One sign, three readers.
Fresh Harvest, an online farmers market in Atlanta, needed a sign for festival booths. We put 93 signs in front of each of 312 local shoppers, a few seconds at a time. The average answer was useful. The spread was worth more: there are three kinds of reader, and each one looks at a different part of the sign.
in 100 would stop at the best sign for everyone. The baseline, with nothing in particular on the sign, is 50.
kinds of reader. Each one is nearly blind to the parts of the sign the other two read.
the same headline, "Visit our booth for 50% off", with two kinds of reader.
Key takeaways
- Lead with an offer. "Visit our booth for 50% off" and "Win a free month of groceries" each add ten points for everyone. No other headline adds more than five.
- Show food, not graphics. The farm photo adds seven points and the produce photo six. The delivery graphic adds one.
- There are three readers. 121 people read the headline, 105 read the small print, 86 read the call to action. The same headline is worth +25 to the first group and −2 to the second.
- One sign can carry all three. The sign built for everyone scores 86, 80 and 62 with the three readers. Tailored signs would add one, six and five points.
- The signs taught something false. Three in five people came away believing Fresh Harvest charges a subscription fee. It does not. The small print has a second job.
The question
An A-frame sign stands beside a booth at an outdoor market. People walk past it. Does it make them stop?
Fresh Harvest had sixteen headlines, sixteen lines of small print, sixteen calls to action, eight images and six background colours. Any sign is one of each. There are far more possible signs than anyone could print, and the team had opinions about most of them.
We asked one question of every person who took part: Imagine you are walking through a local outdoor market or festival. How likely are you to stop at a booth after seeing this sign? Five answers, from keep walking to definitely stop.
How the study ran
312 people in metro Atlanta took part in August 2026: 209 women and 103 men, 134 living inside the Perimeter and 178 outside it. Two in three go to outdoor festivals at least a few times a year. Fewer than half had heard of Fresh Harvest.
Each person saw 93 complete signs, one after another, and rated each one. The signs were assembled by design, so that every part appeared alongside every other part about equally often, and no two people saw the same set. A rating took about three seconds. Nobody was asked what they thought of a headline.
A regression then pulls the parts back out. Every element gets one number: how much it raises or lowers the chance that a person says they would stop, when it is on the sign. The baseline, the chance of a yes with nothing in particular on the sign, is 50 in 100. The score of any sign is the baseline plus its parts. The same arithmetic runs again for every group of people we can name.
What everyone said
The average is clear, and it is the answer most teams would stop at.
Offers win. "Visit our booth for 50% off" and "Win a free month of groceries" each add ten points. Every other headline adds between zero and five. Food wins. The farm photo adds seven, the produce photo and the box photo six each, the grocery photo five. The graphics add one or two. The small print matters less than the headline, but the best line, "Win a free month of groceries!", adds six. The best call to action adds three. The six background colours score within one point of each other, so colour is the designer's call.
Put the best parts together and 78 in 100 would stop. That is a good sign. It is also an average, and an average is a picture of nobody.
Three readers
Instead of averaging the 312 people, we grouped them by how they responded. People whose scores rise and fall together belong together. Three groups came out, and they are not about age, gender or where people live. They are about where the eye goes.
121 people read the headline. For them every headline is positive, from +10 to +25. The small print barely registers. The calls to action are negative, every one of them, from −4 to −13.
105 people read the small print. For them every headline is zero or negative, down to −13, and the small print carries the sign: "Win a free month of groceries!" is worth +17. The calls to action help a little.
86 people read the call to action. For them the line at the bottom does the work, +6 to +11, the small print hurts, and the headlines are mild.
Each reader sees one layer of the sign and looks through the others. Howard Moskowitz called groups like these mind-sets. The average blends them into a person who does not exist.
The sign each reader wants
Build the best sign for each reader from the parts that reader rates highest. Select a reader to see their sign and its parts.
| Part | Their pick | Adds |
|---|---|---|
| 87 in 100 would stop |
The three signs share a line of small print and little else. The headline readers' sign and the call-to-action readers' sign share a headline. The small-print readers would take a headline that the headline readers rate at zero.
The contradiction
Put three parts beside the three readers and the map disagrees with itself.
"Visit our booth for 50% off" is worth +25 to the headline readers, −2 to the small-print readers and +7 to the call-to-action readers. "Win a free month of groceries!" as small print is worth +7, +17 and −8. "Seasonal eating made easy" as a call to action is worth −10, 0 and +11.
The averages of those three parts are +10, +6 and 0. The average says the call to action does nothing. The map says it is worth eleven points to one person in four and costs ten with one in three. A part that does nothing on average can be the most important part on the sign.
A contradiction is not a problem with the data. It is the data saying who is who.
Women and men
Men respond to headlines. Almost every headline moves men between +6 and +13, and most move women by zero. "Guess the price of our box to win" moves men thirteen points and women two.
Women respond to the offer and the food. "Visit our booth for 50% off" moves women twelve points, the most of any headline with any group. The produce photo moves women eight points and men four. The grocery photo moves men nine and women four.
Two signs come out of that. The best sign for women, with the discount headline and the produce photo, scores 82 with women. The best sign for men, with the guessing game and the grocery photo, scores 83 with men.
Age
Some parts flip sign with age. "Organic groceries, fair prices" adds seven points with people aged 35 to 54 and takes ten away with people aged 55 to 64. "Prefer big grocery chains? Keep walking" adds six with people in their late thirties and early forties, and costs eight with people over 55.
Others grow with age. "Win a free month of groceries" adds two with people aged 25 to 34 and twenty with people aged 45 to 54. The best sign for people aged 45 to 54 scores 94, the highest of any sign with any group.
What the signs taught
After the ratings, we asked what people now believed about Fresh Harvest. The signs had taught them things, and not all of them were true.
189 of 312 believed Fresh Harvest charges a subscription fee. It does not. 266 believed every order can be customised, which is true. 158 believed it is a big national brand, and 154 did not. Asked what Fresh Harvest is, 129 said an online farmers market with home delivery, 106 said a local grocery store or farmers market, and 57 said a weekly produce box from one farm.
A sign has two jobs. The headline stops people. The small print has to tell them what they stopped for. Here the small print was chosen to lift the rating, and it did. It did not say there is no fee, and three in five people assumed there was one.
The decision
Three readers do not need three signs. The sign built for everyone carries a strong headline for the headline readers, a strong line of small print for the small-print readers, and a call to action the call-to-action readers can accept. With them it scores 86, 80 and 62. Signs tailored to each would score 87, 86 and 67. One point, six points, five points. Not worth three signs.
Women and men, and the ages, are a different kind of split. Fresh Harvest took a sign for each audience, built from the parts that audience rated highest, with colour left to the designer. Every one of those signs carries the three-reader rule: a headline the headline readers like, small print the small-print readers like, a call to action that does no harm.
And every one of them now has a line of small print with a second job: say, in five words, that there is no fee.
Bring the sign.
Or the pack, the logo, the name, the price. We will design the study around it, put it in front of the right people, and show you who reads what.
Notes and references
- Fieldwork: metro Atlanta, August 2026. 312 people, 29,016 ratings, 93 signs each. Five-point scale; scores are the chance of a top-two answer, with a baseline of 50. Element scores are rounded to whole points, so the parts of a sign may not sum exactly to its score.
- Mind-sets are found by clustering people on their element scores, three clusters. Group sizes: 121, 105, 86.
- Moskowitz, H. R., & Gofman, A. (2007). Selling Blue Elephants. Wharton School Publishing.
- Krantz, D. H., Luce, R. D., Suppes, P., & Tversky, A. (1971). Foundations of Measurement, Vol. I. Academic Press.
- Stevens, S. S. (1957). On the psychophysical law. Psychological Review, 64(3), 153–181.
- MindSurf (2026). Separate to see. Combine to make.