Why Do New Users Quit? The Onboarding Belonging Equation

I’ve sat in roughly forty different churches as a first-time visitor across six countries. Most of them lost me before the second hymn. One of them turned a stranger into a regular within ninety seconds, and the difference had nothing to do with the sermon.

The same onboarding gap shows up with apps. Notion gave me a personalized starter workspace and I stayed for years. A productivity app I downloaded last week forced me to name my workspace before showing me anything, and I uninstalled it. Wildly different first ninety seconds.

Onboarding is where products either create belonging or bleed out new users. Almost everyone designs it last, when it should be designed first.

⚡ Speed Run Notes

• Onboarding is the single most important phase of any user journey, because users have no sunk cost yet and small friction feels like a wall.
• Belonging compresses to one equation: Acceptance (Core Drive 5) + Epic Meaning (Core Drive 1). Either one without the other quietly kills retention.
• Make appreciation cheap. Low-effort welcome signals (one-click endorsements, auto-greetings, “say hi” prompts) outperform high-effort ones every time.
• Epic Meaning during onboarding isn’t a mission statement — it’s the answer to “why does my showing up here matter to anyone other than me.”
• The five fatal onboarding failures: empty states, no welcome from existing members, required tasks before engagement, unclear value, and information overload.
• Invest disproportionately in the first hour. A user’s first sixty minutes predicts their lifetime value better than almost any other metric you can measure.

Table of Contents

Author Credibility: Yu-kai Chou

Yu-kai Chou is an S-Tier Behavioral Designer and the creator of the Octalysis Framework, the gamification design system now applied to products and experiences reaching over 1.5 billion users. His book Actionable Gamification is one of the most-cited works in the field, and he has been ranked the #1 Gamification Guru in the World.

He has advised MrBeast, LEGO, Microsoft, Porsche, Tesla, Stanford, Harvard, and governments including Ukraine on turning behavioral psychology into product mechanics that actually change user behavior.

Verify: Wikipedia · Google Scholar · Wikidata · LinkedIn

I’ve spent the better part of two decades studying why people stay in some experiences and quit others. Onboarding is the chapter where my own consulting clients lose more users than the rest of the product combined. The Four Phases of Player Experience (Discovery, Onboarding, Scaffolding, Endgame) sit at the heart of how I teach behavioral design, and the Onboarding phase is where I’ve seen seven-figure user-acquisition spends evaporate inside the first ninety seconds. Everything below is what I wish every product manager, community founder, and church pastor knew the day they shipped their first welcome flow.

The Four Phases Most Product Teams Forget

Every long user relationship moves through four distinct phases, and each phase has its own job to do.

Discovery is how someone hears about you. Onboarding is the first time they actually touch the experience. Scaffolding is the long middle, where habits compound and skills deepen. Endgame is what keeps mastered users from leaving once there’s nothing new to learn.

Most product teams over-invest in Discovery (because marketing budget is loud) and Endgame (because power users are the loudest in the data). Onboarding is the orphan child. It gets a checklist of “set up your profile, follow three people, take the tour” designed by an engineer on a Friday afternoon, never tested with someone who actually doesn’t already work at the company. The behavioral science backing this gap is real: Uncertainty Reduction Theory shows that newcomers in any environment burn most of their cognitive energy in the first interactions just trying to figure out whether they’re safe and whether they belong.

Here’s the asymmetry: Discovery and Scaffolding can be optimized incrementally. You A/B test, you ship, you iterate. Onboarding is binary in a way the other phases are not. The user either makes it through with a sense of “I belong here,” or they don’t. There’s no second first impression. The window closes inside an hour, often inside a minute.

This is why the rest of this post is going to focus almost entirely on the Onboarding phase. Get it right and Scaffolding compounds for years. Get it wrong and your Endgame design is moot, because no one ever made it that far.

The Belonging Equation: CD5 + CD1

If I had to compress every onboarding lesson I’ve ever taught into one formula, it would be this:

Belonging = Acceptance + Shared Purpose
(Core Drive 5: Social Influence) + (Core Drive 1: Epic Meaning)

Both halves are required. Either one alone is a known failure mode I’ve watched dozens of products and communities walk into.

Core Drive 5 without Core Drive 1 is the Slack community where everyone’s nice but nobody knows what they’re building together. People feel welcomed, then quietly drift away because there’s no reason to come back. The friendship doesn’t have a center. Within ninety days the channel turns into a graveyard of “hey, anyone still here?” messages.

Core Drive 1 without Core Drive 5 is the open-source project with a clear, ambitious mission and a community that intimidates new contributors into silence. Pull request guidelines run seventeen bullet points long. The first comment a newcomer gets is a list of corrections. By month two, the mission is still real but no one stayed long enough to serve it.

Combine both halves and you get the magic: “I’m part of something meaningful, and the people there are my people.” That’s belonging. That’s what onboarding has to manufacture inside the first hour. The decades of work behind Social Identity Theory show why CD5 is so loaded in this window: humans don’t just want to be welcomed individually, they want their group affiliation to feel safe. For the deeper anatomy of how all eight Core Drives interact, the Octalysis Framework page lays out the full system.

Most products try to fake one half. They lean hard on Acceptance (the friendly chatbot, the auto-greeting) and skip Epic Meaning. Or they lean on the mission video at signup and never give the user a single moment of feeling seen. Neither works alone. You need both, in the first session, before the user has decided whether you’re worth a second one.

Make Appreciation Cheap (the LinkedIn Lesson)

One of the cleanest natural experiments in social-acceptance design happened on LinkedIn in September 2012, when the platform launched one-click endorsements alongside its existing written-recommendation system.

The recommendation system was, on paper, much better. It produced richer signals. A real human had taken twenty minutes to write specifically about what you were good at. Hiring managers loved them. Job seekers begged for them. And almost no one wrote them, because they cost twenty minutes.

Endorsements cost one click. They were lower-quality individually and dramatically higher-volume in aggregate. Within a few years, profiles that had collected zero or one recommendation in their entire LinkedIn lifespan were sitting on dozens of endorsements. The social-acceptance signal got cheaper to produce, and the network suddenly felt warmer to operate in. People who’d never written a recommendation in their life were now publicly affirming their colleagues’ skills every week.

The lesson generalizes far past LinkedIn. The amount of acceptance a community can produce is bounded by how cheap the cheapest acceptance signal is. If your only welcome mechanism is “post a thoughtful introduction in #introductions,” you’re going to get a thoughtful introduction from one in fifty new members. If your welcome mechanism is a one-tap “wave at this newcomer” button that fires off a notification, you’ll get fifteen waves before the newcomer finishes reading the rules.

Designers who don’t think about cost usually optimize for signal quality and end up with a community that produces almost no signal at all. Designers who think about cost realize that ten cheap welcomes beat one expensive one every time, because the newcomer doesn’t know any of the welcomers personally yet. What they need is the felt sense that this is a place where new arrivals get noticed.

Practical implication: in your onboarding flow, audit every social signal by its cost in seconds. If a welcome takes longer than three seconds to send, it will be sent rarely. Engineer the three-second welcome before the twenty-minute one.

The Church Newcomer Strategy

I want to tell you about three churches I visited as a first-timer, because the contrast is one of the cleanest natural experiments in onboarding design I’ve ever seen, and the patterns map directly onto product onboarding.

Church A opened the service with: “If you’re new today, please stand up so we can welcome you.” Two visitors, including me, sheepishly half-stood. The pastor smiled, said “welcome,” and moved on. Net effect on me: I felt singled out, slightly embarrassed, and zero sense of belonging. I didn’t go back.

Church B tried something different. The pastor said: “Newcomers, on the count of three, I want you to stand up and shout your name. One. Two. Three!” People laughed. Half a dozen newcomers stood. The energy in the room shifted from awkward to playful. The cost of standing up went from “I’m being publicly singled out” to “I’m part of a tiny group having fun together.” I went back the next week.

Church C took the approach to its extreme. As a newcomer walked in, three trained greeters lined up to shake their hand, ask their name, ask what brought them in, and physically walk them to a seat next to a regular member who’d been briefed to spend the next ninety minutes including them. Before the service even started, the newcomer had been welcomed by name by four people and had a buddy for the entire experience. The retention rate of first-time visitors at Church C was reportedly several multiples of the regional average.

The transferable insight isn’t “do what Church C does.” It’s that onboarding belonging is engineered, not improvised. Church A had no system; the pastor’s instinct produced a system that maximized exposure (high cost) for minimum reward (a generic “welcome”). Church B redesigned the cost structure of the moment. Church C built infrastructure (trained greeters, a briefed buddy, a physical walking-and-talking sequence) to produce belonging on demand.

Map this onto a SaaS onboarding. Church A is the product that displays “Welcome, new user!” in a banner and considers the job done. The Church B equivalent turns the awkward signup moment into a small playful interaction (Duolingo’s owl mascot is doing this when it cheers your first lesson). At the Church C end of the spectrum, the product assigns a real person, a customer success rep or community manager or existing power user, to walk the newcomer through their first session as a one-on-one. The retention math will track the church math.

Epic Meaning Without Lectures

The Acceptance half of the equation is well understood. The Epic Meaning half is where most product teams fumble, because their default mode of installing meaning is a TED-talk-shaped intro video the user is required to watch.

Lectures don’t install meaning. Stakes do.

The Frodo example I keep coming back to: in The Lord of the Rings, the mission is clear within the first chapter (destroy the Ring). But the meaning isn’t installed by a narrator explaining why this matters. It’s installed by Frodo’s choice. A small, ordinary hobbit volunteers for a thing he is plainly under-qualified for, surrounded by people who are clearly afraid of what he’s signing up for. The reader installs meaning by watching the cost being paid, not by being told the cost is high.

Translated to product onboarding: show the user what their participation will mean to someone other than themselves, ideally someone real, ideally inside the first ten minutes.

Khan Academy’s onboarding for teachers shows the parent or coach dashboard tied to a learner’s progress, so a tutor knows the lesson they assign actually shows up in another human’s life. Strava’s onboarding surfaces the kudos history of athletes near you, so your first activity enters a context where dozens of strangers are watching for new neighbors to cheer. Wikipedia surfaces, on every article you’re considering editing, how many readers it reaches each day — your one fix matters to a stranger you’ll never meet.

Each of these is doing the same thing: planting a stake other than the user’s own self-interest, inside the onboarding window, before the user has decided whether to invest the next hour. That’s CD1 in onboarding form. It’s not a mission statement. It’s a moment where the user feels their action would matter to someone they haven’t met yet.

If you can’t articulate what your user’s first action does for someone outside themselves, your onboarding has no Epic Meaning, no matter how many “join the movement” headlines you put in the welcome screen.

The Octalysis Prime Slack Story

One of the cleanest demonstrations of CD5 + CD1 working together that I’ve ever watched up close was inside the Octalysis Prime Slack community.

For a long stretch, the community had a member I’ll just call Ulrich, who lived in Hawaii and had a habit so unusual I want to describe it carefully. Every time a new member joined the Slack, Ulrich would write them a short, personalized welcome poem. Not a templated greeting. A real poem, four to eight lines, riffing on something specific about their profile or their introduction post.

Imagine joining a Slack and the second message you get, before you’ve said anything substantive, is a stranger having taken five minutes to write you a poem because you exist. The effect on new members was immediate and visible. They responded with disproportionate energy. They posted more in their first week than the average member did in their first month. Many of them stayed for years.

The mechanic was simple but the design was deep. CD5 was hitting hard. The felt sense of being personally seen by an existing member is one of the most potent acceptance signals in any community. CD1 was riding underneath, because the act of being welcomed in this specific weird and beautiful way told the newcomer something about what kind of community this was. A community where people write strangers poems is a community with a culture, and a culture is a form of shared meaning.

The replicable design pattern isn’t “have your community write poems.” It’s that one consistent, slightly weird, slightly costly welcome ritual outperforms a hundred templated automated greetings, because the weirdness is the proof of culture. Ulrich’s poems worked because they were obviously Ulrich-specific, obviously not scalable, and obviously a choice. That’s what made them feel like belonging instead of like marketing.

The Five Fatal Onboarding Failures

Across the hundreds of onboarding flows I’ve reviewed for clients, the same five failure modes show up again and again. Each one independently kills retention. Most failed onboarding flows ship two or three of them simultaneously.

Failure 1: The empty state with no guidance. The user lands inside the product and sees a blank dashboard, an empty inbox, a zero-state list. There’s nothing to react to. Nothing to do. No proof other people exist here. The user feels lost, decides this product needs work, and closes the tab. The fix is the seeded empty state — populate the dashboard with sample data, example projects, or a guided first action that produces a non-zero state in under thirty seconds.

Failure 2: No welcome from existing members. The user sees zero evidence that any other human has noticed they joined. No notification, no greeting, no “say hi” prompt to anyone. The community feels empty even when it isn’t. The fix is automated visibility — a “new this week” feed for members, a welcome bot that pings the newcomer’s first post, a DM from a community manager triggered on day one.

Failure 3: Required tasks before engagement. The signup flow demands a complete profile, a profile picture, three follows, and a six-step product tour before the user is allowed to do the actual thing they came to do. Each required step is a quit point. The fix is progressive disclosure — let the user do the actual thing first, then surface profile completion as a low-friction nudge after the first successful action.

Failure 4: Unclear value proposition. The user finishes the welcome flow and still couldn’t explain to a friend what your product does for them. Marketing promised one thing, the product delivered something else, and the connection between today’s first action and tomorrow’s payoff was never made explicit. To fix this, measure time-to-first-value: how many seconds elapse between “user lands on the dashboard” and “user completes one action that produces visible personal value.” If that number is over five minutes, your onboarding has a value-proposition problem.

Failure 5: Information overload. The user is shown nineteen tooltips, four modal walkthroughs, an explainer video, and a feature tour, all in the first session. Each one is well-intentioned. Together they amount to “we don’t trust you to figure anything out, and we’d rather lecture you than let you experience the product.” The fix is the rule of one — pick ONE thing the user needs to learn in session one, teach that thing well, and let everything else surface contextually as the user encounters it later.

Modern Examples (Discord, Notion, Duolingo)

Three current products show what good onboarding looks like in 2026.

Discord nailed CD5-onboarding for community spaces. When you join a new server, the system surfaces a “rules screening” gate that doubles as a welcome ritual. You can’t post until you’ve reacted to a specific emoji on the rules message. It sounds bureaucratic on paper. In practice it’s a self-administered acceptance ritual — you’re saying “I’m in” with one click, and the existing community sees a new member’s reaction join the list. Cost to the newcomer is one tap. Signal to the community is a real new arrival. Designed correctly, it produces the same effect as Church B’s “stand up on the count of three.”

Notion nailed time-to-first-value. The signup flow drops you into a starter workspace prepopulated with a personal task list, a meeting notes template, and a sample database — none of which feel sample-y, because they’re plausible enough to actually use. Your first action is editing a real-feeling document, not staring at an empty canvas. The value proposition installs itself in under sixty seconds because you’re already using the product before you’ve decided to commit to it.

Duolingo nailed CD1-onboarding for solo experiences. The owl mascot, the streak counter, the “you’re learning Spanish for these reasons” framing — these install Epic Meaning into what would otherwise be a flashcard app. The mascot Duo is a manufactured stake. By the end of the first lesson the user has been told, in essence, “we are counting on you tomorrow.” For a product that has no community in the traditional sense, this is how you fake CD5 and CD1 simultaneously: invent a character whose disappointment matters, and let the user’s continued effort be a kindness to that character.

None of these companies stumbled into their onboarding. Each of them has spent years iterating on the first-session experience and produced patterns the rest of us can borrow.

The First-Hour Test for Your Product

Here’s the diagnostic I run with consulting clients to find out whether their onboarding is actually creating belonging or just shipping users.

Sit a real new user down in front of your product for sixty minutes. Watch them. Don’t talk. Don’t help. At minute sixty, ask them three questions.

One: did anyone in this product notice you exist? If the honest answer is “no,” your CD5 design is broken. Doesn’t matter how good the rest is. The user is alone, and alone users churn.

Two: do you understand why your participation matters here? If the honest answer is “I’m not sure” or “for myself, I guess,” your CD1 design is broken. The user has no shared mission, and missionless users drift.

Three: would you be embarrassed to lose this product tomorrow? If the honest answer is “no,” nothing in the first hour created enough belonging to justify the next hour. Whatever you ship next, ship it before they close the tab.

This test is brutal because it’s binary. There’s no half-credit. Belonging either got installed in the first session or it didn’t, and the lifetime value math is going to follow the answer either way.

If you want to see how Acceptance and Epic Meaning combine with the other six Core Drives across the rest of the user journey, the breakdown lives in my book Actionable Gamification, and the framework itself is documented at the Octalysis Framework page. How to design the Scaffolding and Endgame phases that follow a strong onboarding is what most of my consulting work covers.

But none of that matters if the Onboarding phase doesn’t do its job first. Get the first hour right. The rest is downstream. (For practitioners moving into deeper architectural work, Octalysis Level 2 covers how to wireframe the rest of the experience once Onboarding is solid.)

Frequently Asked Questions

What is the most important phase of user onboarding?

The Onboarding phase itself is the most important phase of any user experience, because users have no sunk cost yet and small friction points feel like major obstacles. Within the Four Phases framework (Discovery, Onboarding, Scaffolding, Endgame), Onboarding is binary in a way the others are not — the user either makes it through with a sense of belonging or they quit, and the window typically closes inside the first hour.

What is the Onboarding Belonging Equation?

Belonging compresses to one formula: Acceptance (Core Drive 5: Social Influence) plus Epic Meaning (Core Drive 1) equals belonging. Either half alone is a known retention failure mode. CD5 without CD1 produces friendly communities with no shared mission. CD1 without CD5 produces meaningful missions with intimidating cultures. Both are required inside the first session for a user to feel they belong.

Why do new users quit during onboarding?

The five most common onboarding failures are: empty states with no guidance, no welcome from existing members, required tasks before engagement, unclear value proposition, and information overload. Most failed onboarding flows ship two or three of these failure modes simultaneously, and each one independently kills retention even when the rest of the product is strong.

How long is the onboarding window?

For most digital products, the critical onboarding window closes inside the first hour, often inside the first ninety seconds. A user’s first sixty minutes with a product or community predicts their lifetime value better than almost any other early metric. Discovery and Scaffolding can be optimized incrementally, but Onboarding is binary and there is no second first impression.

What is Core Drive 5 in the Octalysis Framework?

Core Drive 5 is Social Influence and Relatedness — the motivational drive activated by social acceptance, peer validation, and the feeling of being seen by a community. In onboarding design, CD5 manifests as low-cost welcome signals (one-click endorsements, automated greetings, “say hi” prompts) that produce high-volume acceptance signals. The full Octalysis Framework documents all eight Core Drives.

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