Building an MVP to test demand is, in 2026, almost always the wrong move. It anchors you emotionally to a specific solution before you've proven the problem matters. Run these seven experiments first. Each one is designed to take a weekend or less.
1. The pre-sell landing page
Build a single page describing the product as if it exists. Add a "Buy now — $49" button that, when clicked, opens a Stripe checkout. Drive 200 targeted clicks from a relevant community. Anyone who actually pays gets a refund and a personal thank-you. Anyone who clicks "buy" and bounces still tells you something.
2. The waitlist with a question
Same page, but instead of buying, the user enters their email and answers one question: "What's the worst part of [problem] for you right now?" The replies are gold. They tell you whether the problem is real and what language to use in your marketing.
3. The manual concierge
Don't build the product. Do it manually for 10 people. If your idea is "AI tool that writes outreach emails for sales reps," spend a weekend writing emails by hand for 10 reps. If they pay, the demand is real. If they ghost, the demand isn't.
4. The cold-outreach interview
Cold-DM 30 people in your target niche. Don't pitch. Ask about their workflow. The good signal isn't "yes I'd buy that," it's "where have you been all my life" or detailed unprompted complaints. Cross-reference with hair-on-fire signals.
5. The fake door ad
Run a $50 LinkedIn or Reddit ad to a single landing page. Measure click-through and email signups. Compare against benchmark ads in your niche. Cheap, fast, and it tells you whether your positioning lands.
6. The competitor-pricing probe
Sign up for the closest competitor's free trial. Read their reviews. Find the 2-star reviews. Those reviews describe your wedge. Detailed playbook in Competitor Research in 30 Minutes.
7. The 5-person Loom
Record a 3-minute Loom walking through a clickable Figma mock of your idea. Send it to 5 target users with one question: "Would you pay for this?" Their facial reactions, recorded as a return Loom, are the most honest data you'll ever get.
How to interpret the results
If 3+ of the 7 experiments give a strong positive signal, build. If you get polite-but-soft signals across the board, you have a nice-to-have, not a hair-on-fire. Don't build until you can sharpen it.
Use the ShiporDrop quiz to lock in which dimensions you're weakest on, then run the experiments most likely to move those scores.