How to Validate Willingness to Pay Before You Build

TL;DR

“Would you use this?” is a useless question. “Would you pay $X for this?” is slightly better. Actually getting someone to put money down — even a small deposit — is the only real signal. Here’s how to get there fast.

The Most Expensive Assumption in AI

Every AI founder makes the same bet: if we build it well enough, people will pay for it. Some are right. Most are catastrophically wrong. And the difference almost never comes down to product quality.

At IceCream Labs, we’ve seen technically brilliant products fail because the founders skipped one critical step — validating that someone would exchange money for the thing they were building. Not attention. Not praise. Money.

Why “Interest” Is a Lie

Human beings are pathologically polite when you show them a product. They’ll say “this is cool,” “I’d definitely use this,” and “send me the link when it’s ready.” These are social pleasantries, not purchase signals.

We call this the Enthusiasm Gap — the vast, treacherous distance between “I like this” and “here’s my credit card.”

Research consistently shows that stated intent to purchase and actual purchasing behavior have weak correlation. In AI products specifically, the gap is even wider because the demo effect is so powerful. AI demos are inherently impressive. But impressiveness doesn’t pay the bills.

The Willingness-to-Pay Ladder

We use a four-rung framework to move from vague interest to validated demand:

Rung 1: Problem acknowledgment. Does the prospect confirm they have the problem you’re solving? This is table stakes. If they don’t have the problem, nothing else matters.

Rung 2: Solution interest. When you describe your approach (without showing a product), do they lean in? Do they ask questions? Do they reference their own context? You’re looking for intellectual engagement, not just politeness.

Rung 3: Price anchoring. This is where most founders chicken out. Name a price. Say: “We’re thinking about offering this for $500/month. How does that feel for the value I described?” Watch their face, not their words. Flinching is data. Counter-offers are data. “Let me check with my boss” is data.

Rung 4: Commitment extraction. The gold standard. Can you get any form of tangible commitment? A signed LOI. A deposit (even $100). A confirmed pilot with a start date. A written introduction to their procurement team. The form of commitment matters less than the fact that they’re investing something — time, money, reputation — in your product existing.

If you can get 3-5 prospects to Rung 4 before you build, you have something real.

Practical Tactics That Work

The “fake door” test. Put up a landing page with pricing and a “Sign Up” or “Request Access” button. Drive traffic to it (even $500 of LinkedIn ads). Measure click-through on the pricing page, not the homepage. If people read your pricing and still click, that’s a strong signal.

The concierge MVP. Offer to solve the problem manually for 3-5 customers — using AI behind the scenes but with heavy human involvement. Charge them. Even a small amount. If they pay for the messy version, they’ll pay more for the polished one.

The pilot commitment letter. Draft a simple one-page document: “We agree to pilot [Product] for 60 days at $X/month, starting [date].” Send it to your warmest prospects. The ones who sign are your design partners. The ones who ghost you are telling you something important.

The budget question. In every sales conversation, ask: “If you decided to move forward, where would the budget come from? Is there an existing line item this would fall under, or would it require new budget approval?” This question reveals whether your product fits into existing spending patterns or requires creating a new category — which is 10x harder.

What the Numbers Need to Look Like

For an AI startup at the earliest stage, here are rough benchmarks we use:

• Out of 30 prospect conversations, at least 20 should confirm the problem (Rung 1)
• At least 10 should show genuine solution interest (Rung 2)
• At least 5 should engage seriously at your price point (Rung 3)
• At least 2-3 should give you a tangible commitment (Rung 4)

If your numbers are significantly below these, either your problem isn’t painful enough, your solution framing isn’t right, or your price is off. All of these are fixable — but only if you discover them before you’ve spent 6 months building.

The AI-Specific Pricing Challenge

AI products have a unique pricing problem: customers have been trained by free AI tools (ChatGPT, free tiers of everything) to expect AI capabilities at zero cost. This makes the willingness-to-pay bar higher for AI-specific products.

The founders who win here are the ones who price on the outcome, not the technology. Nobody pays for “AI-powered analysis.” People pay for “cut your monthly close from 10 days to 3 days.” Frame the value in their language and their metrics, and the AI becomes invisible — which is exactly where it should be.

---

At IceCream Labs, we don’t write code until we’ve validated willingness to pay. It’s slower at the start and dramatically faster in the long run. Talk to us about validating your AI startup idea.