By Mehdi HassanWhat makes a good logo?
A good logo is a mark people can recognize, remember and tell apart from competitors in every place it appears. That means it survives being shrunk to 16 pixels, printed in one color, glanced at for half a second and placed beside the other logos in its category. Taste is part of it, but taste is the part you can't test.
Most logo advice lists principles: simple, memorable, timeless, versatile, appropriate. They're right, but they're hard to apply to the file you already have. So this guide turns those good logo design principles into ten tests with a clear pass or fail. You can run all ten in about an hour with a laptop, a phone and a screenshot tool. For the wider picture of how a startup should approach its mark, start with our logo design guide for startups.
How do you evaluate a logo? The 10 tests
To evaluate a logo, put it in the situations where it will actually live and see what breaks. Run these in order. The first three are the fastest and catch the most problems.
1. The favicon test
Export the mark at 16 and 32 pixels and open it in a browser tab next to ten other tabs. Pass: you can find your tab instantly. Fail: it turns into a gray smudge or the details merge.
2. The one-color test
Fill the whole logo in pure black, then pure white on black. Pass: it still reads as your logo. Fail: it relies on a gradient, a shadow or color contrast between parts to make sense.
3. The squint test
Blur the logo heavily or step back from the screen and squint. Pass: the overall shape is still distinct. Fail: it collapses into a generic circle or rectangle.
4. The competitor line-up
Put your logo in a grid with five direct competitors, all in one color at the same height. Pass: yours stands out by shape, not just by color. Fail: a stranger couldn't pick it out after looking away.
5. The phone description test
Describe the mark to someone in one sentence and ask them to sketch it. Pass: their sketch is recognizably yours. Fail: the description sounds like every other logo, such as an abstract swirl or a gradient orb.
6. The reverse test
Place it on your darkest brand color, on a photo and in dark mode. Pass: a reversed version works without redrawing. Fail: thin strokes vanish or the mark needs a white box behind it.
7. The lockup test
Try the symbol and name side by side, stacked, and the symbol alone. Pass: each version feels like the same brand. Fail: one layout only works at one size.
8. The name legibility test
Put the wordmark in a website nav bar at about 24 pixels tall and on a phone screen. Pass: the company name reads at a glance. Fail: custom letterforms make the name hard to read.
9. The context test
Mock it on a site header, a social avatar cropped to a circle, a pitch deck cover and an invoice. Pass: nothing needs adjusting. Fail: you find yourself tweaking it for each place.
10. The positioning test
Show it without the name to three people in your target market and ask what kind of company it belongs to. Pass: they guess your category and price tier roughly right. Fail: they guess a different industry or a cheaper product.
Which logo tests matter most for a startup?
For most software startups, the favicon, reverse and competitor tests matter most, because your logo lives in browser tabs, app stores, dark interfaces and comparison pages. Print-heavy businesses should weight the one-color test higher. Here's how we'd prioritize the tests by the kind of company you are.
Test | SaaS and AI products | Consumer apps | Services and agencies |
|---|---|---|---|
Favicon | Critical | Critical | Important |
One-color | Important | Important | Critical (print, signage) |
Squint | Important | Critical | Important |
Competitor line-up | Critical | Critical | Important |
Reverse and dark mode | Critical | Critical | Nice to have |
Name legibility | Important | Important | Critical |
Positioning | Critical | Important | Critical |
A logo that fails a critical test for your type of business is worth fixing soon. A logo that fails only a nice-to-have test can usually wait until your next brand refresh. If you're unsure which logo type suits your context, our guide to the 7 types of logos explains where each one is strong and weak.
What are the most common logo design mistakes these tests catch?
The tests mostly catch logos designed on a big white artboard and never tried anywhere else. Fine detail, thin lines and gradients look great at 800 pixels and fall apart in a tab. The second big pattern is sameness: marks that are well made but interchangeable with the rest of the category.
- Too much detail. Small text inside a symbol, thin outlines or more than two ideas in one mark.
- Color doing the work. Two parts of the logo only separate because they're different colors, so the one-color version is a blob.
- Category camouflage. In AI right now that's the sparkle, the orb and the soft gradient. Creative Bloq and The Drum have both covered why so many AI logos look the same.
- Unreadable custom type. A wordmark that's clever but forces people to decode the name.
- No small version. One horizontal lockup and nothing that works as an avatar or app icon.
How do you score your logo's results?
Score each test as pass, borderline or fail, then look at the pattern rather than the total. Two or more fails in the favicon, squint and one-color group mean the mark itself is too complex. Fails in the competitor and positioning tests mean the idea is wrong, which is a bigger job than simplifying the drawing.
For example, suppose a seed-stage analytics startup runs the tests
Hypothetical example: the founders' logo is a gradient hexagon with a thin line chart inside, next to a custom wordmark. It passes the lockup and context tests. It fails the favicon test (the chart disappears), the one-color test (the hexagon and chart merge) and the competitor line-up (three competitors also use hexagons).
The diagnosis: the wordmark is fine and has some recognition with early customers, but the symbol needs replacing. That's a targeted refresh, not a full rebrand. The new symbol drops the chart, uses one solid shape and is designed at 32 pixels first, then scaled up.
Write the results down before you talk to a designer. A list of specific fails makes a much better brief than "it feels off", and it keeps the conversation on problems rather than personal taste. Our logo design brief template has a section for exactly this.
What should you do if your logo fails the tests?
If your logo fails one or two tests, fix the specific problem and keep the rest. If it fails the competitor and positioning tests, the concept needs rethinking, and that's a redesign. Most early-stage logos sit in the first group: the idea is fine, the execution was never tested at small sizes.
- Fails favicon or squint only: simplify the symbol or design a separate small-size version.
- Fails one-color or reverse: redraw with solid shapes and create proper reversed files.
- Fails name legibility: adjust spacing and letterforms, or switch to a cleaner typeface for the wordmark.
- Fails competitor or positioning: start a new concept, and consider whether the brand around it needs work too.
Our guide to logo redesign without losing recognition covers how to decide which parts to keep. If you're weighing whether an AI tool can handle the fix, the honest comparison in AI logo generator vs hiring a designer explains where each one tends to fail these tests. For budgets, typical market ranges for logo-only work run from about $100 to $2,000, according to tier tables published by several design-service companies, and our breakdown of how much a logo costs explains what changes the price.
Does a good logo need to be clever?
No. A good logo needs to be distinctive and usable, and cleverness is optional. Hidden arrows and double meanings are fun, but customers rarely notice them, and a clever idea that fails the favicon test is still a failed logo. Simple marks with one clear idea age better and adapt more easily.
The same goes for literal symbols. A logo doesn't have to show what you do, and in software it usually can't. What it has to do is look like it belongs to a company at your level, in your category, and look different from the companies next to it. Whether that's a wordmark, a symbol or both is a separate decision, which we cover in wordmark vs logomark vs combination mark.
A logo is finished when it works in a browser tab, not when it looks good on an artboard.
