Felt agency
Felt agency is the sense a person has, as they use a product, that they're the author of their own progress: that they're getting somewhere because of their own competence, not because the product is steering them. An experience with felt agency stays out of the way enough that the person credits the win to themselves, thinking "I'm getting good at this" rather than "this thing did it for me."
Definition
The user's experience of being the author of their own progress through a route, produced when a path leaves the runtime decisions about traversal to the user rather than pre-scripting them at construction-time.
The information gap between construction-time and runtime
Felt agency is an architectural commitment, anchored in an information gap between construction-time and runtime. The provider builds the path with structural knowledge — what the route looks like across many users, what affordances typically fit at each step, which sequence usually works. The user traverses the path with moment-knowledge — what they're trying to do at this step, what energy they have, what their decisiveness state is, what context just came up. Neither party's knowledge substitutes for the other's. The provider's construction-time knowledge can't reach the user's runtime knowledge because the runtime knowledge is generated at runtime.
Felt agency is the design criterion that names what this implies for path architecture: the path has to leave room for the user to make the runtime decisions only they can make. A path that supports user agency lets hopscotch criteria do the work construction-time can't reach. A railroaded wizard tries to operate entirely on construction-time guess; it works only when the guess happened to match this user's actual situation.
Conversation but one side pre-recorded
The compressed framing of how a well-constructed path operates. The provider's affordances are pre-recorded for likely runtime needs; the user's traversal is the live half of the conversation. The live half is where the information gap gets resolved. The provider doesn't try to script both sides — it records one side carefully, leaves room for the other, and the user fills it in at runtime.
Autonomy and efficacy together
A well-designed path produces two feelings at once: "I chose to do this" (autonomy) and "it worked" (efficacy). The two come apart in characteristic shapes. Autonomy without efficacy produces frustrated users who feel responsible for their own failure — they were free to act, but the actions didn't work. Efficacy without autonomy produces users who feel processed — things worked, but they didn't feel like the ones making them work. Either failure breaks felt agency; felt agency is what the joint achievement looks like, the autonomy-and-efficacy dyad operationalized at the route level.
User's relationship to the transition
Felt agency tracks the user's relationship to their own transition. Satisfaction with the offering's features is a separate dimension. A user can be dissatisfied with the offering's features and still experience felt agency — the route delivered, they feel they drove it. A user can be satisfied with the offering's features and fail to experience it — the Clippy pattern, where the offering's helpfulness makes the user feel led rather than leading.
Example
A SaaS onboarding wizard asks twelve qualifying questions, then produces a dashboard "configured for you." Every step is competent; the transition completes. The user finishes with no sense of having authored anything — the offering decided what to do, requested the inputs, and produced the output. Same offering, different shape: a working default workspace, common next actions surfaced as affordances, the user picks which to do first. Same construction-time work; different runtime relationship. The first treats the user's traversal as a problem to be solved by construction-time guess; the second treats it as the live half of the conversation.