TL;DR: Habits are built in four steps — Trigger, Action, Variable Reward, Investment. Nir Eyal’s Hook Model explains why some products become unconscious reflexes while others require constant marketing to maintain usage. The secret isn’t engagement. It’s making the behavior automatic.
What it means
A habit is a behavior done with little or no conscious thought. Checking your phone when bored. Opening Twitter when waiting in line. Reaching for Slack when you want to feel productive without committing to the work. These aren’t choices — they’re reflexes. Products that create habits don’t have to compete for attention each time; they own a slot in the user’s behavioral repertoire. That is what defensibility actually feels like at the level of the individual user (hooked).
The Hook cycle runs in a loop:
- External trigger (notification, email, recommendation) prompts an
- Action (open the app), which delivers a
- Variable reward (something surprising or satisfying), which motivates an
- Investment (follow someone, upload a photo, customize a setting, write a tweet)
The investment loads the next trigger, and the cycle repeats. Over successive loops, external triggers give way to internal ones — emotions like boredom, loneliness, or FOMO that automatically route the user back to the product. When the trigger has migrated inside the user’s head, you have a habit. Until then, you have a notification campaign (hooked).
The argument
Variable rewards create craving — fixed rewards do not. This is the psychological engine and the part most copycat habit designers get wrong. Eyal identifies three types of variable reward:
- Tribe — social validation (likes, comments, status, followers).
- Hunt — material acquisition (deals, information, the next interesting thing in the scroll).
- Self — mastery, completion, competence (Duolingo streaks, Wordle stats, leveling up).
The variability is what matters, not the magnitude. A predictable reward satisfies; a variable one creates anticipation, and anticipation creates the loop. This is exactly why slot machines work and why Twitter is so effective at hijacking attention — the next pull might be the one that pays.
The action must be easier than thinking about it. This is where simplicity-as-strategy meets habit formation. The Action phase succeeds only when the behavior is easier than the alternative. Scrolling a feed is easier than deciding what else to do with your hands. One-click buying is easier than comparison shopping. Every bit of friction you add to the Action step reduces the probability of the habit forming. The formula is unforgiving: Motivation + Ability + Trigger must converge at the same moment (hooked).
Investment is what separates hooks from gimmicks. After the reward, the user puts something into the product — data, content, followers, reputation, customized settings, conversation history. Each investment makes the product more valuable to that specific user and loads the next trigger. This is the compounding mechanism. A Twitter account with 30 follows is more valuable to its owner than a fresh one. A Spotify library with years of listening data delivers better recommendations. Investment creates switching-costs by stealth — the user is locking themselves in and doesn’t notice (hooked).
Frequency × perceived utility determines habit potential. Not every product can or should be habit-forming. A daily-use product with moderate utility (social media) has higher habit potential than an annual-use product with high utility (tax software). Products in the high-frequency, high-utility quadrant — messaging, search, payments — become the deepest habits and are nearly impossible to displace once formed.
The connection to network-effects. Investment in a networked product is simultaneously a network contribution. When you follow 30 people on Twitter, you’ve invested in your own experience and become part of the network that makes the product valuable to others. The Hook Model and network effects are complementary flywheels: hooks drive individual retention, network effects drive collective value. Together they create products that are nearly impossible to leave — and that’s the explicit design goal.
Twitter’s tipping point was 30 follows. Below that threshold, the feed is sparse and the reward is too infrequent to form a habit. Above it, the variable reward of the feed becomes reliable enough to trigger craving. This is the habit-formation equivalent of the atomic-network — the minimum individual investment threshold where the hook starts to hold (cold-start-problem).
The AI assistant version
ChatGPT may be the strongest hook ever built into a non-social product, and it works through a category of variable reward the original Hook Model didn’t quite anticipate (chatgpt-pmf). The trigger is internal almost immediately (“I have a question / I want to think out loud”). The action is a single text box with no friction. The reward is variable in a deeply unusual way: every response is novel, because the underlying generation is non-deterministic and the conversation context keeps shifting. And the investment is conversation history + memory, which feeds back into the next response and makes the relationship feel more personal over time. This is the relationship-effects story told in Hook Model terms.
Loose threads
- The ethics of habit-forming design are real and increasingly contested. Eyal himself has walked back some of his original enthusiasm. At what point does “building habits” become “manufacturing addiction”? The answer is uncomfortable and the line is moving.
- Retention metrics matter more than engagement spikes. Duolingo’s CURR framework (duolingo-growth) and Cohen’s “retention before growth” (pmf-roadmap) both show that habit strength — not novelty — drives durable growth. Beware the novelty-effects trap: early engagement may be curiosity, not hooks.
- distribution and hooks interact: a strong distribution channel gets users to the first trigger, but only the hook keeps them coming back. Most products fail at one or the other. The rare ones that nail both become verbs.