When to Use

You’re looking for startup ideas and want a systematic method for finding real demand (not stated demand). Or you’re evaluating a startup idea and want to understand whether the customer need is genuine or just something people say they care about. Or you’re doing market research and want to avoid the classic trap of building for what people tell you they want rather than what they’ll actually pay for.

The Framework

The Core Insight

Balaji identifies a specific, repeatable pattern for generating startup ideas:

“The gap between stated preference (what is praised) and expressed preference (what is bought) is an inexhaustible source of startup ideas.” — Balaji Srinivasan, The Anthology of Balaji

Stated preference: What people say they value. What they praise publicly. What they tell researchers in surveys. What they advocate for on social media.

Expressed preference: What people actually do. What they buy. Where they spend their time. What they click on. Where they move to.

The gap between these two is not a flaw in human nature to be condemned. It’s an opportunity to be exploited:

“You can condemn hypocrisy. Or you can arbitrage inconsistency.” — Balaji Srinivasan, The Anthology of Balaji

Why the Gap Exists

The gap between stated and expressed preference exists for several structural reasons:

1. Social desirability bias. People say they want what’s socially approved. They buy what they actually want. Example: People say they want to eat healthy. Fast food revenue grows every year.

2. Aspirational vs. actual identity. People describe their aspirational self in surveys and social media. Their credit card statements reflect their actual self. Example: People say they value experiences over things. Amazon’s revenue says otherwise.

3. Cost blindness. People state preferences without considering cost. When they have to pay, their expressed preferences shift dramatically. Example: People say they’d pay more for sustainable products. Market share of sustainable products grows slowly.

4. Information asymmetry. What people say they want reflects their public information. What they actually buy reflects their private information (including things they’ve discovered through personal experience that they wouldn’t share publicly). Example: People publicly praise neighborhood diversity. Private school enrollment patterns tell a different story.

5. Signaling vs. consumption. Stated preferences are signals to other people. Expressed preferences are consumption for oneself. The two serve different functions and don’t need to align. Example: People share articles about minimalism on social media. The self-storage industry is a $48 billion market.

The Arbitrage Method

To use this framework for idea generation, follow these steps:

Step 1: Identify a strong stated preference. Look for something that a large number of people vocally support. The stronger and more unanimous the stated preference, the more likely there’s a gap.

Step 2: Measure the expressed preference. Look at actual behavior data: revenue figures, market share, usage statistics, migration patterns, time spent, money spent. Not surveys, not social media sentiment, not focus groups.

Step 3: Identify the gap. If stated preference diverges significantly from expressed preference, you’ve found the gap.

Step 4: Determine which side to build for. This is the critical decision. You can either:

  • Build for the stated preference (risk: people say they want it but won’t pay)
  • Build for the expressed preference (risk: people buy it but won’t praise it)
  • Bridge the gap (build something that lets people satisfy their expressed preference while maintaining their stated preference identity)

The third option - bridging the gap - is often the most valuable. These are products that let people have their cake and eat it too: satisfy their actual desires while maintaining their public-facing values.

Examples Across Domains

DomainStated PreferenceExpressed PreferenceThe GapOpportunity
Food”I eat healthy”Fast food, delivery, convenience food revenue growsPeople want healthy food that’s as convenient as junk foodHelloFresh, meal prep services
Media”I value quality journalism”Most clicks go to sensational headlines, celebrity gossip, outrage contentPeople want to feel informed, not actually be informedNewsletters that make readers feel smart with minimal time investment
Privacy”I care deeply about privacy”Voluntary adoption of smart speakers, social media oversharing, location sharingPeople value convenience more than privacyPrivacy tools that don’t sacrifice convenience
Environment”I want to reduce my carbon footprint”SUV sales at record highs, air travel growingPeople want environmental virtue without lifestyle sacrificeCarbon offsets, electric SUVs, “green” luxury goods
Education”Every child deserves equal education”Private school enrollment, residential sorting by school districtPeople want the best education for their own childrenEducational tools that deliver private-school quality at public-school prices
Work”I want work-life balance”Ambitious people work more hours than ever, checking email 24/7People want the appearance of balance while actually maximizing career outcomesTools that increase output per hour (not reduce hours)
Social media”Social media is toxic, I should use it less”Screen time increases year over yearPeople want the benefits of social media without the guiltCurated feeds, “slow social media,” digital wellness tools that don’t actually limit access

Connection to the Idea Maze

The stated vs. expressed preference gap is a specific type of “open door” in the idea maze. Previous startups may have failed because they built for stated preferences (what customers said in surveys) rather than expressed preferences (what customers actually paid for).

This connects directly to Balaji’s broader framework:

  • The idea maze tells you what paths have been tried before
  • The preference gap tells you why some paths failed (they built for stated preferences)
  • The physical-to-digital spectrum tells you where the gap is likely to be largest (industries at the “scanned” stage where stated preference is for the old way but expressed preference reveals frustration)

The Political Application

Balaji applies this framework beyond business to politics and governance. The gap between stated preference and expressed preference in politics is enormous:

  • Stated: “I support local small businesses.” Expressed: Amazon Prime membership.
  • Stated: “I want to live in a diverse community.” Expressed: Residential sorting patterns.
  • Stated: “I think tech companies are too powerful.” Expressed: Daily usage of Google, Apple, Amazon, Meta products.

This political application feeds directly into the network state thesis. When the gap between what people say they want from governance and what they actually do (vote with their feet, opt out of institutions, use crypto) becomes large enough, it creates demand for new governance models. The network state is, in some sense, the ultimate preference gap arbitrage: people say they’re satisfied with existing governance, but their expressed preferences (emigration, tax optimization, regulatory arbitrage, crypto adoption) reveal dissatisfaction.

Warnings

Don’t confuse this with cynicism. The point isn’t that people are hypocrites. The point is that human behavior is complex and that stated preferences are a poor predictor of behavior. Survey data, focus groups, and social media sentiment are useful signals, but they must be checked against revealed behavior.

Don’t build something people are ashamed to use. The most successful preference-gap companies don’t embarrass their customers. They bridge the gap by letting customers satisfy expressed preferences in a way that’s compatible with their stated values. Tesla bridges the gap between “I care about the environment” (stated) and “I want a fast, luxurious car” (expressed) by making the environmental option also the desirable option.

The gap can close. Culture shifts over time, and today’s expressed preference can become tomorrow’s stated preference. The company that bridges the gap accelerates this cultural shift. But if you’re betting on a gap that’s rapidly closing (because culture is catching up to behavior), you may be building for a problem that’s solving itself.

Example

Applying the framework to remote work (2020-2026):

Stated preference (2020-2022): “Remote work is the future. I never want to go back to an office. Location independence is everything.”

Expressed preference (2023-2026): Many knowledge workers who moved to lower-cost areas during COVID have returned to or near major cities. Companies that offered fully remote have added hybrid or in-person requirements. Office occupancy in major cities has rebounded.

The gap: People stated they wanted full remote, but their expressed preferences revealed they wanted flexibility and optionality, not permanent isolation from colleagues and urban amenities.

The opportunity: Products and services that deliver the benefits of remote work (flexibility, no commute, focus time) without the costs (isolation, career stagnation, weakened relationships). This is the “bridging” play: hybrid tools, co-working spaces, popup gatherings for distributed teams.

Output

After reading this framework, you should be able to:

  1. Identify stated vs. expressed preference gaps in any market
  2. Use behavioral data (not survey data) to identify real demand
  3. Decide whether to build for stated preference, expressed preference, or bridge the gap
  4. Apply the framework to political and governance contexts
  5. Connect preference gaps to idea maze analysis for deeper startup evaluation

Source: The Anthology of Balaji p. 196