DESIGN METHOD · VERSION 1
A sequence should lead
into the real task.
Snowball’s design moves from easy actions to contact with an avoided task. The five phases below guide that progression; they are not five required screens, and they are not a clinically validated protocol.
The five-phase model
- Easy win: a very manageable action, sometimes unrelated to the target task.
- Momentum: a slightly larger action that still feels achievable.
- Bridge: prepare the environment or object needed for the task.
- Contact: touch the actual work in a small way, such as reading one question.
- Go: do a meaningful piece of the target activity.
A short sequence can combine or omit phases. If you can already open the document, there’s no requirement to drink water or tidy anything first.
Difficulty decides the number of actions
Easy tasks may need 2–3 actions total. Harder or more complex tasks can need more because they involve locating information, understanding a brief, or gathering materials. Added steps should solve those barriers, rather than add unrelated chores.
The web starter gives you a choice of three or six preset actions. It does not infer difficulty, learn your preferences, or adapt the entire route with AI. Those are planned mobile-app behaviors.
“Too much” lowers the next action
In the web tool, each action has two smaller alternatives. At the smallest version, you can try the next action, pause, or stop. This bounds the interaction instead of making you repeatedly request a smaller version forever.
We do not require a timer to finish before marking an action done. The durations are rough effort cues, not measurements of how long you should take.
We distinguish clicks from starting
Finishing cards is not proof that you started the task. The tool asks whether you started and shows your own answer. Skipped actions are not counted as completed actions.
For the planned app, confirmed activation rate means sessions with an explicit Yes outcome divided by all started sessions. Unanswered sessions remain unknown, and Almost is reported separately. We plan to report response rate alongside activation so missing feedback is visible.
Evidence and limitations
We have not published a controlled trial, a user outcome dataset, or a comparison establishing that Snowball works better than another app. The examples illustrate the product design; they don’t prove its effectiveness.
The Centre for Clinical Interventions’ procrastination resources provide broader published context for practical strategies. That organization has not tested or endorsed these Snowball sequences.
What changes next
Version 1 records the current curated website behavior. We’ll update this page when generation, personalization, persistence, or outcome collection changes in the released product. Future reports will need actual data, a defined sample, and visible limitations.
Try a starting sequence →