How Teams Can Test a Logo for Effectiveness Before Launch
Learn how to test a logo step by step, from technical checks and five-second tests to recall, preference tests and consumer interviews that show which concept wins.
Two or three logo concepts usually reach the final round, and the one that wins will sit on packaging, storefronts and app screens for years. The brand and insights teams making that choice need to know which concept consumers will notice, remember and connect to the brand. This guide walks through testing logos step by step. It covers what to measure, which methods and tools fit each question, and how to keep feedback from pushing the decision the wrong way. Fixing a logo before launch costs less than reversing one in public: Cracker Barrel unveiled its first logo redesign in 48 years in August 2025 and restored the original logo within about a week, after a public backlash.
What a team needs before testing a logo
A logo test answers the brand’s question only when the team has settled what the logo is for and who gets to judge it. Put these inputs in place before recruiting anyone:
- Brand objectives: A short written list of what the logo must signal, such as approachable, premium or clinical, drawn from the brand strategy.
- A defined target audience: Category buyers described by screener criteria (what they buy, how often, which brands). Colleagues and friends of the brand are not the audience.
- Two to four finished concepts: Every concept rendered at the same level of polish, so finish quality does not decide the result.
- Vector files: SVG, EPS or PDF masters for each concept, plus single-color and reversed versions.
- Success criteria agreed with leadership: A written record of which criteria matter most for this brand and what score counts as a pass, completed before any results come back.
What to measure in a logo test
Each criterion answers a different question about a logo, so a test plan should name the criterion before it names the method. This scorecard pairs each criterion with a practical test:
| Criterion | What it measures | How to test |
|---|---|---|
| Recall | Whether people can bring the logo to mind without a prompt | Ask for a sketch or spoken description a day or more after exposure |
| Recognition | Whether people pick out the logo and name the right brand | Lineup with competitor logos and close variations |
| Legibility | Whether the name and mark read correctly at a glance | Five-second exposure, then ask what the logo said |
| Scalability | Whether the mark holds up from favicon to billboard | Render at favicon, app icon and large-format sizes |
| Monochrome | Whether the logo works in one color and reversed | Black, grayscale and white-on-dark versions |
| Distinctiveness | Whether the logo signals only this brand in the category | Count how often people attribute it to a competitor |
| Brand fit | Whether associations match the brand objectives | Attribute ratings plus open questions on what it suggests |
Liking is left off on purpose. It is easy to collect, but on its own it says nothing about whether people will remember or attribute the logo, a point Step 3 returns to.
How to test a logo: step by step
Teams start with technical checks, then run audience tests from gut reaction to memory to realistic context, and use interviews to understand why consumers respond as they do. This sequence moves from cheap to expensive, so the team fixes or drops concepts that fail early checks before paying for later tests.
Step 1: Run technical checks before showing anyone
Technical checks need no participants, and the team uses them to remove concepts that cannot survive production. Browsers, app stores and social networks set minimum sizes for favicons, app icons and avatars. Google Search accepts square favicons as small as 8x8 pixels and recommends larger than 48x48. Check each concept for:
- Rendering at 16x16 and 32x32 pixels as a browser favicon, and as a masked app icon
- A single-color version in black and in one brand color
- A reversed version, light on a dark background
- Dark mode, including transparent areas that disappear against dark interfaces
- Vector rendering on desktop and mobile screens
Reviewing concepts only at presentation size on a large monitor is a mistake. Hairline strokes, fine detail and small type are the first things to disappear when a logo shrinks.
Step 2: Run a five-second impression test
Once the concepts survive production checks, the five-second test captures what consumers take in at a glance. The test shows a concept briefly, hides it, and asks what people remember. Five seconds is too short to read copy or notice specific fonts but long enough to form an impression of visual style. Participants should not be warned about the time limit, because the warning primes them to memorize instead of react. Useful questions:
- What company or product do you think this logo belongs to?
- What category is it in?
- What did the name say?
- Which three words describe the brand behind it?
- What, if anything, was hard to make out?
The expected outcome is a read on legibility and first associations for each concept: whether people read the name correctly, place the brand in the right category and choose words that match the brand objectives. A concept that many people place in the wrong category has a problem that liking scores will not reveal.
Step 3: Compare concepts with preference and A/B tests
In a side-by-side preference test, each person sees all concepts and picks the best fit for the brand, and paired comparisons can detect small differences with fewer people than separate-group tests. In a monadic A/B test, each person sees only one concept and the team compares scores across matched groups, so the team assesses each logo without alternatives beside it.
Two practices keep either design honest:
- Randomize order: Rotate which concept appears first or on the left. The first version people see colors their reactions to the rest.
- Rate brand fit, not appeal: Ask people to rate each concept against the brand objectives on a scale, then follow with an open question about why.
Preference measures relative appeal. People may favor a concept’s finish or familiarity, so teams still need separate checks of whether consumers remember the logo, read it at 16 pixels and link it to the brand instead of a competitor. Treat preference as one input on the scorecard, weighed against recall, recognition and fit.
Step 4: Measure recall and recognition after a delay
Preference shows what people like in the moment; memory shows what stays. The memory test runs in two parts, a day or more after exposure:
- Unaided recall: Ask people to sketch the logo or describe it in words. Agree on key features to score (shape, color, layout, name), rather than judging drawing skill.
- Aided recognition: Show a lineup that mixes the test concept with competitor logos and close variations, and ask which logo belongs to the brand.
Keep memory measures unprompted wherever possible, and run recall before recognition so participants try to remember the logo before seeing it again in the lineup. Jenni Romaniuk of the Ehrenberg-Bass Institute warns that prompted measurement of brand assets inflates fame scores by up to about 20 percentage points. A concept that people attribute to a competitor in the lineup is failing on distinctiveness, however well it did on preference.
Step 5: Test the logo in context mockups
Memory tests use the logo on its own, but consumers see it among other brands, so the finalists go into realistic mockups:
- Packaging on a shelf next to the three or four competitors buyers see most
- A storefront sign viewed from a distance
- A social media avatar cropped to a circle
- A display ad and an app icon on a phone home screen
For each mockup, measure standout (how quickly people spot the brand), findability (how long it takes to locate a named brand on the shelf) and fit (whether the logo still feels like the brand in that setting). For a logo refresh, showing both the new and the existing design on shelf shows whether the new mark gains or loses attention against the one consumers already know.
Step 6: Ask consumers why with interviews
After comparing performance in Steps 2 to 5, researchers interview consumers to understand their reactions and decide what to fix. A low fit score might come from a color consumers associate with medicine, a shape they connect to a competitor or a typeface they find cheap, and researchers can explore those causes through open conversation. Researchers ask these questions and follow each answer with another question:
- What the logo reminds people of
- What kind of company they picture behind it
- How it makes them feel
- What they would expect its products to cost
AI-moderated interviews are a newer way to run this kind of conversation at scale, with hundreds of consumers instead of a dozen. Strella runs these interviews for logo concept tests. Researchers embed each concept as an image in the discussion guide, and participants, who are real consumers, talk through their reactions voice-to-voice with the AI moderator. Results synthesize as interviews finish, and every theme links to the participant’s own words and video, so teams hear consumers describe each concept in their own voices at the decision meeting. Prose used Strella to test a new brand and creative direction.
Step 7: Score the results and make the decision
With the tests complete, the team combines separate results into one comparison it can defend in front of leadership:
- Set weights from brand goals: No widely accepted weighting covers all these criteria. A new brand in a crowded category might prioritize distinctiveness and recall, while a refresh of a well-known mark might prioritize recognition and continuity.
- Score consistently: Score each concept per criterion on the same scale, then apply the weights agreed before testing.
- Read the reasons beside the totals: Use interview verbatims to understand the scores and identify fixes.
If one concept leads on the top-weighted criteria and researchers find no fixable flaw in the interviews, pick it; if the lead comes only from preference, or researchers trace a weak score to one fixable cause such as a thin stroke or a misleading color, iterate on that concept and retest only the criteria it failed.
Methods and tools for logo testing
Sample size and timing depend on the decision the team needs to make and how precise the result has to be:
| Method | What it measures | Typical sample | Typical turnaround | Example tools |
|---|---|---|---|---|
| Interviews (AI-moderated or human-moderated) | Associations, emotions, reasons behind scores | 9 to 17 per similar segment | AI-moderated: a few days from launch to results. Human-moderated: a few weeks, including recruiting | Strella |
| Five-second test | First impression, legibility, perceived category | About 40 | Same day with a panel | Online panel testing platform |
| Preference test | Which concept fits the brand best | Depends on the gap the team needs to see | Same day with an online panel | Online panel testing platform |
| Monadic A/B test | Fit and recognition per concept, without comparison | 40 per group for a 20-point gap; about 110 for a 10-point gap | 1 to 2 weeks | Survey platform with separate test cells |
| Recall and recognition | Memory and correct brand attribution | Same participants as the earlier test | Length of the chosen delay plus response time | Survey or interview platform with image stimuli and delayed follow-up |
Worked example: testing two logo concepts
A hypothetical oat-milk brand shows how the steps fit together. The brand is relaunching and has two concepts. Concept A is an abstract geometric monogram. Concept B is a wordmark with a small leaf drawn into the first letter. The brand objectives are approachable, natural and easy to spot on a crowded shelf, and the audience is people who buy plant-based milk at least weekly. The team ran the five-second questions from Step 2, brand fit ratings against the three objectives, an unaided description and competitor lineup two days later, a shelf mockup, and interviews.
The hypothetical results by criterion:
| Criterion | Concept A | Concept B |
|---|---|---|
| Read the name correctly in five seconds | 58% | 91% |
| Side-by-side preference | 54% | 46% |
| Brand fit (approachable, natural), out of 5 | 3.1 | 4.2 |
| Described key features after two days | 22% | 41% |
| Picked correctly from competitor lineup | 47% | 78% |
| Found on shelf mockup within five seconds | 61% | 74% |
Illustrative interview comments show the pattern. On Concept A, a participant might say, “It looks sleek, like a bank or a tech startup, not something I’d drink.” On Concept B: “The leaf tells me it’s plants before I even read the name.” The team picks Concept B despite A’s narrow preference lead, because B wins every weighted criterion and the interviews explain A’s appeal: people found it sleek but could not say what the brand sold. Before finalizing, the team thickens the leaf stroke, which both the interview comments and the 16-pixel check flagged as too fine.
Troubleshooting biased or misleading logo feedback
Most misleading logo results trace back to who was asked, how and in what order. The common failure modes and their fixes:
- Leading questions: Wording such as “How much do you like the bold new color?” signals the expected answer. Rewrite questions as open and neutral, and have someone outside the design team review the guide before launch.
- Novelty and familiarity bias: People often rate an unfamiliar logo lower at first, and Walsh, Winterich and Mittal (2010) found that consumers strongly committed to a brand evaluate a redesigned logo more negatively. Split results by brand commitment and weight recall and recognition above first-exposure liking.
- Asking designers or employees: People who made the logo or live with the brand tend to overvalue their own work and cannot set aside what they already know. Keep internal reviews to checking concepts against agreed criteria, and take audience judgments from target consumers only.
- Too few participants for numbers: Percentages from five or ten people swing widely from one group to the next. Use small groups for the reasons behind reactions, and the sample sizes in the methods table for any figure the decision depends on.
- Order effects: The concept shown first or on the left can gain or lose from position alone. Randomize or counterbalance order in every test that shows more than one concept.
FAQ
How many participants do you need to test a logo?
Sample size follows the difference the team needs to detect: about 40 people per concept shows a 20-point gap, and about 110 shows a 10-point gap. Interviews usually surface the main themes within 9 to 17 conversations per similar segment.
What questions should you ask when getting feedback on a logo?
Open questions about associations, fit and memory produce more useful feedback than asking whether people like a logo. Ask what company or category the logo suggests, which words describe the brand behind it, how well it fits the brand’s stated objectives and, a day later, what people remember of it.
How do you test a logo at small sizes and in black and white?
Export each concept as a 16x16 and 32x32 pixel favicon, a masked app icon, and black, grayscale and white-on-dark versions, then check that the name stays legible and the mark stays recognizable. Any concept that loses its key feature in these exports needs heavier strokes or simpler detail before consumer testing.
Can AI-moderated interviews replace a logo survey?
No. The two methods answer different questions and work best together. A survey or panel test counts how many people recognize, recall or prefer each concept, while AI-moderated interviews explain the associations and feelings behind those counts through follow-up questions, so teams typically run the counting tests and use interviews to decide what to fix.
How long does logo testing take?
Teams should plan for a day to a few weeks, depending on recruitment, methods and the chosen recall delay. Panel-based tests can return results the same day, and A/B tests usually take 1 to 2 weeks.

