5 Times Customers Said "I'd Buy It" — and Nobody Did

In interviews, 9 customers told me, "I'll definitely use this once it's out." I launched it. 0 of the 9 paid. They weren't lying. That's simply how people answer.

The most expensive mistake in user research isn't skipping research. It's doing the wrong research and coming away confident. You spend months building on the grounds that "customers said they wanted it," and then face a result nobody uses.

The core point is simple. What people say and what people do are different. Good user research is the skill of closing that gap. Today we look at moments when that gap came to light, and at methods a solo founder can use as they are.

Trap 1. "It would be nice" is not "I'll buy it"

In interviews, "A feature like that would be handy" comes up almost every time. Nobody turns down something better for free. But there is a canyon between "that would be handy" and "I'll open my wallet."

The test is simple. Ask about past behavior, not future intent.

• Bad question: "Would you use something like this?" (→ everyone says "yes")
• Good question: "The last time this problem came up, what did you actually do?" (→ real behavior)

A question for your business: Of all the "that would be nice" comments I've heard, how many are backed by behavior where the customer actually spent money or time?

Trap 2. Customers don't know the solution — they only know the problem

A common beginner mistake in research is asking customers, "What features do you need?" Customers are experts in their own problems, not in solutions. Designing the solution is our job.

There's an often-quoted analogy: ask people what they want and they'll say "a faster horse." The anecdote's source is unclear[^1], but the truth it points to is clear. Customers express their needs only within the range of what they already know.

Method: Don't ask about features — dig into the problem and its context. "What frustrated you most at that moment?" is 100 times more useful than "What features do you want?"

Trap 3. The people who watched behavior instead of words won

In Airbnb's early days, the founders didn't send out surveys. They flew to New York, borrowed a camera, visited hosts' homes one by one, and photographed the listings themselves. The real bottleneck wasn't traffic but 'trust (photo quality)', and that surfaced in the field, not in interviews. Bookings for New York listings switched to professional photos rose 2–3x, and New York revenue doubled that month.[^2]

Slack is even more dramatic. It started as a game company (Glitch), and the game failed. But the team kept using the tool they had built for internal communication. The actual behavior of its users (themselves) pointed to the product.[^3]

In both cases, they looked not at "What do you want?" but at **"What are they actually doing?"**

A question for your business: Among the things my customers (or I myself) are already doing, is there a clue to a product hiding?

Trap 4. 5 people are enough (but the right 5)

Many people put off research, saying "the sample is too small to mean anything." Yet most usability problems surface even with a handful of interviews. Nielsen Norman Group's classic recommendation is that 5 users in a usability test uncover about 85% of problems.[^4]

For a solo founder, this is liberating. Even without the resources to survey hundreds of people, you can look deeply at 5 well-chosen people. There is one condition — those 5 must be your real target users. Friends and acquaintances are not a sample.

Method: Watching the behavior of 5 real customers in depth for an hour each beats a survey of 200.

Trap 5. Ask only confirming questions, and you get only confirmation

People love their own ideas. So they unconsciously ask questions that steer toward "yes, you're right." Ask "Isn't this great?" and the customer will politely say "It's nice." In that moment, we've bought the answer we wanted to hear.

Good research deliberately creates chances to be wrong. It asks questions that could break your hypothesis.

A question for your business: In my last customer conversation, did I ask even one question that could have shown my idea was wrong?

One sentence that runs through all 5 traps

The heart of user research is not "asking customers for the answer." It is reading the answer from customers' behavior.

• Past behavior instead of intent
• Problem and context instead of solutions
• Field observation instead of surveys
• 5 real people instead of a large sample
• Chances to disprove instead of confirmation

For a solo founder, this is actually an advantage. Large companies rely on massive surveys and panels and can't look deeply at a single customer. We can spend an hour with one real customer as soon as tomorrow.

If you do just one thing today

In your next customer conversation, change one question.

• ❌ "Would you use something like this?"
• ✅ "The last time that problem came up, what did you actually do?"

This one-line difference separates "the answer you want to hear" from "the real answer." If you'd like to check together whether your research questions are only digging for confirmation—

Book a free consultation → We'll spend 30 minutes taking apart your customer interview questions together.

---

Sources (✅ 4/4 verified, WebSearch 2026-06-01)

[^1]: There is no primary evidence that "faster horses" was said by Henry Ford (unsourced, misattributed). It does not appear in Ford's autobiography or among the Henry Ford Museum's 200 verified quotes, and the earliest attribution dates to 2001. — HBR "Henry Ford, Innovation, and That 'Faster Horse' Quote" (2011) · Snopes · Quote Investigator. https://hbr.org/2011/08/henry-ford-never-said-the-fast [^2]: Airbnb's founders photographed listings themselves in New York → bookings 2–3x, New York revenue doubled that month. The bottleneck was trust, not traffic. — BU Boston Hospitality Review "The Making of Airbnb" · Paul Graham "Do Things That Don't Scale". https://www.bu.edu/bhr/2016/01/08/the-making-of-airbnb/ [^3]: Slack began as the internal messaging tool of Glitch (Tiny Speck), a failed game — the team's behavior of using it "nonstop" pointed to the product. Launched in 2013, later sold to Salesforce for USD 27.7 billion. — Building Slack · Startup Archive (Ben Horowitz). https://buildingslack.com/the-death-of-glitch-the-birth-of-slack/ [^4]: 5 users in a usability test uncover about 85% of problems (per-user discovery rate L≈31%, Nielsen & Landauer model). — Nielsen Norman Group, Jakob Nielsen, "Why You Only Need to Test with 5 Users" (2000). https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/

This article is STAR-T's flagship piece for the user research sector. All 4 figures and anecdotes were verified against primary or authoritative sources.

---

If you're wondering whether you need to rethink your customer research from the start,
a 2-minute diagnostic will show you where to add structure first. Feel free to answer only as much as you're comfortable with.

STAR-T AI Business Operations Diagnostic →