How to test a logo before launch
Monadic test, A/B and focus group — what's the difference and when to use what
These three methods are often confused, although they solve different tasks. The monadic method is an industry standard: one respondent is shown only one logo, without comparison to alternatives. The logic is simple — in real life, a consumer never sees two competing logos side by side on a screen, they encounter one brand and form an impression in isolation. Comparative presentation creates a systematic error: people begin to evaluate not the logo as such, but the difference between the variants, which distorts the results.
A/B testing, on the contrary, is built on direct comparison — but not in a survey, but in real conditions: two logo variants are shown to different audience segments on a website or in advertising, and behavioral metrics are observed — clicks, time on page, conversion. This works for quickly testing a narrow hypothesis (for example, which of two ready finalists converts better), but does not suit deep understanding of associations.
Focus group — a qualitative method where 6-10 people discuss the logo live or online, and a moderator records not only evaluations but also the reasons for reactions. Here you can dig into obscure associations that won't surface in a questionnaire, but the result depends heavily on the moderator and does not scale to a larger audience.
| Method | What it measures | Sample size | When to apply |
|---|---|---|---|
| Monadic survey | First impressions, associations, without distortion by comparison | 100-300 people per variant | Evaluation of each logo variant separately before final selection |
| A/B test | Behavioral response — clicks, conversion, recognizability | Depends on traffic, usually from 1000 impressions per variant | Final check of 1-2 finalists on real platforms |
| Focus group | Deep associations, emotional reactions, explanation of 'why' | 6-10 people, 2-3 groups | Early stage, search for hidden negative associations |
How to test a logo independently, without an agency
A full quantitative study does not require a team of analysts. For monadic testing, a survey constructor is sufficient — Yandex.View, Google Forms with mailing through targeted advertising, or platforms like Testograf. Respondents can be recruited through targeting in social networks by interests of the target audience — this is cheaper than a professional agency panel and provides a comparable result with a sample of 100-150 people per logo variant.
Practical algorithm for independent testing:
- Prepare 2-3 final logo variants under identical display conditions (one background, one size, no brand labels).
- Conduct a five-second test: show the logo for exactly 5 seconds, then remove it and ask to describe the impression in their own words — this mimics real consumer behavior, which dedicates seconds rather than minutes to a brand.
- Ask a series of monadic questions across four categories: perception (like/dislike), visualization (what comes to mind upon seeing), associative chain (what words come to mind), and category fit (does it resemble a competitor from the same niche or a brand from another sector).
- Collect 100-150 responses for each variant — this is sufficient for a statistically stable picture based on basic metrics.
- Run the finalist through a technical and legal checklist (see below) until approval.
Eye-tracking without special equipment is not possible, but its logic is reproducible: ask the respondent which element of the logo they looked at first. Effective logos capture attention on the key element in the first second of contact — if the responses differ or people mention random details, the composition requires refinement.
Technical test: a case with a logo that 'broke' on the favicon
A demonstrative situation that repeats itself with dozens of startups: the team asserts a logo with a thin gradient line and a small slogan inside the mark, everyone is satisfied at the presentation on a 4-monitor. A month after launch, in the browser favicon and app icon, a gray unreadable square is visible — details merged into a spot already at 32x32 pixels, and at 16x16 the logo has become an amorphous spot without a single recognizable element. Redesign required not cosmetics, but a complete review of the entire concept of the mark, because thin lines and gradients do not scale down in principle.
This situation is not an exception, but a typical result of the absence of technical testing. An obligatory checklist before launch:
- Reduce the logo to 16x16 pixels (favicon size) and check whether the mark remains recognizable without text.
- Convert the file to a black-and-white variant — if the mark loses meaning without color, then identity is held only by the palette, not by the form.
- Check color reproduction in CMYK for printed media — RGB colors on screen and CMYK on paper differ, especially bright shades of blue and green may settle and fade during printing.
- Check contrast according to WCAG 2.1 — the logo must be readable by people with color perception disorders, the minimum contrast ratio for text elements is 4.5:1.
- Check the logo on a smartphone screen in a real interface — in the site header, in social networks, in a messenger — not only in a designer's mockup.
- Ensure that the selected color palette does not reduce the effectiveness of perception: incorrect color choice can undermine the effectiveness of the logo by 20-30% even with an ideal composition.
Evaluation criteria: why Paul Rand's test still works
The classic seven criteria of Paul Rand remain a working tool precisely because they are not about aesthetics, but about the functionality of the sign in real usage conditions.
| Criterion | What is checked | What happens when it fails |
|---|---|---|
| Uniqueness | Differentiation from competitors in the category | Legal refusal to register, consumer confusion |
| Readability | Perception of form and text at a quick glance | Loss of recognizability on small media |
| Adaptability | Performance on different media and sizes | Destruction of the sign on favicon, merch, packaging |
| Memorability | Ability to stay in memory after brief contact | Low brand recall percentage without cues |
| Universality | Working in different cultural and market contexts | Negative associations on new markets |
| Timelessness | Resistance to fads | Need for frequent rebranding |
| Simplicity | Minimum elements for maximum meaning | Detail blending when reducing scale, complexity of reproduction |
An important detail often overlooked: up to 68% of rebrandings are initially perceived negatively by the audience, but are accepted over time as new associations accumulate. This means that a sharply negative initial reaction in a test is not always a signal of concept failure, especially if the logo significantly differs from the previous brand visual language. It's important to distinguish between emotional rejection of novelty and structural issues — unreadability, weak uniqueness, technical non-functionality.
Legal and technical verification before registration
Consumer testing can be conducted perfectly and still result in logo rejection during trademark registration. Rospatent experts evaluate similarity of marks based on three independent criteria: phonetic — how the name or element sounds, graphic — how it visually appears, and semantic — what meaning it conveys. A refusal to register is sufficient if there is similarity in at least one of the three points, even if the logo is completely original in the other two. This means the verification must be conducted not only based on the image, but also on the sound of the brand name and the semantic field it creates.
Before submitting an application, one should:
- Check the designation through open registries of Rospatent and international databases (WIPO, if planning to expand beyond Russia).
- Check for phonetic similarity of the name with already registered brands in the same product category.
- Ensure that the vector logo file is preserved without loss in a format accepted by print, outdoor advertising, and digital platforms — EPS, AI, SVG, with correct paths and without rasterized elements inside.
The systemic error of most teams is to run these three layers of testing sequentially, rather than in parallel. The consumer test confirms the logo, the designer delivers the final files, and only then does the lawyer discover a conflict with someone else's trademark or the technologist realizes the mark is not readable in 16x16. The correct order is the reverse: legal and technical checks of the finalists should be conducted before the consumer test, so that only variants that can be physically registered and used across all media reach the audience.
Monadic test, A/B and focus group — what's the difference and when to use what
These three methods are often confused, although they solve different tasks. The monadic method is an industry standard: one respondent is shown only one logo, without comparison to alternatives. The logic is simple — in real life, a consumer never sees two competing logos side by side on a screen, they encounter one brand and form an impression in isolation. Comparative presentation creates a systematic error: people begin to evaluate not the logo as such, but the difference between the variants, which distorts the results.
A/B testing, on the contrary, is built on direct comparison — but not in a survey, but in real conditions: two logo variants are shown to different audience segments on a website or in advertising, and behavioral metrics are observed — clicks, time on page, conversion. This works for quickly testing a narrow hypothesis (for example, which of two ready finalists converts better), but does not suit deep understanding of associations.
Focus group — a qualitative method where 6-10 people discuss the logo live or online, and a moderator records not only evaluations but also the reasons for reactions. Here you can dig into obscure associations that won't surface in a questionnaire, but the result depends heavily on the moderator and does not scale to a larger audience.
| Method | What it measures | Sample size | When to apply |
|---|---|---|---|
| Monadic survey | First impressions, associations, without distortion by comparison | 100-300 people per variant | Evaluation of each logo variant separately before final selection |
| A/B test | Behavioral response — clicks, conversion, recognizability | Depends on traffic, usually from 1000 impressions per variant | Final check of 1-2 finalists on real platforms |
| Focus group | Deep associations, emotional reactions, explanation of 'why' | 6-10 people, 2-3 groups | Early stage, search for hidden negative associations |
How to test a logo independently, without an agency
A full quantitative study does not require a team of analysts. For monadic testing, a survey constructor is sufficient — Yandex.View, Google Forms with mailing through targeted advertising, or platforms like Testograf. Respondents can be recruited through targeting in social networks by interests of the target audience — this is cheaper than a professional agency panel and provides a comparable result with a sample of 100-150 people per logo variant.
Practical algorithm for independent testing:
- Prepare 2-3 final logo variants under identical display conditions (one background, one size, no brand labels).
- Conduct a five-second test: show the logo for exactly 5 seconds, then remove it and ask to describe the impression in their own words — this mimics real consumer behavior, which dedicates seconds rather than minutes to a brand.
- Ask a series of monadic questions across four categories: perception (like/dislike), visualization (what comes to mind upon seeing), associative chain (what words come to mind), and category fit (does it resemble a competitor from the same niche or a brand from another sector).
- Collect 100-150 responses for each variant — this is sufficient for a statistically stable picture based on basic metrics.
- Run the finalist through a technical and legal checklist (see below) until approval.
Eye-tracking without special equipment is not possible, but its logic is reproducible: ask the respondent which element of the logo they looked at first. Effective logos capture attention on the key element in the first second of contact — if the responses differ or people mention random details, the composition requires refinement.
Technical test: a case with a logo that 'broke' on the favicon
A demonstrative situation that repeats itself with dozens of startups: the team asserts a logo with a thin gradient line and a small slogan inside the mark, everyone is satisfied at the presentation on a 4-monitor. A month after launch, in the browser favicon and app icon, a gray unreadable square is visible — details merged into a spot already at 32x32 pixels, and at 16x16 the logo has become an amorphous spot without a single recognizable element. Redesign required not cosmetics, but a complete review of the entire concept of the mark, because thin lines and gradients do not scale down in principle.
This situation is not an exception, but a typical result of the absence of technical testing. An obligatory checklist before launch:
- Reduce the logo to 16x16 pixels (favicon size) and check whether the mark remains recognizable without text.
- Convert the file to a black-and-white variant — if the mark loses meaning without color, then identity is held only by the palette, not by the form.
- Check color reproduction in CMYK for printed media — RGB colors on screen and CMYK on paper differ, especially bright shades of blue and green may settle and fade during printing.
- Check contrast according to WCAG 2.1 — the logo must be readable by people with color perception disorders, the minimum contrast ratio for text elements is 4.5:1.
- Check the logo on a smartphone screen in a real interface — in the site header, in social networks, in a messenger — not only in a designer's mockup.
- Ensure that the selected color palette does not reduce the effectiveness of perception: incorrect color choice can undermine the effectiveness of the logo by 20-30% even with an ideal composition.
Evaluation criteria: why Paul Rand's test still works
The classic seven criteria of Paul Rand remain a working tool precisely because they are not about aesthetics, but about the functionality of the sign in real usage conditions.
| Criterion | What is checked | What happens when it fails |
|---|---|---|
| Uniqueness | Differentiation from competitors in the category | Legal refusal to register, consumer confusion |
| Readability | Perception of form and text at a quick glance | Loss of recognizability on small media |
| Adaptability | Performance on different media and sizes | Destruction of the sign on favicon, merch, packaging |
| Memorability | Ability to stay in memory after brief contact | Low brand recall percentage without cues |
| Universality | Working in different cultural and market contexts | Negative associations on new markets |
| Timelessness | Resistance to fads | Need for frequent rebranding |
| Simplicity | Minimum elements for maximum meaning | Detail blending when reducing scale, complexity of reproduction |
An important detail often overlooked: up to 68% of rebrandings are initially perceived negatively by the audience, but are accepted over time as new associations accumulate. This means that a sharply negative initial reaction in a test is not always a signal of concept failure, especially if the logo significantly differs from the previous brand visual language. It's important to distinguish between emotional rejection of novelty and structural issues — unreadability, weak uniqueness, technical non-functionality.
Legal and technical verification before registration
Consumer testing can be conducted perfectly and still result in logo rejection during trademark registration. Rospatent experts evaluate similarity of marks based on three independent criteria: phonetic — how the name or element sounds, graphic — how it visually appears, and semantic — what meaning it conveys. A refusal to register is sufficient if there is similarity in at least one of the three points, even if the logo is completely original in the other two. This means the verification must be conducted not only based on the image, but also on the sound of the brand name and the semantic field it creates.
Before submitting an application, one should:
- Check the designation through open registries of Rospatent and international databases (WIPO, if planning to expand beyond Russia).
- Check for phonetic similarity of the name with already registered brands in the same product category.
- Ensure that the vector logo file is preserved without loss in a format accepted by print, outdoor advertising, and digital platforms — EPS, AI, SVG, with correct paths and without rasterized elements inside.
The systemic error of most teams is to run these three layers of testing sequentially, rather than in parallel. The consumer test confirms the logo, the designer delivers the final files, and only then does the lawyer discover a conflict with someone else's trademark or the technologist realizes the mark is not readable in 16x16. The correct order is the reverse: legal and technical checks of the finalists should be conducted before the consumer test, so that only variants that can be physically registered and used across all media reach the audience.
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
