The Trigger Moves Inside

The Core Insight

Of smartphone owners, 79 percent check the device within 15 minutes of waking. A 2011 study counted 34 checks a day, and industry insiders put the real figure near 150 sessions. Nir Eyal opens Hooked with those numbers and one claim: a company's economic value now tracks the strength of the habits it creates.

A habit is an automatic behavior triggered by a situational cue, run with little or no conscious thought. Habits steer nearly half of daily actions from the basal ganglia. For a product, reaching that layer means unprompted use. The user feels something, and the product comes to mind first, with no ad in between.

Most product teams treat retention as a messaging problem and spend on push campaigns, email, and re-engagement ads. Eyal argues retention is an architecture problem, solved by a four-phase loop that moves the cue from the ad budget into the user's emotions. He taught the loop at Stanford's business school in 2012, after building a startup that placed advertising inside social games.

The Framework

The Hook Model cycles users through four phases until the product cues itself.

• A trigger cues the behavior, external at first, internal once the habit sets.
• An action is the simplest behavior done in anticipation of a reward.
• A variable reward pays out on an unpredictable schedule, which separates a hook from a plain feedback loop.
• An investment is a bit of user work that improves the product for the next pass.

The Habit Zone decides whether habit is reachable at all. Plot how often the behavior occurs against how useful the user perceives it over alternatives, and above a threshold curve the behavior goes automatic. A behavior of thin utility becomes a habit on frequency alone. A rare behavior never does, whatever its utility. Google runs on frequency with a thin edge over Bing, and Amazon runs on utility at lower frequency. Life insurance sits outside the zone and does fine.

The prize inside the zone stacks four ways: longer customer lifetimes, pricing power, faster viral growth, and defense. At a two-day viral cycle, 20 days of growth reaches 20,470 users. Halve the cycle to one day and the same 20 days passes 20 million.

Defense comes from Harvard's John Gourville. Users overvalue the old and makers overvalue the new, so an entrant must be nine times better to displace a habit. The QWERTY layout shipped in the 1870s and still beats the Dvorak layout patented in 1932, on switching cost alone.

Habit products enter as vitamins and become painkillers. Eyal defines the crossover as the point where skipping the action stings, an itch rather than a pain. Phil Libin's smile graph at Evernote drew the payoff in 2011. In month one, 0.5 percent of users paid. By month 33 the paying share reached 11 percent, and by month 42 it reached 26 percent. The longer the habit ran, the more the same users paid.

Key Ideas

Triggers Climb From Paid to Felt

External triggers come in four kinds. Paid triggers buy strangers. Earned triggers ride press and App Store features, and fade when the limelight moves. Relationship triggers travel on invites and word of mouth. Owned triggers occupy space the user granted, an icon or a push, and they alone prompt repeat use until the habit forms. Eyal's stress test is blunt: a social network that buys an ad for every return visit goes broke.

The destination is an internal trigger, a thought or emotion coupled to the product, and negative emotions couple strongest. Boredom, loneliness, indecision, and fear of missing out cue products with no prompt at all. A 2011 study tracked 216 undergraduates and tied depressive symptoms to heavier email, video, gaming, and chat. To find the emotion, Eyal borrows Toyota's 5 Whys. His worked example asks why a manager checks email, five whys deep, and lands on fear of being out of the loop. Instagram climbed the whole ladder, cross-posts, press, push notifications, then fear of losing a moment. Facebook paid 1 billion dollars for it in 2012, at 150 million users.

Start With Ability

BJ Fogg's model gates the second phase: behavior happens when motivation, ability, and a trigger arrive together in sufficient degrees. Motivation runs on three lever pairs, pleasure against pain, hope against fear, acceptance against rejection. Eyal's order is fixed. Raising motivation costs time and money, so start with ability, and make the product so simple that users already know how to use it. Simplicity means shrinking whichever of six factors is scarcest for this user at this moment.

• Time prices how long the action takes.
• Money prices the fiscal cost.
• Physical effort prices the labor involved.
• Brain cycles price the mental focus required.
• Social deviance prices how odd the action looks to others.
• Non-routine prices how far the action sits from existing habits.

The case studies all subtract. Twitter built its tweet button after finding links inside 25 percent of tweets, and the 140-character cap raised users' ability to create. Bias then stacks on ease. A car-wash card holding ten squares with two pre-punched beat a blank eight-square card by an 82 percent higher completion rate, on identical work.

Craving Runs on Anticipation

In the 1940s, Olds and Milner let mice self-stimulate the nucleus accumbens, and the mice crossed an electrified grid to keep pressing. Brian Knutson's fMRI work at Stanford located the pull: the region fires in anticipation of a payout and quiets on receipt. Skinner added the multiplier in the 1950s. Pigeons fed on a fixed schedule pecked steadily, and pigeons fed after a random number of taps pecked far more. Eyal's image for the predictable reward is the fridge light, which comes on every time, and nobody opens the fridge for it.

The rewards that run this circuit come in three types. Rewards of the tribe pay in acceptance and status. Stack Overflow draws 5,000 answers a day for upvotes and badges. League of Legends cooled its trolls with Honor Points that only other players confer. Rewards of the hunt pay in resources and information. Gamblers feed 1 billion dollars a day into American slot machines, and Pinterest crops images at the fold to bait the next scroll. Rewards of the self pay in mastery and completion, inbox zero and the next level in World of Warcraft. Mailbox manufactured that feeling by deferring low-priority mail, and Dropbox bought it in 2013 for a rumored 100 million dollars. Email runs all three at once: who wrote it carries tribe, what it holds carries hunt, sorting to zero carries self.

Variable Rewards Fail Three Ways

Cash is the wrong currency for most itches. Mahalo paid bounties for answers, peaked at 14.1 million monthly users, and slid, because payouts came too seldom and too small to matter. Quora launched in 2010, paid nothing beyond upvotes and status, and won the same market. Eyal warns that variable rewards are no magic fairy dust.

Coercion breaks the loop. In August 2012 Quora auto-opted everyone into a views feature exposing who read what, and a user revolt forced opt-in within weeks. A French study doubled bus-fare giving by closing with the line that the listener stays free to refuse. A meta-analysis of 42 studies and more than 22,000 participants found the same doubling. The technologies that win, in his telling, are the ones nobody makes us use.

Predictability drains the loop last. FarmVille reached 83.8 million monthly players in 2009 and booked more than 36 million dollars in 2010. The sequels re-skinned the same farm. Zynga carried a valuation above 10 billion dollars in March 2012, and the stock lost more than 80 percent by that November. Experiences with finite variability wear out with use, so multiplayer and user creation outlast solo plots. World of Warcraft held more than 10 million players eight years after release.

Investment Comes After the Reward

The fourth phase asks the user for work, and the placement carries the trick. The ask lands after the variable reward, while the brain is still justifying effort spent. In a Stanford experiment, people who used a helpful computer did almost twice as much work when the machine later asked for help. The IKEA effect prices the bias. In a 2011 experiment, makers valued their own origami at five times the bystander price, near expert-made levels. Consistency compounds it. Among homeowners who first took a three-inch safe-driver window sticker, 76 percent later accepted an ugly yard sign, against 17 percent approached cold.

The work stores value in five forms, content, data, followers, reputation, and skill, and every form raises the price of leaving. On Twitter's IPO day in November 2013, a Bloomberg commentator said a day's work rebuilds the technology. He was right, and it saved nobody: App.net shipped an ad-free version many judged better and still failed against years of stored followers. Investment also loads the next trigger. In 2010, 26 percent of mobile apps got exactly one use after download. Any.do asks for calendar access so its reminder fires right after a meeting, aimed at the anxiety of forgetting a task. Every Tinder match notifies both parties, and by mid-2013 the app counted 3.5 million matches from 350 million daily swipes.

Two Questions Gate the Build

Ian Bogost calls habit technology the cigarette of this century, and Eyal answers with a matrix instead of a rule. You ask two things: whether you use the product yourself, and whether it improves users' lives. Facilitators answer yes twice. Eyal argues they hold the moral high ground and the best odds, since they build for a user they know from inside. Peddlers believe in the product without using it, which describes most advertising. The odds sink when you design for a customer you do not know. Entertainers use products that improve nothing, which is art, hits-driven and quick to fade. Dealers answer no twice, and the book files them with casinos. Bogost built Cow Clicker to mock FarmVille, watched it hook players anyway, and shut it down.

Practical Applications

Run the book's five diagnostic questions on the product you own today.

• Name the internal trigger, the recurring itch the product relieves.
• Name the external trigger that brings users back this month.
• Find the simplest action done in anticipation of the reward, then cut steps until only it remains.
• Check that the reward satisfies and still leaves the user wanting more.
• Find the bit of work that stores value and loads the next trigger.

Excavate the trigger with user narratives and the 5 Whys. Write one session as a story from the user's side, ask why until an emotion appears, and stop there. Watch what users do over what they say, because declared preferences lie.

Then run Habit Testing, the book's port of build-measure-learn. Decide how often a user of your category must return, daily for a social product, weekly for a movie recommender. Count who does through cohort analysis. Eyal's floor for habitual users is 5 percent, and below the floor either the audience or the product is wrong. Above it, codify the Habit Path those users share. Twitter found that new users who followed 30 members tipped into retention, so it rebuilt onboarding to push following first. Modify your funnel toward the path, measure, and repeat, because the test never finishes.

Stage the investment: small work early, harder work later, every ask after a reward. If users refuse the work, shrink the ask instead of raising the motivation.

Who This Is For

Builders of products that need unprompted return get the most: consumer apps, social products, prosumer tools, anything whose retention chart decides the next round. Founders of rare-use or contract products sit outside the frequency axis and lose little by skipping it.

The mechanics carry an ancestry the book names and then walks past. Skinner's intermittent schedule is the slot machine's engine, and the casino floor appears here as an exhibit for hunt rewards rather than a warning. The ethics gate is self-assessed. The builder grades their own product on the two questions, and industry estimates put pathological users near one percent even for slot machines. Beyond a short duty-to-protect passage, responsibility lands on the user. The evidence is 2013 vintage and vendor-reported, heavy on survivors, and one exhibit, FarmVille, collapses inside the book's own pages.

The Decision

The gate comes before the loop. Write the two matrix answers down, and treat squirming on either as Eyal treats it, a failed test.

Then test the hook on your own product this week. Write the internal trigger as one sentence naming an emotion and the moment it fires. Pull cohort data and count the share of users who return at your category's natural frequency, then compare that share to the floor. A trigger sentence that names a feature instead of a feeling means the trigger work is unfinished. A share under the floor means the loop leaks, and the five questions locate the leak. Both answers fit on one page. Write that page before the next feature.