INNOVATIONMonths to result

The Unspoken Customer Desire Method

Discover what people deeply want but cannot articulate

Problem it solves

stagnant innovation

Best for

Product teams and researchers frustrated by focus groups that produce misleading data, who need methods to uncover genuine customer preferences that self-report conceals.

Not ideal for

Businesses with well-understood markets where customer preferences are already clearly mapped and segmented.

Overview

Why this framework exists

The Unspoken Customer Desire Method addresses a fundamental flaw in traditional market research: people do not know what they want, and even when they do, they cannot articulate it. Howard Moskowitz proved this through decades of food research—when asked, people describe the version of themselves they want to be (dark, rich coffee drinker) not who they actually are (milky, weak coffee drinker).

The method replaces asking with testing. Instead of focus groups that ask 'what do you want?' it creates dozens of product variations and measures actual preference through blind testing. The data reveals truths that self-report conceals: one-third of Americans craved extra-chunky spaghetti sauce, but in 20 years of focus groups, no one ever mentioned it.

The broader philosophical point is that self-knowledge is limited. We construct narratives about our preferences that serve our self-image rather than reflecting our actual experience. Only by systematically exposing people to variations and measuring response can we discover what they truly want—and often, the biggest opportunities hide in the gap between what people say and what they actually prefer.

Core principles

4 total
  1. The mind knows not what the tongue wants—self-report is unreliable for preference discovery
  2. People describe the version of themselves they want to be, not who they actually are
  3. Testing beats asking every time for uncovering genuine preferences
  4. The biggest opportunities hide in the gap between stated and revealed preference

Steps

3 steps
  1. Create a Wide Variation Test
    Instead of asking customers what they want, create the widest possible range of product or service variations. Moskowitz created 45 spaghetti sauces varying along every dimension. The key is breadth—you need enough variation to capture preference landscapes you cannot even imagine yet. Budget for testing at least 15-20 meaningfully different variations.
    Pro tipVary along dimensions that seem unusual or unlikely. Extra-chunky was not an obvious dimension to test, but it revealed the largest unserved market.
    WarningDo not pre-filter variations based on what you think customers want. The whole point is that you do not know what they want—let the data reveal it.
  2. Measure Revealed Preference, Not Stated Preference
    Have people experience the variations and measure their actual response (ratings, behavior, repeat selection) rather than asking what they think they want. The gap between stated and revealed preference is where the biggest insights hide. Most people say they want dark coffee but most actually prefer milky—only revealed preference testing catches this.
    Pro tipBlind testing removes the social desirability bias that corrupts stated preference. People rate what they actually experience rather than what they think they should prefer.
    WarningEnsure sample sizes are large enough to detect clusters. Small samples may show noise that looks like clusters or miss real clusters that only emerge with more data.
  3. Look for Clusters, Not Averages
    Analyze the data for natural groupings rather than central tendencies. The average preference in a room of coffee drinkers might be 60 out of 100—satisfying nobody. Three coffee clusters, each scored at 78, represent dramatically different experiences. Use cluster analysis, not mean analysis, as your primary tool.
    Pro tipThe difference between 60 and 78 on a satisfaction scale is the difference between a product that makes people wince and one that makes them deliriously happy. That gap is the entire value of this method.

Checklist

Saved in your browser

Examples

1 cases
The Diet Pepsi Mess That Started Everything

Pepsi asked Moskowitz to find the perfect sweetness between 8 and 12 percent. He tested systematically and got a mess—no bell curve. Most researchers would have picked the middle and moved on. Moskowitz refused. Years later, the insight struck: the mess meant there were multiple perfect sweetness levels for different groups, not one universal optimum.

OutcomeThis realization led to the entire horizontal segmentation revolution in the food industry and eventually to the extra-chunky Prego discovery that generated $600 million.
Malcolm Gladwell, TED Talk 2004

Common mistakes

2 traps
Trusting Focus Group Feedback
Twenty years of focus groups never surfaced the extra-chunky insight. People in focus groups perform—they describe the sophisticated version of their preferences rather than their actual messy reality. Use focus groups for understanding language and framing, not for product direction.
Averaging Across Distinct Clusters
When you average across three distinct preference clusters, you get a compromise that satisfies none of them. The average hides the truth. Always look for clusters first and averages second.

Origin story

How this framework came to be

Moskowitz spent years frustrated by the messy Pepsi data before the insight struck: there was no single answer because there was no single question. When Campbell's hired him for Prego, he deliberately designed 45 variations testing every conceivable dimension—not because he expected all 45 to be viable, but because he needed the full landscape to see where clusters would emerge. The extra-chunky discovery validated the entire method: a massive unserved market hiding in plain sight, invisible to traditional research.

Source

Traced to primary
Source · VIDEO
Choice, Happiness, and Spaghetti Sauce
Malcolm Gladwell · 2004
Open source →

Related frameworks

Browse all Innovation →