Logo quality assessment criteria
Paul Rand's test as a coordinate system
The methodology of the legendary designer remains a working tool precisely because it translates the subjective 'pretty' into measurable points. A logo is evaluated on seven parameters, maximum — 75 points. Simplicity weighs more than the rest — 15 points, the remaining six criteria each give 10. If the total is below 60, the mark needs to be reconsidered, not just 'polished'.
| Criterion | What is checked | Points |
|---|---|---|
| Simple (simplicity) | Is the form readable within 1-2 seconds | 15 |
| Distinctive (uniqueness) | Does not get confused with competitors or foreign marks | 10 |
| Memorable (memorability) | Stays in memory after a brief glance | 10 |
| Visible (visibility) | Works in different sizes and on different backgrounds | 10 |
| Adaptable (adaptability) | Maintains recognizability across all media | 10 |
| Universal (Universal) | Not bound to slang and local trends | 10 |
| Timeless (Timeless) | Will not become outdated within 3-5 years | 10 |
The test is good as a filtering sieve, but it does not answer the question of whether you will register this mark and whether it will offend the audience in another country. Separate checks are needed for this — legal and semantic.
Checking for uniqueness: where to look and whom to involve
Designers often assure the client that 'there is definitely nothing like this anywhere' — at first glance. This is not a check, but an illusion. The real procedure looks different.
- Search in the Rospatent database (the Russian Federal Trademark Register) — if the business operates in Russia.
- Check via the WIPO Global Brand Database — an international database covering applications under the Madrid System.
- Classification of the figurative element according to the Vienna Classification — the international system of coding graphical signs (for example, the category "stars", "geometric figures", "animals"). This allows narrowing the search to signs with similar elements.
- Manual examination by a patent examiner — automatic search by images is extremely limited, algorithms poorly recognize stylized and abstract forms.
The key mistake is to think that online image checking services (such as reverse image search) replace legal expertise. They find exact matches, but do not detect similarity to the degree of blending — and it is precisely this criterion upon which Rospatent refuses registration.
Black-and-white version and technical adaptability
The requirement that "the logo must work in a black-and-white version" sounds like a formality, but it is based on specific practical logic. If a sign relies only on color contrast, it fails when printed on a fax machine, engraved, embossed, or in documents where color is technically impossible. The monochrome version is a stress test for the form: if the silhouette is readable without color, then the form carries meaning by itself, not masked by colors.
There is also an unexpected bonus: perception studies show that black-and-white logos raise the sense of brand status by 15-20% compared to colored analogs — they are associated with coldness and artistry, which is often perceived by the audience as premium quality.
Test on the favicon — not scaling, but reassembly
Favicon 16×16 pixels is only 256 pixels, less than 1% of the area of the logo at resolution 200×200. Simply reducing the file size almost always results in a mess instead of the sign. The working approach is to create a simplified version separately, often leaving only the monogram or the key geometric element. The contrast of such a version must not be lower than 3:1, and according to the WCAG AA standard — 4.5:1, especially taking into account the dark mode of browsers, where the background is unpredictable.
Guard band and format set
A separate quality criterion is the presence of a protective field around the mark, at least 1/4 of its height on all sides. Without this field, the logo 'sticks' to neighboring elements on the packaging, website, or sign. The complete designer acceptance package should include:
- Vector files AI, EPS, SVG — for printing and any scaling without quality loss.
- Raster PNG, JPG files — for use on the web and in presentations.
- RGB color profiles for screens and CMYK for printing — otherwise the color 'floats' when output to paper.
It is also necessary to resolve the question of the mark's architecture here: a polymorphic logo with multiple variations is good for digital ecosystems and social networks, but is absolutely inappropriate for a sign or packaging where a single fixed form is needed. If the business does not plan a digital-first strategy, a simple geometric form without variations will prove more practical.
Semantic audit: errors that are invisible to the designer
The designer evaluates the form, but rarely checks cultural codes — this is the area where logos most often fail, especially on international markets. Typical traps:
- Hidden images that are read ambiguously — a symbol that is invisible at the development stage but noticeable to the audience (a classic example from global practice — the arrow in the FedEx logo, which was found by accident, but such 'discoveries' can also be unfortunate).
- Colors with opposite meanings in different cultures — white as mourning in Asia, green as a religious symbol in Arab countries.
- Numerical symbolism — the number of elements (petals, lines, stars) matching unfortunate numbers in a specific culture.
- Gestures and silhouettes of animals that are offensive or taboo in certain regions.
The only way to avoid this is to show the final version of the carrier to the target culture before launch, not after. Design reviews within the agency almost never catch such things because the team looks at the form, not the context of perception.
How to measure the effectiveness of a logo in practice
The Rand criteria are expert evaluations, not audience reaction data. To understand whether a sign works for conversion and memorability, separate measurements are needed.
| Method | What it shows | When to apply |
|---|---|---|
| A/B testing on the website/landing page | The impact of the logo variant on clicks, time on page, conversion to the target action | When choosing between 2-3 final variants prior to launch |
| Memory test (tachistoscope test) | Speed and accuracy of brand recognition after brief exposure (fractions of a second) | Checking the memorable criterion without subjective evaluations |
| Focus groups with open-ended questions | Associations, emotional reactions, cultural readings | Early stage, before finalization, especially for new markets |
| Recognition survey (brand recall survey) | Percentage of audience recalling the brand by silhouette without text | Through 3-6 months after launch, for assessing real adoption |
The main methodological error is testing the logo in a vacuum, without context of use. A mark that is well-recognized on a white background of a presentation may completely lose legibility on a product packaging among competitors on the shelf. Therefore, an A/B logo test is valid only when it is embedded in a real usage environment — a website, app, or storefront — and not shown to respondents as an isolated image.
Collecting the evaluation in a unified checklist
Full logo acceptance is not a single test, but a sequence of filters, each with its own area of responsibility:
- Expert evaluation according to Rand's methodology — filters out weak forms before spending money on a lawyer.
- Legal check at Rospatent and WIPO with the involvement of a patent agent — protects against refusal of registration and lawsuits.
- Technical test — black-and-white version, favicon 16×16, protective field, set of formats in RGB and CMYK.
- Semantic audit with carriers of target market culture — excludes reputational failures.
- Behavioral testing — A/B test in a real environment, recall survey conducted several months after launch.
Skipping any of these steps does not automatically make the logo bad — but turns its evaluation into a lottery, where the stake is money for rebranding in a year or two, when the problem surfaces on its own.
Paul Rand's test as a coordinate system
The methodology of the legendary designer remains a working tool precisely because it translates the subjective 'pretty' into measurable points. A logo is evaluated on seven parameters, maximum — 75 points. Simplicity weighs more than the rest — 15 points, the remaining six criteria each give 10. If the total is below 60, the mark needs to be reconsidered, not just 'polished'.
| Criterion | What is checked | Points |
|---|---|---|
| Simple (simplicity) | Is the form readable within 1-2 seconds | 15 |
| Distinctive (uniqueness) | Does not get confused with competitors or foreign marks | 10 |
| Memorable (memorability) | Stays in memory after a brief glance | 10 |
| Visible (visibility) | Works in different sizes and on different backgrounds | 10 |
| Adaptable (adaptability) | Maintains recognizability across all media | 10 |
| Universal (Universal) | Not bound to slang and local trends | 10 |
| Timeless (Timeless) | Will not become outdated within 3-5 years | 10 |
The test is good as a filtering sieve, but it does not answer the question of whether you will register this mark and whether it will offend the audience in another country. Separate checks are needed for this — legal and semantic.
Checking for uniqueness: where to look and whom to involve
Designers often assure the client that 'there is definitely nothing like this anywhere' — at first glance. This is not a check, but an illusion. The real procedure looks different.
- Search in the Rospatent database (the Russian Federal Trademark Register) — if the business operates in Russia.
- Check via the WIPO Global Brand Database — an international database covering applications under the Madrid System.
- Classification of the figurative element according to the Vienna Classification — the international system of coding graphical signs (for example, the category "stars", "geometric figures", "animals"). This allows narrowing the search to signs with similar elements.
- Manual examination by a patent examiner — automatic search by images is extremely limited, algorithms poorly recognize stylized and abstract forms.
The key mistake is to think that online image checking services (such as reverse image search) replace legal expertise. They find exact matches, but do not detect similarity to the degree of blending — and it is precisely this criterion upon which Rospatent refuses registration.
Black-and-white version and technical adaptability
The requirement that "the logo must work in a black-and-white version" sounds like a formality, but it is based on specific practical logic. If a sign relies only on color contrast, it fails when printed on a fax machine, engraved, embossed, or in documents where color is technically impossible. The monochrome version is a stress test for the form: if the silhouette is readable without color, then the form carries meaning by itself, not masked by colors.
There is also an unexpected bonus: perception studies show that black-and-white logos raise the sense of brand status by 15-20% compared to colored analogs — they are associated with coldness and artistry, which is often perceived by the audience as premium quality.
Test on the favicon — not scaling, but reassembly
Favicon 16×16 pixels is only 256 pixels, less than 1% of the area of the logo at resolution 200×200. Simply reducing the file size almost always results in a mess instead of the sign. The working approach is to create a simplified version separately, often leaving only the monogram or the key geometric element. The contrast of such a version must not be lower than 3:1, and according to the WCAG AA standard — 4.5:1, especially taking into account the dark mode of browsers, where the background is unpredictable.
Guard band and format set
A separate quality criterion is the presence of a protective field around the mark, at least 1/4 of its height on all sides. Without this field, the logo 'sticks' to neighboring elements on the packaging, website, or sign. The complete designer acceptance package should include:
- Vector files AI, EPS, SVG — for printing and any scaling without quality loss.
- Raster PNG, JPG files — for use on the web and in presentations.
- RGB color profiles for screens and CMYK for printing — otherwise the color 'floats' when output to paper.
It is also necessary to resolve the question of the mark's architecture here: a polymorphic logo with multiple variations is good for digital ecosystems and social networks, but is absolutely inappropriate for a sign or packaging where a single fixed form is needed. If the business does not plan a digital-first strategy, a simple geometric form without variations will prove more practical.
Semantic audit: errors that are invisible to the designer
The designer evaluates the form, but rarely checks cultural codes — this is the area where logos most often fail, especially on international markets. Typical traps:
- Hidden images that are read ambiguously — a symbol that is invisible at the development stage but noticeable to the audience (a classic example from global practice — the arrow in the FedEx logo, which was found by accident, but such 'discoveries' can also be unfortunate).
- Colors with opposite meanings in different cultures — white as mourning in Asia, green as a religious symbol in Arab countries.
- Numerical symbolism — the number of elements (petals, lines, stars) matching unfortunate numbers in a specific culture.
- Gestures and silhouettes of animals that are offensive or taboo in certain regions.
The only way to avoid this is to show the final version of the carrier to the target culture before launch, not after. Design reviews within the agency almost never catch such things because the team looks at the form, not the context of perception.
How to measure the effectiveness of a logo in practice
The Rand criteria are expert evaluations, not audience reaction data. To understand whether a sign works for conversion and memorability, separate measurements are needed.
| Method | What it shows | When to apply |
|---|---|---|
| A/B testing on the website/landing page | The impact of the logo variant on clicks, time on page, conversion to the target action | When choosing between 2-3 final variants prior to launch |
| Memory test (tachistoscope test) | Speed and accuracy of brand recognition after brief exposure (fractions of a second) | Checking the memorable criterion without subjective evaluations |
| Focus groups with open-ended questions | Associations, emotional reactions, cultural readings | Early stage, before finalization, especially for new markets |
| Recognition survey (brand recall survey) | Percentage of audience recalling the brand by silhouette without text | Through 3-6 months after launch, for assessing real adoption |
The main methodological error is testing the logo in a vacuum, without context of use. A mark that is well-recognized on a white background of a presentation may completely lose legibility on a product packaging among competitors on the shelf. Therefore, an A/B logo test is valid only when it is embedded in a real usage environment — a website, app, or storefront — and not shown to respondents as an isolated image.
Collecting the evaluation in a unified checklist
Full logo acceptance is not a single test, but a sequence of filters, each with its own area of responsibility:
- Expert evaluation according to Rand's methodology — filters out weak forms before spending money on a lawyer.
- Legal check at Rospatent and WIPO with the involvement of a patent agent — protects against refusal of registration and lawsuits.
- Technical test — black-and-white version, favicon 16×16, protective field, set of formats in RGB and CMYK.
- Semantic audit with carriers of target market culture — excludes reputational failures.
- Behavioral testing — A/B test in a real environment, recall survey conducted several months after launch.
Skipping any of these steps does not automatically make the logo bad — but turns its evaluation into a lottery, where the stake is money for rebranding in a year or two, when the problem surfaces on its own.
Recent articles
MegaFon, BMW, VTB and more 240+ companies. See how it looks in the showcase reel:
Our branding agency is approached to order:
Branding agency services
