You: "Would you use a feature that lets you export your data to CSV?"
User: "Yeah, sure."
You: writes down "Strong demand for CSV export"
Narrator: The user would never, ever use CSV export.
Welcome to the free ice cream problem.
What Is Free Ice Cream?
Here's the thought experiment:
Walk up to random people on the street and ask: "Would you like some free ice cream?"
Almost everyone says yes. Even people who:
- Don't particularly like ice cream
- Just ate lunch
- Are lactose intolerant
- Are diabetic
- Have no way to eat it right now
- Weren't thinking about ice cream until you asked
Why? Because it's free and delicious. There's no apparent downside. Saying yes costs nothing. So why not?
This is exactly what happens in user research when you ask about features without context or tradeoffs.
The Pattern in Product Research
You're conducting user interviews. You want to validate feature ideas. You ask:
- "Would you use dark mode?"
- "Would you like to be able to share this with friends?"
- "Should we add a calendar integration?"
- "Would notifications be helpful?"
- "Do you want more customization options?"
Users respond: "Yeah, sure." "That could be useful." "Why not?" "Yeah, I guess so."
You tally up the yeses. You go back to your team: "Users want all of these features!"
Six months later, you ship them. Usage: 3%.
You've been offering free ice cream.
The Tell: The Unenthusiastic "Yeah"
Learn to recognize the sound of free ice cream acceptance:
- "Yeah, sure" (with a shrug)
- "I could see myself using that" (won't)
- "That might be nice" (doesn't care)
- "Why not?" (has no actual use case)
- "I mean, more options are always good" (won't use the options)
- "Yeah, I guess" (is being polite)
Compare this to genuine demand:
- "Oh my god, YES. I've been trying to do this for months!"
- "Wait, you can do that? When is this shipping?"
- "I would use that literally every day"
- "That's exactly what I need"
The difference between "yeah, sure" and "holy crap yes please" is the difference between free ice cream and solving a real problem.
Why Users Do This
Users aren't lying. They genuinely think they might use the feature. They also have zero cost associated with the choice so there's no immediate downside. Here's why they're wrong:
1. They Don't Consider Opportunity Costs
When you ask "Would you use feature X?" users hear: "Would you use feature X if it magically appeared, fully formed, with no trade-offs?"
They don't think: "Would I use feature X enough to justify it being built instead of features Y and Z, or making the app more complex, or delaying other improvements?"
You're asking them to evaluate a free lunch. Of course they say yes. From the Facebook poster department, one of my favorite sayings: It's only prioritization if it hurts.
2. They Imagine Their Best Case Scenario
User's mental model: "Would I use CSV export?"
What they imagine: "One day I might need to analyze my data in Excel, and having CSV export would be convenient."
Reality: They've never opened Excel for personal data analysis in their life and never will.
People are optimists about their future behavior. They imagine a version of themselves who is more organized, more analytical, and more likely to use advanced features than they actually are.
3. People Want to Be Helpful
You're sitting there with a clipboard or a prototype, clearly hoping for positive feedback. Users want to be nice. They want to help you. You are probably going to give them a Starbucks gift card in twenty minutes.
Saying "no, I wouldn't use that" feels negative and unhelpful. Saying "yeah, sure" feels supportive and constructive.
They're being polite, not dishonest.
(Note that this is not universal, though common in most parts of North America. I've worked with Eastern Europeans, Germans, and people from the northeast of the USA, and generally they will tell you that your free ice cream sucks.)
4. They Can't Predict Their Future Behavior
Humans are terrible at predicting what they'll actually do. We're decent at describing what we've done in the past. We're horrible at forecasting future behavior. Look at how many people will not buy an EV because they want the optionality of taking their family on a rally through the Kalihari desert one day where there aren't any charging stations.
"Would you use this?" is asking users to predict the future. They can't. Neither can you. But they'll give you an answer anyway because you asked.
How to Avoid the Free Ice Cream Trap
1. Ask About Past Behavior, Not Future Intent
Bad: "Would you use a feature to export your reading history?"
Good: "Tell me about the last time you wanted to do something with your reading history."
If they can't describe a recent, specific instance where they needed this, they probably don't need it.
2. Force Trade-Offs
Bad: "Would you like dark mode?"
Good: "We can build either dark mode or improved search this quarter. Which would you use more?"
When users have to choose, you see real priorities.
3. Make It Real
Bad: "Would notifications be useful?"
Good: "Here's a prototype with notifications. Use it for a week and tell me if you kept them enabled."
Actual usage data beats hypothetical responses every single time.
4. Look for Workarounds
Bad: "Would this feature solve your problem?"
Good: "How are you solving this problem today?"
If users have created elaborate workarounds, you've found real demand. If they shrug and say "I don't, really," it's not a real problem.
5. Listen for Energy
Pay attention to emotional intensity, not just the words:
- Excitement = real demand
- Enthusiasm = probably real demand
- "Yeah, sure" = free ice cream
- Indifference = free ice cream
- "I guess" = definitely free ice cream
The energy level tells you more than the yes/no answer.
6. Ask What They'd Pay For
Bad: "Would you use this feature?"
Good: "Would you pay $5/month for this feature?"
Suddenly all the free ice cream disappears and you see what people actually value.
The A/B Testing Corollary
This problem exists in A/B testing too. You ship a feature because users said they wanted it. Usage is terrible. You conclude: "Users don't know what they want."
No. Users accurately told you they'd take free ice cream. You incorrectly interpreted that as product-market fit.
Before building, ask:
- How often did they request this organically (without being prompted)?
- What workarounds are they currently using?
- How much energy/emotion is behind the request?
- Would they pay for it?
- Can they describe a specific recent instance where they needed it?
When "Yes" Actually Means Yes
To be clear: not every positive response is free ice cream. Real demand exists. Here's how to recognize it:
They bring it up unprompted. You didn't ask about the feature—they complained about its absence.
They describe specific, recent pain. Not "I could imagine needing this" but "Last Tuesday I spent an hour trying to..."
They've built workarounds. They're exporting to CSV, manipulating in Excel, and importing back because you don't have the feature they need.
They show emotion. Frustration, excitement, relief—strong feelings indicate real problems.
They're willing to sacrifice. "I'd trade feature X for feature Y in a heartbeat" reveals true priorities.
Conclusion: Stop Offering Free Ice Cream
The free ice cream problem is everywhere in product development:
- User interviews where people agree to features they'll never use
- Surveys where every option gets rated "somewhat important"
- Feature requests that sound good but solve no real problem
- A/B tests where you optimize for stated preferences instead of actual behavior
The solution isn't to ignore users—it's to ask better questions.
Stop asking if they want features. Start asking about their problems, their behavior, their workarounds, their frustrations.
Stop offering free ice cream. Start looking for people who are desperately hungry and willing to pay for a meal.
The difference between "yeah, sure" and "oh my god yes please" is the difference between wasted development time and product-market fit.
Listen for the energy. Demand the specifics. Force the trade-offs.
And when someone says "yeah, sure" to your feature idea?
That's not validation. That's just someone grabbing the sample at Trader Joe's while they shop for other food.
What features have you built based on "yeah, sure" responses that nobody used? How do you separate free ice cream from genuine demand?