—never all three at once.

The after file should include a master on a clear field, a stacked lockup if the name is long, a horizontal lockup, and a one-colour master. Those are production states of the same drawing. They are not four different concepts.

How to read a redesign without getting lost in taste

When two public versions of a mark sit side by side, score them on the table above, version against version. A redraw can be quieter and still be better if the silhouette, spacing, and one-colour test improved. A redraw can be more fashionable and still be worse if it introduced thin strokes, a gradient dependency, or a generic geometry that several competitors already use.

Taste still matters at the end, as editorial interpretation: whether the metaphor fits the offer, whether the type’s voice matches the name, whether the mark will feel dated because it copies a current app-icon fashion. Mark that layer as interpretation. The audit layer is finished when you can point at a path, a gap, a contrast failure, or a construction fault.

Gaps in this brief

No named logo file, brand owner, or measured artwork was supplied, so this guide cannot certify a specific mark as good or bad, quote real bounding-box sizes, or reproduce a published “nine worst logos” list. Those lists change with the writer and the year; treating any of them as a factual ranking would invent evidence. If you are comparing live assets, put the actual files on the artboard and run the checks above. If you need legal clearance to redraw or publish a mark, that sits outside a visual audit.