{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-06-07T08:39:54.792Z"},"content":[{"type":"documentation","id":"6adaf54c-27c9-416e-beee-5ff3caacc741","slug":"fake-door-testing-guide","title":"Fake Door Testing (Painted Door Test): Validate Demand Before You Build","url":"https://www.koji.so/docs/fake-door-testing-guide","summary":"Fake door (painted door) testing shows users an entry point to a feature that does not exist yet and measures click-through to gauge real demand before building. It directly attacks the 35% of startups that fail from no market need and the 64-80% of features that go unused. Its weakness is that clicks reveal how many want a feature but not why. Koji closes that gap by auto-interviewing everyone who clicks using AI-moderated interviews, structured questions (six types), and automatic thematic analysis — turning a demand signal into actionable insight about what to build.","content":"**Fake door testing (also called a painted door test) is a validation method where you show users an entry point to a feature that does not exist yet — a button, menu item, or pricing tier — and measure how many people try to use it.** The clicks reveal genuine demand before you invest engineering time, and a short follow-up turns each interested user into qualitative insight about *why* they wanted it.\n\nThe principle is older than the name: do not build it until you have evidence people want it. A fake door is the cheapest possible experiment to generate that evidence.\n\n## Why fake door testing matters\n\nBuilding the wrong thing is the most expensive mistake in product development, and it is shockingly common:\n\n- [CB Insights](https://www.cbinsights.com/research/report/startup-failure-reasons-top/) found that roughly **35% of startups fail because there was no market need** for the product.\n- The [Standish Group's CHAOS research](https://www.mountaingoatsoftware.com/blog/are-64-of-features-really-rarely-or-never-used) reported **64% of software features are rarely or never used**, and a later [Pendo analysis](https://www.antmurphy.me/newsletter/why-you-should-remove-features) put that figure at **80%**.\n- The canonical example: before building Dropbox, Drew Houston posted a short demo video of a product that did not fully exist. As documented in Eric Ries's *The Lean Startup*, the [beta waitlist jumped from 5,000 to 75,000 sign-ups overnight](https://www.shortform.com/blog/dropbox-mvp-explainer-video/) — validating demand before a single sync feature was finished.\n\nA fake door test is the operational answer to those numbers. Instead of shipping a feature and hoping, you spend a day building a convincing entry point and let real behavior tell you whether to proceed.\n\n> \"The only way to win is to learn faster than anyone else.\" — Eric Ries, *The Lean Startup*\n\n## How fake door testing works\n\nThe mechanics are simple:\n\n1. **Create the entry point.** Add the UI element a user would click if the feature were real — an \"Export to PDF\" button, a new nav item, a premium plan tile, or an in-app card promoting the capability.\n2. **Instrument the click.** Track who clicks, from where, and how often. Click-through rate is your primary demand signal.\n3. **Be honest at the door.** When the user clicks, show a transparent message: \"This feature is coming soon — want early access?\" Never leave users feeling tricked. Capturing an email here turns interest into a recruitment list.\n4. **Compare against a benchmark.** A click rate is only meaningful relative to expectations or a control. Decide your threshold *before* you run the test.\n5. **Decide: build, refine, or kill.** Strong demand justifies investment; weak demand saves you from the 64-80% unused-feature trap.\n\nFake doors work best for evaluating *demand* for a clearly describable feature. They are not a substitute for usability testing (which evaluates whether a built thing works) or for discovery interviews (which uncover problems you did not know existed).\n\n## What metrics to track\n\n- **Click-through rate (CTR):** the share of users exposed who engage the fake door. Your headline demand metric.\n- **Reach and segment:** which user segments clicked? A feature that excites power users but no one else changes your business case.\n- **Repeat intent:** users who return and click again signal durable, not novelty, demand.\n- **Opt-in rate:** of those who clicked, how many left an email for early access? This filters idle curiosity from real intent.\n\n## The ethics: do not actually trick people\n\nFake door testing has a reputation problem because it has been done badly. The rule is simple: **measure intent, never waste it.** Show a clear \"coming soon\" message, offer a real way to be notified, and limit exposure so you are not repeatedly disappointing the same users. A well-run painted door test leaves users feeling heard, not deceived — and gives you a warm list of people to talk to next.\n\n## The biggest weakness of fake doors — and how to fix it\n\nA fake door tells you *how many* people want something. It does not tell you *why*, *what they expected behind the door*, or *whether your solution would actually satisfy them*. A high click rate on \"Export to PDF\" might mean users want PDFs — or it might mean they want to share read-only reports and assumed PDF was the only option. Build the literal feature and you may still miss the real job.\n\nThat gap between a number and its meaning is where most fake door tests stop short. Closing it requires talking to the people who clicked.\n\n## The modern approach: pair the painted door with AI follow-up\n\nThis is where [Koji](https://www.koji.so) makes fake door testing dramatically more valuable. The opt-in list your painted door generates is a perfectly qualified audience — people who just raised their hand for this exact capability. Koji lets you interview all of them, automatically.\n\n- **Auto-interview every clicker.** Send each opt-in a link to an [AI-moderated interview](/docs/ai-moderated-interviews) in voice or text. Koji runs them 24/7, so you can talk to everyone who clicked instead of a hand-picked handful — turning a quantitative signal into qualitative depth within days.\n- **Quantify *and* explain in one conversation.** Koji's [structured questions](/docs/structured-questions-guide) — six types: open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — let you ask \"how likely would you be to use this weekly?\" on a scale, then have the AI immediately probe for the reasoning. You capture the demand number and the story behind it together.\n- **Surface what they actually expected.** Automatic [thematic analysis](/docs/thematic-analysis-guide) clusters responses so you learn whether \"Export to PDF\" really meant PDF — or sharing, archiving, or compliance. That is the difference between building the feature and building the *right* feature.\n- **Move at experiment speed.** Because Koji automates moderation, transcription, and synthesis, you get real-time insight while the test is still warm — no PhD in research methods required. Compared with manually scheduling interviews, teams using AI-assisted research see far faster time-to-insight.\n\nA fake door without follow-up answers one question: should we build it? A fake door plus Koji answers the harder one: what exactly should we build, and for whom?\n\n## Fake door vs. related validation methods\n\n- **[Smoke test](/docs/smoke-test-product-validation):** often used interchangeably; a smoke test usually validates a whole product concept (like a landing page), while a fake door validates a single feature inside an existing product.\n- **[Wizard of Oz testing](/docs/wizard-of-oz-testing-guide):** the feature appears to work, but humans perform the work behind the scenes. Use it to test the *experience*, not just demand.\n- **[Concierge MVP](/docs/concierge-mvp-guide):** you deliver the outcome manually for a few customers. Higher effort, deeper learning.\n- **[Pretotyping](/docs/pretotyping):** the broader discipline of testing \"would people use it\" before \"can we build it.\" Fake doors are a core pretotyping tactic.\n\n## Real-world patterns that work\n\nFake doors show up in several reliable formats:\n\n- **The feature button.** An \"Export to PDF\" or \"Schedule send\" button that, when clicked, reveals a coming-soon message and an early-access opt-in. Best for validating demand for a specific capability.\n- **The pricing tier.** A new plan or add-on shown on the pricing page. Clicks on \"Upgrade\" or \"Contact sales\" measure willingness to pay before you build the underlying functionality.\n- **The empty-state promo.** A card in an empty dashboard promoting a capability you are considering. Intent runs high here, because the prompt appears exactly when the user feels the gap.\n- **The demo or explainer.** Dropbox's screencast is the classic: present the product as if it exists and measure sign-ups. Best for validating a whole concept rather than a single feature.\n\n## A fake door test checklist\n\nBefore you launch, confirm that:\n\n- You defined a success threshold *in advance*, ideally against a control or a comparable existing feature.\n- The entry point looks genuine — a half-hearted fake door under-measures real demand.\n- The coming-soon message is honest and offers a real way to be notified.\n- You are capturing emails so interested users become a research and beta list.\n- Exposure is limited so the same users are not repeatedly disappointed.\n- You have a plan to follow up with clickers to learn *why* — the step that turns a number into a decision.\n\n## When not to use a fake door\n\nFake doors are wrong for some situations. Avoid them in safety-critical or trust-sensitive flows — payments, security settings — where a non-functional element erodes confidence. They are also weak for measuring *satisfaction* or *usability*, since a click proves interest, not that your eventual solution will work. And in small or high-touch B2B accounts, a single disappointed champion can cost more than the learning is worth; there, a direct conversation beats a painted door. Always match the method to the risk.\n\n## Frequently asked questions\n\n**Is fake door testing unethical?** Not when done transparently. Always show a clear \"coming soon\" message, offer a way to be notified, and avoid disappointing the same users repeatedly.\n\n**What is a good click-through rate?** There is no universal number — set your success threshold before the test, ideally against a control or a comparable existing feature.\n\n## Related Resources\n\n- [Smoke Test Product Validation](/docs/smoke-test-product-validation) — validate an entire concept before building\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide) — quantify and explain demand in a single conversation\n- [Wizard of Oz Testing](/docs/wizard-of-oz-testing-guide) — test the experience with humans behind the curtain\n- [Concierge MVP Guide](/docs/concierge-mvp-guide) — deliver the outcome manually to learn deeply\n- [Assumption Testing Guide](/docs/assumption-testing-guide) — identify and validate your riskiest assumptions\n- [MVP Validation Guide](/docs/mvp-validation-guide) — choose the right validation method for your stage","category":"Research Methods","lastModified":"2026-06-07T03:21:27.615133+00:00","metaTitle":"Fake Door Testing (Painted Door Test): Validate Demand Before You Build — Koji","metaDescription":"Fake door testing measures real demand for a feature before you build it. Learn how painted door tests work, what metrics to track, the ethics, and how to interview every clicker with AI.","keywords":["fake door testing","painted door test","feature validation","demand testing","product validation","validate feature demand","fake door MVP","pretotyping"],"aiSummary":"Fake door (painted door) testing shows users an entry point to a feature that does not exist yet and measures click-through to gauge real demand before building. It directly attacks the 35% of startups that fail from no market need and the 64-80% of features that go unused. Its weakness is that clicks reveal how many want a feature but not why. Koji closes that gap by auto-interviewing everyone who clicks using AI-moderated interviews, structured questions (six types), and automatic thematic analysis — turning a demand signal into actionable insight about what to build.","aiPrerequisites":["mvp-validation-guide","assumption-testing-guide"],"aiLearningOutcomes":["Run a fake door test to measure feature demand before building","Choose and benchmark the right demand metrics like CTR and opt-in rate","Apply fake door testing ethically without deceiving users","Convert clickers into qualitative insight with AI follow-up interviews"],"aiDifficulty":"beginner","aiEstimatedTime":"9 min read"}],"pagination":{"total":1,"returned":1,"offset":0}}