How to Read a Perception Map

What the map is made of

The map is the output of audience perception mapping: an audience responds to images instead of describing them, and the pattern shows where the group agrees, where it splits, and why. The output has three groupings, and each one carries different information.

Positive. Images the audience responded to as right for whatever was asked: most American, most premium, most representative of the brand, most likely to make you want the drink.

Negative. Images actively marked as wrong for it. This is a real response, not an absence of one, and the distinction matters for everything that follows.

Neutral. Images that drew no strong response either way, or that split the group. These get discarded in most analyses and shouldn't be.

Figure 1 · The three groups

The clustering is the finding. People who have never met, responding individually, land in the same places. Human audiences do this without exception, and it isn't people agreeing with each other. Separate people bring a shared cultural read to the same image and independently arrive at the same place. The map is a record of that shared read.

Pass one: what the positives have in common

Start with the positive cluster, and do not start by picking a favorite from it.

The question is what those images share. Not what they look like individually, but the quality that runs across them. Sometimes that's obvious at a glance. More often the shared quality is not the one a designer would name first, because designers read images as elements and audiences read them as complete things.

Two things to watch for.

The shared quality may be functional rather than visual. In the study on what reads as American, the winners and losers used the identical visual vocabulary. Same stars and stripes, same vintage type. What separated them was whether the image stood for something or sold something. A palette-level read of that map would have found nothing.

Coincidental similarities are the trap. Twelve images assembled by one person for one board share plenty of qualities that mean nothing. Sourced similarly, cropped similarly, in a similar register. Every one of those is available as an explanation and most are wrong.

Which is why the positive cluster can't be read alone.

Pass two: the negatives as control

The negative cluster is what makes the positive cluster interpretable, and it's the pass most teams skip.

The rejected images were in the same room, on the same board, seen by the same people. Whatever separates the two piles is the only variable that moved. Everything the two piles have in common can be ruled out.

Figure 2 · The negatives as control

So the question in this pass is not "what's wrong with these." It's "what do the negatives have that the positives don't, and the reverse." Hold the two collections against each other and the dividing line appears. That line is the thing you design against for the rest of the project.

This is also the pass where a model earns its place. Give an AI one collection of images and it will describe it. Give it two contrasting collections and ask what divides them and you get a reason, and the reason is what no single image can tell you on its own.

Pass three: the middle

The neutral and contested group is the one worth slowing down on, because it holds two different situations that look identical on the map.

Genuine indifference. Nobody had a reaction. The images weren't doing anything for anyone. Usually not interesting, occasionally useful as evidence that a whole register is inert for this audience.

A real split. Half the group said yes and half said no, and the averages cancelled. This looks like indifference and is the opposite of it. It means there are two audiences in your sample reading the same image in opposing ways, and if you're about to make one brand for both of them, that's the most important thing on the page.

Figure 3 · Indifference or a split

The way to tell them apart is to segment. Turn a segment on and off and watch whether the middle resolves. If a contested image jumps to positive for one group and negative for another, it was never neutral. That's also the rule of thumb worth adopting generally: if you're even slightly curious whether a group might see it differently, segment, because segments can always be blended back together afterwards and cannot be recovered if you didn't collect them. What happens when the segments disagree is its own decision, and it's a strategy decision rather than a design one.

The Halloween study on spooky against scary produced a third possibility. Its neutral group turned out to be coherent in its own right, dark subject matter handled with intellectual restraint, occult diagrams and mythic studies, reading as neither comforting nor threatening. Not indifference and not a split. A third category the question hadn't anticipated.

What to do with what you find

The map is input to the creative work, not a replacement for it. Nobody should be picking the top-scoring image and building a brand on it.

Name the quality, not the image. The deliverable from a map is a sentence: the audience sorted on X. That sentence is what carries forward into work they've never seen.

Make the shared alignment core. Where every segment agrees, that becomes central to the brand. Where segmentation shows a difference, mark it and decide whether to lean toward one audience, which is a strategy question rather than a design one. It's what lets you make global decisions about a brand and separate, targeted decisions inside a campaign.

Treat the map as the north star for the rest of the project. The signal is what the creative direction gets built on top of. It also becomes the reference point for every review conversation afterwards, and those conversations are where most of its value gets paid back. If someone wants to move away from what the audience converged on, they now need evidence rather than preference.

Reading time

A map takes about five minutes to read and a few days to stop arguing about, which is the reverse of how creative reviews usually go.

The five minutes is the three passes above. The few days is everyone in the room absorbing that a question they'd been debating for weeks has an answer, and that the answer isn't any of the versions they were defending. That part is uncomfortable and it's the point. Everyone has agreed, and they know why, because they can see why.

Frequently asked questions