The Onboarding Mistake That Kills Innovative Products

I watched an innovative product’s onboarding fail in real time last year. The app looked like a small miracle. Layered rewards, interactive tutorial, a future-self narrative opener, visual progress everywhere. The entrepreneur building it had done his homework. He had applied gamification principles the way a good chef applies salt.

And I still came out of the tutorial confused.

Not confused about the buttons. Not confused about what to tap. Confused about something far more fundamental. I did not know what the product actually was. Was I playing a game? Was I ordering real food? Did the $4 I just spent come from my wallet or from a make-believe wallet? Would a burger show up at my door, or had I just decorated a virtual burger that would live forever on a blockchain nobody would ever check?

The tutorial had taught me how to use every feature and failed to tell me what the product actually did.

This is the trap of innovative-product onboarding. When you are building something new in category terms, great UX can quietly destroy your conversion rate, because the playbook written for familiar products is missing the one thing your users need most.

⚡ Speed Run Notes

• Traditional onboarding teaches HOW to use features. Innovative products first need to teach WHAT the product actually is — the category itself.
• The metric that matters for new-category products is comprehension rate, not completion rate. How many users, after the tutorial, can explain the product to a friend?
• Uber and Airbnb both launched into “wait, what?” reactions. They did not design for 100 percent instant comprehension. They designed to move the baseline from 20 out of 100 to 35 out of 100 to 45 out of 100.
• Four moves establish context before features: a future-self narrative opener, a 3-panel product-arc sequence, precise language that separates virtual from real, and simplified starter choices that get users to first value fast.
• The answer to “fun onboarding versus clear onboarding” is not to pick one. It is to layer them. Boring clarity first, for 30 seconds. Then the gamified tutorial earns its fun.

About Yu-kai Chou

Yu-kai Chou is the creator of the Octalysis Framework, author of Actionable Gamification and 10,000 Hours of Play, and one of the world’s most cited voices in gamification and behavioral design.

For more than fifteen years he has advised governments, founders, and major brands on onboarding, motivation, and behavior-change design, often helping teams diagnose why polished products still lose users in the first five minutes.

This article comes directly out of that work. The context-clarity failure below is one of the most common reasons innovative products quietly bleed conversion even when the interface looks impressive on the surface.

The Day I Realized “Great Onboarding” Was Wrecking Products

Let me give you the concrete case, because abstractions do not move founders. An entrepreneur I was advising was building a product at the intersection of three then-novel categories: food delivery, NFT ownership, and a virtual-restaurant simulation game. You could design a burger. You could “own” that burger as a digital asset. And other players could actually order that burger and have it delivered to their physical home.

Three categories in one product. Each one already weird on its own. Stacked together, borderline uninterpretable on first contact.

The onboarding flow was excellent by conventional measures. It opened with a future-self narrative: “In the future, you own a chain of NFT restaurants. You are already successful. You just have to take the steps to get there.” That opener activates Core Drive 1 (Epic Meaning & Calling) and Core Drive 2 (Development & Accomplishment) at the same time, and it is a technique I teach in my Octalysis Framework because it collapses the first-step-is-scary problem by telling the user the destination is assured.

From there the tutorial did everything you are supposed to do. Progressive disclosure. Experience points for every action. Gold coins for acquired ingredients. Level-ups with satisfying visual feedback. Rare items that tickled Core Drive 7 (CD7): Unpredictability & Curiosity. Each step was bounded, each reward was immediate, each click felt like earned progress.

And at the end of it I could not answer the question a six-year-old would have asked first: “Is this real?”

I pulled up the entrepreneur’s screen and said something close to this: “If this is the first day I downloaded your app, and I just went through this, I walk away thinking this is a virtual burger game. A little Farmville with meat. I don’t think a real person is going to drive to my actual house and bring me food at the end of it.” His face did that thing founders’ faces do when they see their blind spot. A half-second of pain, then a full second of “oh no, how many users already churned because of this.” (For a real-named product where I ran the same diagnosis end to end, see my Apollo Neuro onboarding audit — the failure mode is identical even when the category is wellness rather than restaurants.)

The onboarding had flawlessly taught the how. It had never taught the what.

The Missing Metric: Comprehension Rate

Most product teams track completion rate. What percentage of users who started the tutorial finished it. A 70 percent completion rate is considered healthy; 85 percent gets a celebratory Slack emoji. Completion rate is cheap to measure, easy to A/B test, and almost always moves in the direction your dashboard wants it to.

For innovative products, completion rate is the wrong metric.

The metric you actually need is comprehension rate. How many users, five minutes after finishing the tutorial, can explain your product to a friend in one sentence that is not factually wrong?

A user who completes your tutorial but walks away believing your NFT food ordering app is a game is a user who did not, functionally, finish your onboarding. They finished the clicks. They did not finish the cognitive installation. And the moment they get a push notification saying “Your burger is being delivered!” they are going to feel confused rather than excited, because a delivery is not something they expect from a game.

Comprehension rate is harder to measure. You cannot pull it from an analytics dashboard. The cheapest approach I have used with clients is a ten-person guerrilla test: recruit ten strangers, have them use your product cold for five minutes, then ask them one question: “in one sentence, what does this product do?” Score each answer as correct, partially correct, or wrong. That is your baseline.

Ten people is not statistically rigorous and I don’t care. Ten people will tell you whether you are at 20, 35, or 50 percent comprehension, and that is the only resolution you need to make decisions.

The Uber and Airbnb Pattern (Why Weird Becomes Normal)

Here is the part I want you to sit with, because it is the part that most founders get wrong when they hear “your category is confusing.”

Uber was a weird product. Airbnb was a weird product. “I’ll get into a car with someone I summoned from an app on the internet” was not a sentence anyone was saying in 2009. “A stranger will sleep in the empty bedroom of the house I live in” was even weirder. Both products had comprehension-rate problems on day one. Both products scaled to tens of billions of dollars in market cap anyway.

They did not get there by designing onboarding flows that achieved 100 percent comprehension. That target is mathematically impossible when your product creates a new category, because a significant slice of your users do not yet have a mental model to hook the explanation onto. You cannot Bullet Point your way into a slot that doesn’t exist in the user’s head.

What Uber and Airbnb did was compound two things over time. First, they improved the design so that the comprehension rate moved. Maybe 20 out of 100 early users understood what the product was on first contact. A year later it was 35 out of 100. A year after that, 45. The design got clearer, the copy got sharper, the first-time experience shed friction. None of those individual improvements were dramatic. All of them were measurable.

Second, and this is the part founders underestimate: the users who did get it told their friends. This is Core Drive 5 (Social Influence & Relatedness) in its raw form. Once a category becomes a conversation topic between friends, the comprehension rate on first contact is no longer the only lever. Your users are now onboarding your next users for you. If you want to see how that compounds, my post on Core Drive 5: Social Influence and Relatedness digs into why peer validation moves faster than any ad campaign ever will.

So the design question is not “how do I get 100 percent of users to understand my weird product.” The design question is “how do I move the baseline from 20 to 35 to 45, one cohort at a time, while Core Drive 5 does the rest of the work for me.”

Four Techniques That Establish Context Before Features

Once you accept that you are designing for a comprehension-rate slope rather than a single fixed number, the design moves become concrete. Here are the four that have moved the needle most in the client work I have done.

1. Open with a future-self narrative (activates CD1 and CD2)

The entrepreneur I was advising had already nailed this one. The opening line was not “Welcome. Let me teach you the app.” It was “In the future, you own a chain of NFT restaurants. You are already successful.”

That sentence does three jobs at once. It collapses the first-step intimidation because the future is already promised. It plants CD1 (Epic Meaning & Calling) by telling the user they are on a narrative arc toward something significant. And it pre-loads CD2 (Development & Accomplishment) because every tutorial step becomes a down-payment on the destination, not an obstacle between the user and the destination.

The failure mode is using a future-self narrative for the emotional lift and then forgetting to tell the user what they are a future-self OF. Your user needs to know what kingdom they are the king of before “you are the king” means anything.

2. Insert a 3-panel product-arc sequence before the tutorial

This is the single highest-leverage fix I made on the NFT food app. Before the interactive tutorial started, we added three static panels. Each panel had one sentence and one illustration.

Panel one: “Design food items you wish existed.” Panel two: “Other players order and pay for them — with real money.” Panel three: “We deliver the physical food. You earn a share of every sale.”

Thirty seconds. Three illustrations. Zero cleverness. The entrepreneur’s first reaction was to object. In his words: “this feels like a boring PowerPoint in the middle of my beautiful game.” I told him that his beautiful game was what was eating his comprehension rate, because users were exiting the tutorial with the mental model of a game rather than the mental model of a restaurant business. The 3-panel sequence is not there to entertain. It is there to install the product’s cause-and-effect chain in the user’s head so that every feature they later learn has a slot to fit into.

The test for whether your 3-panel sequence works: a user who closes the app after just those three panels, never touching the tutorial, should still be able to explain your product correctly. If they can’t, the panels are not doing their job yet.

3. Use precise language that separates virtual, real, and hypothetical

This one sounds like a copy-editing detail. It is not. It is the difference between a user who trusts your product and a user who is waiting for the other shoe to drop.

On the NFT food app, the word “store” was doing double duty. It meant the virtual simulation and it meant the real fulfillment pipeline. We split it. “Your virtual store” and “real store deliveries.” The word “burger” got the same treatment. “Virtual burger design” and “real burger orders.” Money became “real dollars” or “in-game coins” depending on which one was involved.

A companion character in the tutorial (we used a cartoon fox) also got better lines. Instead of “Congrats! You designed a burger!” we wrote “Soon you’ll be able to share this burger with friends and family — and they can start ordering it.” That one sentence turned the tutorial from a game moment into a preview of the category.

When every noun and every verb is explicit about which layer of reality it lives in, your user stops doing background work to keep the layers straight. That freed attention becomes the budget they spend on learning your product.

4. Simplify early choices to get to first value fast

The NFT food app’s first tutorial screen asked users to design a burger from scratch. Pick a bun. Then a patty, then the cheese, then sauces, toppings, and a garnish. It was a beautiful choice-architecture puzzle. For a first-time user who did not yet understand the category, it was paralysis.

The fix was a “Surprise Me” button. One click, a randomized burger gets generated, the user moves to the next step. The user who wants to customize can still customize; they just can’t be forced into customization before they know why customization matters.

This is a direct application of the Core Drive 3 (CD3) — Empowerment of Creativity and Feedback — trap I see constantly: giving the user meaningful choice before the user has any basis to make a meaningful choice. Choice without context is friction, not empowerment — Schwartz called this the paradox of choice: too many options before context produces paralysis, not empowerment. Once the user has been through one full loop (design, sell, deliver, earn), the second time around the customization options mean something, because they have the mental model to know which customization would actually matter.

If you want to go deeper on how onboarding specifically hands off to the next phase of the player journey, my post on the scaffolding phase walks through how to prevent the 80 percent drop-off that happens between tutorial completion and habit formation.

The Fun-Versus-Clear Tension (And the Layering Fix)

Founders who have sat with this long enough eventually stumble into what feels like an unsolvable tradeoff. The two options look mutually exclusive.

Option A is fun, interesting, gamified onboarding. The kind that gets you quoted in design case studies. It activates multiple Core Drives, produces beautiful screen recordings for your pitch deck, and makes the user feel something.

Option B is clear, somewhat boring, explanatory onboarding. The kind that feels like a corporate training module. It installs the product’s category, leaves nothing ambiguous, and makes the user understand something.

Most founders pick Option A, because fun is easier to fundraise against than boring. Almost every innovative-product onboarding I have reviewed picked Option A. And almost every one of those was leaking users who had “completed” the tutorial without understanding the product.

The answer is not to pick. The answer is to layer.

Open with 30 seconds of boring clarity — the 3-panel product-arc sequence, narrated by a companion character, using precise language. No rewards. No points. No choice architecture. Just category installation. Then, and only then, hand off to the fun gamified tutorial. The fun tutorial now does exactly what it was designed to do: build skill, give feedback, produce satisfaction. That only works because it is no longer also trying to teach the user what the product is while teaching the user how to use it.

The trap is that Option A feels too valuable to “waste” the first 30 seconds on something boring. It isn’t a waste. It is the difference between a user who converts and a user who evaporates.

There is a related decision, whether your product should even look like a game at all, that I unpack at length in my post on implicit versus explicit gamification. If you are already asking whether the gamified layer is the right shell for your product, that one is the next piece to read.

The 5-Question Pre-Ship Checklist

Before you ship the next version of your onboarding, answer these five questions out loud. If any answer is “I don’t know” or “probably yes,” you are not ready.

1. Can a user who just finished my tutorial explain my product to a friend in one sentence that is not factually wrong? If you have not actually tested this with ten real strangers, you do not know the answer. Intuition is worthless here. The founder always thinks the product is obvious because the founder has been marinating in it for two years.

2. Is my product a new category or an incremental improvement on a familiar one? If it is incremental (a better to-do list, a faster messaging app, a cleaner expense tracker), you can skip most of this advice. Familiar categories have pre-installed mental models. You are not doing category installation, you are doing feature positioning. Different problem.

3. Have I established the full product arc before I start teaching features? Every feature has a place in the cause-and-effect chain of your product. If the user does not know the chain, every feature feels like an orphan. Three panels. Thirty seconds. Before the tutorial.

4. Am I using precise language to distinguish virtual from real, hypothetical from actual, and earned from simulated? Open your onboarding copy. Flag every noun that could mean either of two layers of reality. Rewrite each one to be unambiguous. This is unglamorous work that moves comprehension rate more than a new illustration system ever will.

5. What is my current comprehension rate, and what is my target for the next cohort? Ten-person guerrilla test. Score the answers. Write the number on a sticky note. Ship changes. Re-test. Move from 20 to 35 to 45. That is the job. My deeper treatment of the first five minutes of onboarding walks through the rest of the measurement system if you want the full toolkit.

Frequently Asked Questions

What is comprehension rate in onboarding?

Comprehension rate is the percentage of users who, after finishing your onboarding, can correctly explain in one sentence what your product actually does. It is distinct from completion rate, which measures only whether the user clicked through every step of the tutorial. For innovative products in new categories, comprehension rate is the more important number because a user who completes the tutorial but cannot explain the product has not been successfully onboarded.

Why do innovative products need different onboarding than familiar products?

Familiar products can assume the user already has a mental model of the category. A user opening a to-do list app already knows what a to-do list is. Onboarding only needs to teach the specific features. Innovative products, those creating a new category or combining multiple novel concepts, cannot assume a mental model exists. They must install the category first, before teaching any features, because otherwise every feature feels unmoored.

How do I test my product’s comprehension rate?

The cheapest viable test is a ten-person guerrilla study. Recruit ten strangers who match your target user profile. Have each one use your product cold for five minutes. Then ask one question: “In one sentence, what does this product do?” Score each answer as correct, partially correct, or wrong. The resulting ratio is your baseline. Statistical rigor is not the point. Decision-grade resolution is.

What is a future-self narrative in onboarding?

A future-self narrative is an opening framing that tells the user who they will become after using the product, rather than what the product does. The Octalysis Framework maps this to Core Drive 1 (Epic Meaning and Calling) combined with Core Drive 2 (Development and Accomplishment). An example: “In the future, you own a chain of successful restaurants. You are already there. You just have to take the steps.” This frame reduces the intimidation of step one by making the destination feel inevitable.

Should I sacrifice fun to add clarity to my onboarding?

No. The two are not actually in conflict once you layer them. Open with 30 seconds of boring-but-clear category installation — a 3-panel product-arc sequence with precise language. Then hand off to the fun gamified tutorial. The fun tutorial works better once it no longer has to also teach the user what the product is. Fun is a multiplier on comprehension, not a substitute for it.

What to do next

If you are building anything new in category terms (an NFT product, a new consumer category, a behavioral-design intervention that combines disciplines) your onboarding is almost certainly leaking users who “completed” the tutorial without installing the category. Start with the 10-person comprehension test this week. Use the result as the baseline that every future onboarding change is measured against.

If you want the broader framework that context clarity fits into, the Octalysis Framework is the map. For the canonical breakdown of where onboarding fits in the 4 Experience Phases of a player journey, that’s the upstream piece. If you want the deep dive on motivation itself, Actionable Gamification is the book. And if you want the life-scale version of the same principles applied to your own story, 10,000 Hours of Play is where that thread lives.

Context first. Features second. Fun always. But fun earns its place only after the category is installed in your user’s head.

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?