--- title: "Hook Model (Nir Eyal): measuring each step of the habit loop | Skene" description: "Nir Eyal's Hook Model builds habits through trigger, action, variable reward and investment. Most steps leave events; the reward step barely does, so measure with care." url: https://www.skene.ai/resources/frameworks/hook-model doc_version: "51e269c7d116" last_updated: 2026-10-30 --- Resources / Frameworks # The Hook Model: four steps to a habit, and the one step events cannot see Hook Model. Credit: Nir Eyal, _Hooked_. By [Teemu Kinos](https://www.skene.ai/about), co-founder of Skene. The Hook Model, from Nir Eyal's book _Hooked_, describes how products build habits through a four-step loop. A trigger prompts an action, the action delivers a variable reward, and the user makes an investment that loads the next trigger. It is useful to any team designing engagement loops, with a caveat worth applying: a habit is only worth building if it serves the person who forms it. ## How it works Each step has a different relationship with data. Triggers, whether a notification or an internal cue, are partly visible as sends and opens. Actions are the easiest to count. Rewards happen in the user's head and leave little trace beyond whether the person comes back. Investments, such as saved content, followed people or configured settings, show up as stored state. That uneven visibility is the model's measurement trap. Teams optimise the steps they can count and infer the reward from return visits. When one of the counted events breaks, the loop looks broken, and a team can spend a quarter redesigning a reward that was working. ## What it assumes your events already do | Event | What the framework uses it for | What it reports when the event breaks | | --- | --- | --- | | Trigger delivered and opened `notification_opened` | The trigger step, as far as it is visible. | A new delivery channel records opens under a different event, and triggers appear to stop working. | | Core value action `report_viewed` | The action step. | A renamed action breaks the link between trigger and action, so the loop looks broken at its first step. | | Return visit `session_started` | The proxy for the reward. | A change in session tracking reads as rewards losing their pull. | | Stored investment `item_saved` | The investment step. | Saves made through a new surface go uncounted, and users look less committed than they are. | Event names are illustrative; yours will differ. The [dependency map](https://www.skene.ai/resources/frameworks#dependency-map) shows which other frameworks share each event. ## The reward that was redesigned for no reason _Illustrative example. The company and the figures are invented to show the mechanism._ A team measures its loop as `notification_opened` followed by `item_saved` within ten minutes. A new app version adds an in-app inbox that fires `notification_opened` for every notification in the list each time the inbox loads. The share of opens followed by a save collapses. The team concludes the reward has gone stale and spends a quarter on new content formats. The habit had not changed. The open event had. ## Checks before you trust it - Track each step with its own named event, including where triggers are delivered. - Measure return and investment as well as action, so one broken event cannot make the whole loop look dead. - Re-check open tracking whenever notification delivery changes. - Ask users about the reward directly; it is the one step the events cannot show. ## Heard in The Question Interviews in [The Question](https://www.skene.ai/resources/the-question) that bear on this framework: - [Interview 01: the founder who uses product data to retire a recurring idea](https://www.skene.ai/resources/the-question/anonymous-founder). Product data confirms or kills a hypothesis that people formed first, and the expensive gap is the number nobody collects. ## Check the events under it Before Hook Model goes into a plan, find out whether the events it needs exist and still fire the way it assumes. The [free audit](https://www.skene.ai/signup) maps what your product collects today: sign in, connect a GitHub repository or Supabase, and Skene returns one audit, with no card. On Pro, Skene Cloud reviews every pull request on a linked GitHub repository for removed or broken tracking, so a pull request that removes or renames one of these events is flagged before it merges. ## Related - [Cohort retention analysis: why the blended number hides the trend](https://www.skene.ai/resources/frameworks/cohort-retention-analysis) - [The aha moment: finding your own number instead of borrowing Slack's](https://www.skene.ai/resources/frameworks/aha-moment) - [Viral loops and the K-factor: invites times conversion, and the invite nobody counted](https://www.skene.ai/resources/frameworks/viral-loop-k-factor) - [Jobs to be done: the progress a customer hires your product to make](https://www.skene.ai/resources/frameworks/jobs-to-be-done) - On the blog: [Retention loops vs growth loops: which to build first](https://www.skene.ai/resources/blog/retention-loops-vs-growth-loops) - Glossary: [Side effect](https://www.skene.ai/resources/glossary/side-effect) - Glossary: [Moved event](https://www.skene.ai/resources/glossary/moved-event) - Glossary: [Event name](https://www.skene.ai/resources/glossary/event-name) - Glossary: [Cohort](https://www.skene.ai/resources/glossary/cohort) ## Sitemap - [Markdown sitemap](https://www.skene.ai/sitemap.md): every page on this site, with its markdown counterpart. - [Index for language models](https://www.skene.ai/llms.txt): the same set as a link-dense index. - [Full corpus](https://www.skene.ai/llms-full.txt): reference content concatenated as plain text.