Constellations, patent pending

Case study · Higher education

The College Use Case

How test structure shapes what you see. We should’ve suspected that prospective students like fun. The first test proved it and nothing else. The second test, restructured, showed what they actually respond to.

Quick answer

How a Constellations test is structured decides what it can tell you. A liberal arts college tested campus imagery with prospective students, high school juniors and seniors already in contact with admissions. The first round mixed academic and extracurricular scenes evenly across two Audience Perception Maps, and the energetic scenes dominated both, confirming only that teenagers prefer concerts to classrooms. The second round gave each map one theme: academic settings on one, student life on the other. That surfaced which academic environments felt welcoming and which parts of campus life stood out. Blending everything works for a brand-identity question. It does not work when one category is louder than the other. Mix them, and the test measures energy, not preference.

Using visual research to understand college appeal

A liberal arts college partnered with Constellations to explore how prospective students perceive different aspects of campus life. The goal was to understand which visual cues, environments, activities, atmospheres, shape the impressions high school students form early in the admissions process.

Constellations maps audience reactions to images to identify what resonates most. Here the participants were high school juniors and seniors who had either attended an on-campus event or were in active communication with admissions officers. Their responses were meant to help the college refine how it communicates its culture and environment to future applicants.

Test 1: mixed content across both maps

The first round used a curated set of professionally photographed marketing images to build two Audience Perception Maps within a single survey. Each map carried a mix of academic and extracurricular scenes: classroom settings alongside field trips, concerts and student-led events.

Student workers from the admissions office invited prospective students to take the survey while they waited in line at an on-campus event. Responses were collected on iPads. The people running it reported that participants found it intuitive and quick, typically a minute or two to complete.

The results showed the limitation in the structure right away. Images of energetic, activity-based scenes dominated the feedback on both maps, overshadowing the more subdued classroom and academic settings.

Figure 1 · Test 1, mixed content

Two mixed Audience Perception Maps where every response cluster sits on an activity tile Two boards side by side, each a grid of twelve tiles with academic and activity scenes evenly mixed. Cyan response marks appear only on the activity tiles of both boards. A key panel at right identifies the two tile types and the response marks. APM 1 · mixed ACADEMIC AND CAMPUS LIFE, EVENLY MIXED APM 2 · mixed SAME MIX, SECOND MAP KEY Academic scene Activity scene Responses landed here Every hit is on an activity tile. The academic tiles were never compared with each other. They were compared with a concert.
Read where the marks are not. Not one academic tile drew a response on either map, which does not mean no classroom appealed. It means every classroom was being judged next to a concert, and lost.

That imbalance made it impossible to tell which academic themes held the most appeal, or which extracurricular environments resonated most strongly. The high-energy visuals pulled participant attention, and the finer distinctions between content types never surfaced. A heatmap shows you where attention went. It cannot show you where it would have gone without the loudest tile on the board.

Why the brand method did not transfer

In prior brand-focused research, blending a diverse range of visuals across multiple Audience Perception Maps has worked well for surfacing a brand’s core identity. It gives clear creative direction by revealing what an audience intuitively associates with the brand, which is how an enterprise rebrand found its direction without a single rework.

In this context the same method produced results that were too broad. It confirmed a predictable dynamic: high school students prefer lively, social imagery over classroom scenes. Unsurprising, and of limited strategic value.

Figure 2 · Which structure for which question

Decision fork matching a blended test structure to brand questions and a one-theme-per-map structure to category comparisons A fork. The question at the top splits into two. A brand-association question leads to blending everything across the maps, which surfaces the core identity. A within-category question leads to one theme per map, which compares like with like. A dashed crossed-out path between the two structures is labelled: mixed maps for a category question measure energy, not preference. What do you need the test to tell you? What does this audience associate with the brand as a whole? Which scenes inside a category win, against their own kind? Blend everything across the maps One theme per map Surfaces the core identity Compares like with like mixed maps for a category question measure energy, not preference
The tool did not change between the two rounds. The question did. A blended board answers what the audience associates with the whole, and the college was not asking about the whole. It was asking about the parts.

So the framework was revised. Rather than mixing content types across maps, each map was given a distinct thematic focus, one devoted to academic settings, the other to student life. That created an intentional comparison structure and clarified both sides of what the college wanted to know.

Test 2: academic and campus life, separated

The restructured test split the content into two distinct Audience Perception Maps. One featured exclusively academic and classroom imagery. The other focused solely on campus life outside the classroom.

The clearer division let participants engage with each theme directly. Rather than confirming a general preference for “fun” imagery, the revised test surfaced specific insight: which types of academic environments felt most welcoming, and which aspects of student life stood out as most compelling.

Figure 3 · Test 2, one theme per map

Two single-theme Audience Perception Maps, each showing response clusters on specific tiles within its own category Two boards side by side. The left holds only campus life scenes, the right only academic scenes. Each board now shows cyan response marks of varying size on a few of its own tiles. A panel at right notes that participants, imagery and test length were unchanged, and that each map now answers its own question. APM · campus life only WHICH PARTS OF STUDENT LIFE STAND OUT APM · academic only WHICH CLASSROOMS FEEL WELCOMING WHAT CHANGED Same participants. Same kind of imagery. Same one-minute test. Each map now answers its own question. The academic signal was there in Test 1. It was being outshouted, not absent.
This is the same audience that ignored every classroom a round earlier. Given a board of classrooms, they had preferences. The signal was never missing from Test 1. It was being outshouted.

By isolating the categories, the college could distinguish not just what was broadly appealing but which particular scenes aligned with the values, aspirations and expectations of prospective students. That is the difference between a board that records a preference the room already had and a test that finds one it did not.

Refining the approach, expanding the insight

Further rounds are planned to deepen the college’s understanding of how prospective students perceive academic life, student culture and the broader campus experience. Each iteration sharpens the research focus and extends the ways the tool can frame a more useful question and draw a clearer signal from visual feedback.

The larger lesson travels. Structure is not a setup detail. It is the hypothesis. Put a loud category next to a quiet one and the test measures volume. Put each category on its own map and the test measures preference. The participants did not change between the two rounds. What they were allowed to compare did.

As the tool moves into new contexts, from corporate brand strategy to higher education, the value holds: it moves teams from assumption to evidence, and surfaces preferences and perceptions that usually go unspoken.

Frequently asked questions

Why did the first test only confirm the obvious?

Because academic and activity scenes shared the same maps, and activity scenes are louder. Every classroom was being judged next to a concert, so the responses clustered on the energetic tiles and the test could only report that teenagers prefer fun. The academic preferences existed, but the structure gave them nowhere to show up.

When should a test blend imagery across maps, and when should it separate it?

Blend when the question is about the whole: what does this audience associate with the brand? That structure surfaces core identity. Separate when the question is about the parts: which scenes inside a category win against their own kind? Mixing categories with unequal energy answers neither question well.

Who took part in the college study?

High school juniors and seniors who had attended an on-campus event or were in active contact with admissions officers. Student workers from the admissions office invited them to take the survey on iPads while they waited in line at an event. Participants typically finished in a minute or two.

Can a Constellations test be run in person rather than online?

Yes. The college study was administered on iPads by admissions staff at a live event, which put the test in front of exactly the audience it needed while they were already on campus. The people running it reported the format was intuitive and quick for participants.