Hook Model: Trigger, Action, Variable Reward, Investment
The Hook Model helps you diagnose or design a habit loop as Trigger, Action, Variable Reward, and Investment, so you can see why people come back (or why they do not) instead of treating retention as a feature list.
Framework Card
- Name:
- Hook Model
- Goal:
- Diagnose or design a repeatable engagement loop as Trigger, Action, Variable Reward, and Investment.
- Flow:
- Trigger → Action → Variable Reward → Investment
- Best For:
- Explaining why users return or drop after first use; Checking a product habit loop for a missing phase; Designing a simpler action or a better-loaded next trigger
Why it matters
A first session can look fine: the user signed up, clicked around, maybe finished a task. Then they do not return, or they return and nobody on the team can say what pulled them back.
Teams often answer with more notifications, a new badge, or a longer onboarding. Those tactics sit on one phase of a loop. If there is no cue the user already feels, the action is still hard, the reward is fully predictable, or nothing the user did last time made the next visit easier, the product is asking for willpower every time.
The Hook Model is a way to look at that cycle as a system: what prompts the behavior, how cheap the action is, what makes the outcome worth repeating, and what the user stored that loads the next prompt. The point is diagnosis, not a promise that “habit-forming” equals growth.
What it is
The Hook Model is a four-phase cycle described by Nir Eyal in *Hooked: How to Build Habit-Forming Products* (2014). Each pass through the loop is meant to make the next pass more likely.
| Phase | Question it asks |
|---|---|
| Trigger | What cue starts the behavior this time: something in the environment, or something in the user’s state? |
| Action | What is the simplest behavior the user performs in hope of a reward, and is it easy enough? |
| Variable Reward | What does the user get, and is the exact outcome unknown enough to stay interesting? |
| Investment | What did the user put in (time, data, content, social ties, money) that stores value and helps start the next loop? |
External triggers include notifications, emails, icons, and messages from other people. Internal triggers are states such as boredom, uncertainty, or a habitual thought. Over time, products that form habits often rely less on pings and more on those internal cues. That path is a teaching claim of the model, not a guarantee for every app.
Variable rewards are often grouped as tribe (social response), hunt (new information or resources), and self (progress or completion). Predictable prizes go stale faster. Investment is not “make them pay.” It is stored work that improves the next visit and often becomes the next trigger (a half-finished lesson, a playlist, a thread waiting for a reply).
The Hook Model does not count acquisition funnels, does not replace a jobs interview, and does not by itself make a product ethical or successful.
How it works
A useful Hook pass is a loop check, not four posters.
1. Name the behavior and the user
Pick one repeating behavior (open the app to do X) and one user in the case. Mixing several products or jobs in one loop produces a fake diagnosis.
2. Map Trigger
List external cues that currently fire, and any internal states the case actually supports. If the only cue is a paid push, say so. Do not invent loneliness as a trigger because a textbook mentioned it.
3. Test Action
State the action and whether it is easy given motivation and ability (the Fogg-style test used at this step). High motivation will not save an action that takes too many taps. Low friction will not save an action nobody wants.
4. Check Variable Reward
Name what the user gets after the action. If every visit yields the same predictable screen, the reward may be too fixed. If the “reward” is only a guilt badge, say that too. Tribe, hunt, and self are prompts for types of variability, not required checkboxes.
5. Check Investment, then close the loop
Name what the user stored, and whether that store loads the next trigger or improves the next reward. If Investment is missing, each visit starts from zero. If Investment exists but never cues a return (a profile nobody sees), the loop is still broken.
The implication is which phase to fix first, not a growth forecast.
How it compares
When another lens fits better, or when you need a complementary view, these frameworks do different jobs. They are not interchangeable labels for the same question.
| Framework | What it helps you see | How it differs from Hook |
|---|---|---|
| AARRR Model | Acquisition, activation, retention, referral, revenue as a metrics path | Counts stages of growth. Hook explains a psychological repeat loop, not the dashboard. |
| Fogg Behavior Model | Whether one behavior will occur given motivation, ability, and a prompt | Whole-method behavior design. Hook uses a similar test mainly at the Action step. |
| Jobs-to-be-Done | The job the customer hires the product to do | Progress the customer wants. Hook is how use might become automatic, not why the job exists. |
| Marketing funnel | Linear stages from awareness toward purchase or another conversion | Campaign or site path. Hook is a cycle that may sit after someone already uses the product. |
The Hook Model is the lens for a Trigger, Action, Variable Reward, Investment cycle. Other methods help when the question is growth accounting, a single-behavior equation, the customer’s job, or a linear funnel.
When to Use This Framework
- Product engagement design. People try the product once and do not come back, and you need to see which phase is missing.
- Retention when repeat behavior matters. Growth depends on return use, not only on getting a first visit.
- Habit-loop diagnosis. Users do return, but the team cannot name the cue, the easy action, the reward, or the stored effort.
Example
A concrete example makes the structure easier to reuse when you are under uncertainty.
Example: Language app with a strong first lesson and weak return
A language app’s first lesson is polished. Day-7 return is poor. The team’s instinct is more reminder emails.
A loop-oriented pass might look like this in spirit:
Trigger: External: scheduled email. Internal: little evidence that “I have five minutes on the train” is attached to this app yet.
Action: Start a new lesson. Setup still asks for a daily goal every session. Friction is high relative to a tired evening user.
Variable Reward: The lesson score is the same star animation every time. New words appear, but the social and progress variability is thin.
Investment: The user has a weak streak and no saved phrases. Nothing in the account makes tomorrow’s open feel like continuing a story they own.
Implication: Reminder volume is not the first missing piece. The action is still heavy, the reward is mostly fixed, and Investment does not load a next cue. A useful Hook outcome is to name those gaps. It does not invent a retention rate.
Example: Team chat that already has a loop
A small company lives in a chat tool.
Trigger: Internal: “I wonder if anyone replied.” External: badge on the icon, @mention ping.
Action: Open the thread (one tap).
Variable Reward: Hunt and tribe: new messages, reactions, an unexpected decision.
Investment: History, channels, files, and people already in the graph. Leaving is costly because the store is the work.
Implication: The loop is intact for this behavior. Hook here explains why return happens. It does not tell you whether the tool is the right job for the customer (that is a Jobs-to-be-Done question) or how acquisition is performing (AARRR).
Takeaway
What the Hook Model can help with
- Mapping a repeating product behavior as cue, action, reward, and stored effort
- Finding a broken phase (no internal trigger, too much friction, fixed reward, investment that does not load the next visit)
- Designing a cheaper action or an investment that actually cues return
- Separating “we sent more notifications” from “we have a loop”
What the Hook Model cannot replace
- Pirate metrics. AARRR counts acquisition, activation, retention, referral, and revenue. Hook does not produce those numbers.
- A full Fogg redesign. The Fogg Behavior Model is the equation for one behavior (motivation, ability, prompt). Hook uses a similar test at Action. It is not a substitute for Fogg’s whole method.
- Jobs-to-be-Done. Why a customer hires the product for a job is a different question from how a use loop repeats.
- A linear marketing funnel. Awareness-to-purchase stages describe a path through a campaign or site. Hook is a cycle after use. A marketing funnel page is a different job.
- An ethics or strategy finish. The loop can be used to help a user or to extract attention. Filling four phases does not make the product good, legal, or a business.
Honest scope: Hook structures a habit-loop diagnosis. It does not replace the methods above, and it should not be sold as guaranteed retention.
Frequently asked questions
It is for seeing a product habit as a four-phase loop: a cue, an easy action, a variable reward, and an investment that stores value and helps start the next visit. It is most useful when first use and return do not match, or when return is unexplained.
No. A notification is one kind of external trigger. A Hook is the whole cycle, including whether the action is easy, whether the reward stays interesting, and whether the user stored something that loads the next cue.
AARRR and funnels describe stages and counts along a path. Hook describes a repeating cycle around a behavior. You can have funnel numbers without a habit loop, and you can have a loop without healthy acquisition.
Four case-tied phases plus a loop implication: which phase is missing or weak, and whether Investment actually loads the next Trigger. If the output is four slogans or a generic “make it addictive” line, you still have inventory.
It can. The same phases can support a useful routine or an extractive one. The model describes the loop. It does not decide whether the product should exist, and it does not replace legal, ethical, or jobs-level judgment.