Chapter 3. What Does “First” Mean?

Your FTUE began before the user ever opened the app.

People arrive with expectations already formed by ads, thumbnails, and what their friends have said. A friend’s offhand “This game is fun,” or a single screenshot glimpsed in a store has already drawn a picture of our game in their mind. A team that watches only the first launch screen therefore misses the entire moment when the user made up their mind before reaching it.

“First impressions matter.” Everyone knows that. But ask exactly when a game’s “first impression” is formed, and the answers diverge.

The first screen? Plausible. Yet users begin judging long before they see it: when a friend says, “This game is fun,” when they swipe through three store screenshots, or when they press Download and wait for loading to finish. At every one of those points, expectations rise or fall a little. A “first impression” is not a single point but a succession of judgments that are revised again and again.

The word we need to stop and examine is first: the F, or First-Time, in FTUE. If there is a first time, there must be a second, so how are those “times” divided? The single lump we casually call “the first” actually contains many beginnings of different kinds, and we cannot design it if we treat it as one point. We are left holding only the result—“D1 is low”—without being able to identify which “first” is leaking people. This chapter divides that one point into eleven thresholds. Chapter 4 takes up the question of what kind of “experience” users have on those thresholds.

A Funnel That Cannot Tell You Where People Dropped Out

Consider a scene. These are hypothetical numbers, not actual data. Suppose a mobile game receives ten thousand new visitors in a day. Ten thousand people see an ad and enter its store page. Four thousand of them install it. Three thousand install and actually open the app. Fifteen hundred finish the tutorial. Three hundred open it again the next day.

Ten thousand came in; three hundred remained. It is easy for the team to look at this funnel and lament that “conversion is low.” The hard part is identifying which segment leaked the most. The drop from ten thousand to four thousand—people who saw the store page but did not install—and the drop from four thousand to three thousand—people who installed but did not open—are entirely different problems. The former concerns store screenshots and first impressions; the latter concerns download size, permission requests, or the initial loading experience. If both are lumped together as a “first-impression problem,” the leak will remain even after the screenshots have been revised a hundred times.

This is what happens when “first” is treated as a single point. Unless we divide the funnel into fine-grained stages, we patch the wrong place without knowing where the leak is. We therefore need to divide “first” into units we can actually grasp.

Eleven Thresholds, Three Phases

This book divides “first” into eleven thresholds, from T0 through T10. Eleven are hard to memorize at once, so we group them into three phases: before the user encounters the game, within the first session, and during settling in. Let us follow in prose how the user’s mind moves through these three phases.

The first phase takes place before the game is opened. Long before launching it, the user first encounters us by hearing its name through a friend’s recommendation, a community post, or an ad (T0, first word of mouth). The first thing planted here is expectation. Next, the user sees a store thumbnail or trailer (T1, first exposure), decides at a glance “what kind of game this appears to be,” and predicts its genre, price, and fun (T2, first expectation). “Oh, it’s an idle game.” “It’s probably free.” “Cute collectibles.” This prediction becomes a promise—the baseline against which the user will later compare the actual game. They have not played for even one second, but an image of our game already exists in their mind.

The second phase occurs within the first session, and it is the most densely packed. While the user downloads and launches the game, logs in, grants permissions, and waits for it to load (T3, first entry), a high startup cost makes them leak away. This first entry also carries the cold-start problem: deciding what to show first when we know nothing about the user. It can be addressed by asking a few initial questions about their tastes or presenting the safest defaults. (Appendix C covers AI-based solutions.) Once safely inside, the user recognizes from the first screen “what kind of game this is” (T4, first recognition). Then they tap or move something for the first time (T5, first action); if there is no reason to act, their finger stops. When the system responds (T6, first response), they feel that their action has reached the world. Progressing further, they make a mistake, get stuck, or lose for the first time (T7, first failure). This is where a failure that invites another try diverges from one that is merely irritating. At last, they complete the core task for the first time and receive a reward (T8, first reward), producing the realization: “Oh, so this is what it is.” This moment has a name of its own.

The aha moment. This is the moment when the user first feels the product’s core value. In a messenger, it might be sending the first message to a friend and receiving a reply; in a video-editing app, creating the first video. In a game, it may be winning the first battle or completing the first character. The metric that measures whether the user reached this aha moment is called activation, and it matters far more than sign-up or download counts. Even if a game has ten thousand downloads, if only fifteen hundred people reach the aha moment, then only fifteen hundred have actually tasted its first fun.

Many services are known to have defined their own aha moment as one clear action and focused on leading users to it. The collaboration messenger Slack is often said to have used the point when a team had exchanged a certain volume of messages as its “cross this line and they stay” benchmark; Facebook is often said to have used making a certain number of friends early after sign-up. The exact figures vary by service and period, but the principle is the same. Rather than pursuing a vague “good experience,” they identify a single action: “What action, performed once, makes the user feel the core value?” Our game has such an action too. We must first make clear whether it is completing the first character or experiencing the first act of cooperation; only then can we concentrate the design on reaching that point.

One more concept matters here. The time it takes to reach the aha moment is called time to value. The longer it takes to reach the first fun, the more people leak along the way, so one major goal of first-session design is to shorten that time. Long company logos, unskippable intros, and endless terms-and-conditions screens are all obstacles that increase time to value.

The third phase is settling in. After seeing the first fun, the user reveals their preferences for the first time or encounters another person’s traces (T9, first choice and social contact): customizing a character, checking a friend’s record, or sharing something for the first time. Then they return in the next session, the next day, or the next week (T10, first return). Whether they had a reason to come back is decided here, and the relevant metrics are D1 and D7. The endpoint of the first experience is this moment of return. T10 lies outside FTUE’s boundary at T8, but the seeds scored there are planted inside that boundary. The reason to return must be created before the first session ends. In that sense, the first experience does not end at T8; it is scored at T10. Business events such as payment are not part of the first experience but occur much later, and they are covered separately in the dashboard chapters in the latter half of this book.

Let us make one point explicit. The core of FTUE runs from T1 to T8, from first exposure to first reward, while the two thresholds of settling in (T9 and T10) connect that first fun to relationships and return. Trying to open the user’s wallet on day one and losing the chance to deliver the first fun is one of the most common mistakes in first-experience design. This mistake does not arise from ignorance. It usually arises from a fight over metric ownership. Marketing and monetization departments want to see first-day revenue appear in their own reports, and that pressure puts a payment pop-up before the first reward (T8). Payment is not a threshold but an output of a good first experience. The moment an output is inserted where an input belongs, the entire funnel narrows. The fight over whose number becomes the first experience’s report card is addressed head-on in the dashboard discussion in Chapters 22 through 24. (Appendix A contains a blank map for mapping all eleven thresholds onto your own game.)

In operational terms, FTUE spans T1 through T8, NUX spans T3 through T10, and UX continues without end from T0 onward. There is a reason NUX starts at T3 rather than T8. Settling in does not begin after the first fun; its score starts being recorded at installation. If entry is arduous and the quality of the first encounter is low, the probability of returning tomorrow is already reduced on the spot. Conversely, a smooth first session plants the seed of return in advance. The two ranges therefore overlap from T3 through T8. But if their responsibilities blur merely because they overlap, the division of responsibility established in Chapter 2 collapses, so we need one rule: when people leak in the overlap (T3–T8), FTUE bears primary responsibility; NUX takes the lead after T8. If people leave during the first session, suspect FTUE first. If they reach the first fun but do not return the next day, then suspect NUX. When a quick explanation is needed, the model can be reduced to just three phases: before the game, first session, and settling in. But whenever you need to locate a leak, return to the eleven thresholds. These eleven thresholds, T0 through T10, are this book’s canonical coordinates for dividing “first.” Even when later chapters introduce business events such as payments or revenue, those are not thresholds but outputs that occur much later. The ladder of the first always ends at T10.

The Game Conventions Hidden at Every Threshold

Reading the thresholds only as stages in time reveals only half the picture. Each threshold is also a point where a game’s genre conventions first collide with a new user. First recognition (T4) contains the question “What genre does this appear to be?” Whether an ordinary person reads the first screen as an “RPG” or a “decorating app” changes both what they expect and where they tap next. The question hidden inside first action (T5)—“What control convention does this demand?”—points out that gestures obvious to gamers, such as using a virtual joystick or pressing and holding, are a foreign language a newcomer must learn. The same applies to the “reward grammar” promised by first reward (T8). Level-up pop-ups, three stars, and gacha presentations signal achievement to gamers, but to an ordinary person they may pass by as meaningless screen effects. At each threshold, then, ask once more: is the user colliding with a wall of time, or with a genre promise they have never learned?

This second reading matters because the prescriptions for the two problems are opposites even when people leak at the same threshold. A wall of time can be reduced, removed, or postponed. But a wall of genre promises does not improve no matter how fast it becomes; the promise itself must be removed or replaced with something familiar. A game may spend weeks cutting one second of loading after half its users leave at first action (T5), while those users were actually leaving before the foreign language of the virtual joystick. Funnel numbers tell us only which threshold leaks. This dual reading tells us why. This is where the genre curse discussed in Chapter 1 acquires a specific address on the eleven thresholds.

▶ Three Questions to Apply to Your Screen

Am I treating only the first launch as FTUE?

Have I put pre-install touchpoints—ads, the store, and word of mouth—inside my area of responsibility?

Do I treat T0 through T2 as design targets?

If even one answer is “no,” redraw the stages before the first screen before polishing that screen itself.

What Must Be Defined First?

When we redraw the funnel with eleven thresholds, the earlier ten-thousand-person funnel looks entirely different. The fall from ten thousand to four thousand is a T1–T2 problem: the first impression and expectation formed in the store. The fall from four thousand to three thousand is a T3 problem: the cost of entry among people who installed but did not open. The fall from three thousand to fifteen hundred is a problem somewhere between T4 and T8, from the first screen to the first reward. Then what kind of problem is the drop from fifteen hundred to three hundred? We must not jump to a conclusion. Fifteen hundred is the number who “finished the tutorial,” not the number who reached the first fun (T8)—and that was precisely the point of Chapter 2. The interval therefore represents one of two possibilities: either we taught them but did not get them to T8, or we gave them fun but no reason to reach T10 and return. We can distinguish the two only by measuring arrival at the aha moment separately from tutorial completion. T9, first choice and social contact, can also be a quiet source of leakage in between. If users never had a chance to reveal their preferences or see another person’s traces, one entire reason to return is missing. Each interval has a different handle for repair.

The first thing to do, then, is place your game’s new-user traffic on these eleven thresholds and find the single threshold where the number falls most sharply. We cannot fix every threshold at once. Plugging the largest leak is a hundred times more effective than changing the color of the first screen.

Return to our hypothetical game. Suppose a short-form character-collection conversation game shows that users watch the first video to the end—passing first recognition and viewing—but half leave at the step immediately before collecting their first character (just before T8). The problem is not the first impression. It is the failure to give users a reason to move from “watching is fun” into the act of collecting. A common mistake here is to make the store screenshots more spectacular. The leak is at T8, but T1 gets repaired. Without separating the thresholds, the team repeats this wasted effort.

Applied to MEJE Aidong World, it looks like this. The moment a fan first meets and brings home an Aidong—a puppy resembling their favorite idol—is T8, the aha moment. Everything else follows only after the single achievement, “I have my own Aidong.” The first-experience design for Aidong World therefore concentrates on making T8 arrive as quickly and clearly as possible: shortening the time to choose, pet, and name an Aidong—the time to value—and amplifying the reward of the moment it comes to the user’s side.

Finally, let us raise our way of seeing thresholds by one level. The eleven thresholds are not only a diagnostic tool; they are design variables that can be rearranged. A well-made first experience changes their order. A playable ad pulls first action (T5) and a taste of first reward (T8) ahead of installation (T3). Wordle’s green-square sharing converts my first social contact (T9) into someone else’s first word of mouth (T0). It follows the same logic as placing a supermarket sample before purchase. Design is not limited to sending users down the ladder from top to bottom. After asking which leaky stage of the funnel to plug, ask one more question: which threshold can we move in front of which other threshold?

Reference Content

Examples from other media that reveal the concepts in this chapter.

General Apps

Collaboration messenger Slack’s “2,000 messages” benchmark: it is widely reported that once a team had exchanged a cumulative 2,000 messages, Slack considered the team to have “really used” the service, and retention among teams that crossed that point was very high.
Interest-selection onboarding in TikTok and Spotify: the first screen asks users to choose several topics or favorite artists, seeding an initial feed that otherwise has no information with those choices.
Wordle’s one game per day and shareable green squares: everyone receives the same single puzzle each day, building twenty-four hours of anticipation, while a colored grid that conceals the answer lets users show off the result and simultaneously creates a reason to return and word of mouth.
Sign-up funnel analysis: visits, sign-ups, first actions, and returns are separated to identify exactly which stage leaks.

Real-World Workflows / Psychology

Free supermarket samples: one taste shortens the distance to the first conversion, a purchase.

The remaining reference content is collected in the “Chapter 3 Appendix” at the end of this chapter. (In the printed book, it is grouped into Appendix D.)

Design Note ▶ Try It Yourself

Think about the new-user numbers for your game and complete this sentence.

“The threshold where our game loses the most people is T(　), and users leave there because (　　　).”

If you have no data, write down your best guess. It might be first exposure (T1), because “it does not show what kind of game this is”; first action (T5), because “I do not know what to tap”; or first return (T10), because “there is no reason to come back.” The pre-game phase (T0–T2) may seem impossible to measure because it produces no in-game logs, but proxy metrics exist. The percentage of ad viewers who proceed to the store and the percentage of store-page viewers who install act as thermometers for T1 and T2. If next-day retention after installation differs according to the ad creative that brought a user in, that is a sign that T2, first expectation, made a promise misaligned with the actual game. Trends in search volume for the game’s name become an indirect thermometer for T0, first word of mouth. The moment you identify one place, you know what to repair next.

Add one more line beneath the threshold you chose: “What piece of gamer common sense should the user not have needed to know at this threshold?” If the leak is caused not by time or friction but by a genre promise the user never learned, the repair handle is not screen speed. It is removing that promise or replacing it with something familiar.

If you cannot identify one place, you cannot repair any of the eleven thresholds. Start by pinpointing the single stage with the sharpest fall.

In one sentence: “First” does not happen once but eleven times. From T0, first word of mouth, through T10, first return, it divides into three phases: before the game, the first session, and settling in. The core of FTUE runs from T1 through T8, and its first reward—the aha moment at T8—is the finish line of the first fun. You can see where the leak is only by separating the thresholds. Next chapter: But is the “experience” we believe we give users at those thresholds truly an experience, or only a shadow shaped like one? Is pressing a button an experience?

Chapter 3 Appendix: Reference Content Collection

This collection checks the tool presented in Chapter 3—the funnel reading that divides “first” into eleven thresholds from T0, first word of mouth, through T10, first return, and identifies the stage where people leak—against cases from other media. It is grouped into video games and platforms, video and content, publishing/comics/music, mobile/apps/services, and offline/everyday life. For each case, pair the following questions with the chapter’s discussions of the funnel, time to value, and threshold rearrangement: Which threshold is operating here? What action constitutes the aha moment? What device rearranges the order of the thresholds?

Video Games / Platforms

Outer Wilds: in this 2019 game, “understanding” accumulates instead of stats, and the first reward arrives as an insight when scattered clues suddenly lock together. What to examine: an aha moment can be a cognitive event rather than a system reward such as a level or item, along with the cost that such a moment is difficult to bring forward.
Minecraft’s first night: while making the first pickaxe and building the first shelter before sunset, the user reaches the realization, “So this is what kind of game it is.” What to examine: the time pressure of day and night creates the first session’s objective on its own, allowing the system to manage time to value.
Stardew Valley’s first harvest: planting seeds, waiting several in-game days, and finally harvesting the first crop remains a clear first achievement. What to examine: a case where the aha moment is designed to arrive several days later rather than within the first session, and the small rewards that sustain the user in between.
It Takes Two: this 2021 two-player-only cooperative game reveals its “core value” only when the pair solves their first cooperative puzzle together. What to examine: a game whose aha moment requires another person adds the threshold of recruiting a partner to first entry (T3).
Steam’s two-hour refund policy: Steam officially offers refunds regardless of reason when playtime is under two hours and the purchase was made within fourteen days, turning the first two hours after purchase into a de facto evaluation period. What to examine: if the first session is also a refund review, time to value becomes a variable directly tied to revenue.

Video / Content

The Save the Cat! method: the screenwriting guide Save the Cat! specifies the structure of the first ten minutes by page count, establishing tone with an opening image on page one and introducing the catalyst around page twelve. What to examine: an approach to craft that manages the opening phase through coordinates—what happens on which page—rather than intuition, much like the eleven thresholds.
Episode-one drop-off curves in streaming: a funnel in which viewers leave from episode to episode, where the key question is which episode loses the most people. What to examine: saying “At which episode does it break?” rather than “Viewership is low” reveals what to repair, using the same language as threshold-level diagnosis.
Cold opens (Breaking Bad, Lost): a teaser that drops viewers into the middle of an event before the opening credits, binding them to the story before they change channels. What to examine: a rearrangement that postpones the explanatory section wholesale to accelerate first recognition (T4).
Review embargoes: the film and game industries contractually control when reviews may be published, managing the time when first word of mouth (T0) is released. It is also widely observed that an embargo lasting until just before release can itself be read as a bad sign. What to examine: the pre-game phase (T0–T2) is a managed design target rather than luck, along with the backfire in which control becomes a negative signal.

Publishing / Comics / Music

The hook of a novel’s first sentence and Chapter 1 drop-off: the point where a reader closes a book may be the cover, the first line, or the first chapter, and each stage has a different cause. What to examine: drop-off always occurs stage by stage regardless of medium, so prescriptions must also differ by stage.
Weekly Shōnen Jump reader surveys and early cancellation: readers vote each week for the works they liked, and serials rise or fall on those rankings; it is commonly reported that a series can be canceled as early as three weeks after debut. What to examine: how a market in which opening-phase results immediately decide survival solidifies a structure that concentrates resources in the first installment.
The web-novel first-episode “golden time” and cathartic opening: because readers may leave immediately if Episode 1 ends in frustration, standard practice is to preserve continued-reading rates by including satisfying developments and a small complete arc within the opening episodes. What to examine: the publishing version of shortening time to value by pulling first reward (T8) into Episode 1.
Streaming’s thirty-second rule and shortened intros: once a play had to pass thirty seconds to count toward streams and royalties, intros reportedly shrank from around twenty seconds in the 1980s to roughly five, while more songs moved the chorus forward. What to examine: a measurement standard that changed the form of content, demonstrating that what we measure determines what we make.

Mobile / Apps / Services

Playable ads: an advertising format that lets users directly control a miniature stage before installation, pulling first action (T5) and a taste of first reward (T8) ahead of installation (T3). What to examine: the chapter’s final argument—that thresholds are not a fixed ladder but design variables that can be rearranged—in its clearest form.

Offline / Everyday Life

Sales pipeline stages: conversions are tracked stage by stage from lead to contract to identify where the largest drop occurs. What to examine: funnel thinking was a basic practice of sales management even before digital products, and only by dividing the stages can responsibility and prescriptions be separated.
Test-driving a car: a procedure that puts the first action—the steering wheel—before the major decision to purchase. Dealerships standardize even the test-drive route and accompanying guidance to manage the first few minutes of impression. What to examine: in high-priced products, the same threshold rearrangement used in supermarket sampling becomes a process complete with route design.
Apartment model homes: the convention of building a full-scale sample unit of a home that does not yet exist, bringing the first impression and expectation (T1–T2) of an unrealized product forward into a physical experience. What to examine: the device that creates expectation (T2) also becomes a promise, and a mismatch with the actual move-in experience can return as a dispute.
The IKEA effect: experiments have reported that people assign greater attachment and value to objects they assembled themselves. What to examine: the achievement of first reward (T8) grows when users complete the first result with their own hands, so automatic progress that completes it for them can instead diminish the reward.