Journey Maps Gain AI Agents: Nielsen Norman Group Rewrites the 101

Every journey map, no matter its format, rests on five components — actor, scenario and expectations, journey phases, actions, mindsets and emotions, and opportunities. That's the framing a new Nielsen Norman Group primer, "Journey Mapping 101," puts at the center of the discipline. The primer's sharpest shift, though, concerns who the actor is: AI agents can now be the ones making the journey, not just the people who commission it.

The definition has not changed. A journey map is a visualization of the process a person goes through to accomplish a goal. You compile user actions into a timeline, add thoughts and emotions, and refine the narrative into a map. The terms "user journey map" and "customer-journey map" remain interchangeable. The authors concede that "customer" can mislead — in B2B products, buyers and end users are not always the same people — but argue that agreeing on content matters far more than agreeing on the label.

What are the five components?

• Actor. The persona whose experience the map shows. One actor per map; a university mapping both a student and a faculty member needs two separate maps, because their experiences differ. Actions must be grounded in research data.
• Scenario and expectations. The situation and goal — say, switching mobile plans to save money, expecting to find all decision information easily. Real scenarios require research; anticipated scenarios can rest on assumptions, but those assumptions must be labeled.
• Journey phases. High-level stages that structure everything else. For a flight: plan the trip, compare flights, book tickets, pack, check in, clear security, board, fly, collect luggage. Healthcare and ecommerce produce their own phase structures.
• Actions, mindsets, and emotions. Actions form a narrative of key steps, not a granular log. Mindsets work best as direct quotes from user research. Emotions plot as a single line across phases — where the user is delighted, frustrated, or between.
• Opportunities. Actionable insights paired with ownership and success metrics, answering where to improve, what to do, who owns each change, and how impact gets measured.

Journey maps suit scenarios that unfold over time, involve sequenced steps, or span multiple channels. A single dropdown selection does not warrant one.

How does a journey map differ from its neighbors?

The primer separates the tool from four lookalikes, using a ridesharing service as the running example:

• Experience map: a generic person's broad experience — how people get from A to B by walking, biking, driving, transit, or an app like Uber — revealing cross-cutting pain points such as unpredictable travel times.
• Service blueprint: the same journey plus backstage operations — driver matching, location tracking, saved-payment charging — connecting customer-facing interactions to the systems behind them.
• User flow: one task inside the product, such as requesting a ride, drawn as a flowchart with alternative paths for cases like no drivers available. No emotional layer.
• User-story map: functionality organized around user activities for planning releases. A journey map showing business travelers anxious about early flights becomes the story: "As a business traveler, I want to schedule a ride in advance so that I can plan my departure." Journey maps are for discovery, user-story maps for implementation.

Why map at all? Two benefits. The process forces conversation and a shared mental model — the primer calls fragmented understanding a widespread problem, since siloed success metrics leave no one responsible for the whole experience. And the artifact communicates understanding memorably, becoming a basis for decisions.

What changes when the actor is an AI agent?

This is the primer's most consequential section. People are no longer the only actors; agents research products, complete transactions, resolve issues. In some cases people never touch the company's website or app at all.

Two approaches emerge. You can map the person's journey with the agent's activities alongside — showing what the agent does, what blocks it, when the person must approve a decision, and the person's uncertainty or anxiety about the agent's choices. Or the AI agent becomes the primary actor, and the map reflects its point of view: tools accessed, decisions made, obstacles hit, points requiring human approval. Agent maps omit thoughts and emotions, for obvious reasons. The evidence rule holds either way — study agents through logs and observation, and draw human feelings from user research, not inference from agent behavior.

AI also accelerates the mapping craft itself: drafting hypothesis maps that surface assumptions, coding interview transcripts by stage and lane, clustering support tickets and reviews, prototyping interactive maps stakeholders can filter by channel, and generating "how might we" questions for identified pain points.

The caveat is blunt and worth quoting-scale weight: a journey map is only as reliable as the evidence behind it. The primer tells practitioners to treat AI outputs as starting points, demand supporting quotes and participant IDs, and verify them against originals. Be especially cautious, it warns, about AI-generated interpretations of users' thoughts and emotions — those require inferring what participants mean rather than what they say.

So the promise comes with its own audit trail. AI speeds the artifact; it cannot replace the research and team alignment that make the map worth building. As agents increasingly act on people's behalf, journey maps look set to become the shared instrument for seeing where human and machine activity meet — one evidence-backed phase at a time.