A car dealership group spends two sessions mapping its service booking journey. The map is good: fourteen steps from the reminder email to driving away, an actions row, what the customer is thinking, an emotion curve that dips hard around the courtesy-car handover. In the review, the operations director asks which team owns the wait between drop-off and the first call back. Nobody can answer from the map, because no row on it names a team, a system or a channel. It describes the wait accurately and cannot say whose it is.
That is a lane problem, not a research problem. The journey map lanes you choose are the rows of the map, and they decide which questions the finished artifact can answer. It is a smaller decision than the research behind a map, and a more consequential one than most teams treat it as.
What a lane actually is
A lane is one horizontal row of a journey map that holds a single kind of information across every step of the journey. Reading down a column shows everything happening at one moment. Reading across a lane shows how one dimension of the experience changes over time.
Steps and stages run left to right as time, and lanes run top to bottom as dimensions. Some tools call lanes layers, and process modelling uses swimlane for something different again: a row holding one actor's activity rather than one type of information.
The lanes worth having, and what each one costs
Almost every map that gets used carries the same short spine: what the customer does, what they are thinking or trying to achieve, how the experience feels, where the interaction happens, and where it breaks down. Those five rows are what a cross-functional group needs in order to agree on what is happening before arguing about what to do. Usage data from journey mapping tools shows the same pattern at scale: goals, an experience curve, process and channel information, and pain points appear in the large majority of real maps, a touchpoint row in about half.
Everything beyond that spine is a purchase. No lane is free once drawn, because every row multiplies the cells someone has to keep true when the journey changes. A twelve-step map with six lanes is 72 cells. Take it to ten lanes and it is 120, and the extra 48 do not fill themselves.
| Optional lane | What it lets the map answer | Worth carrying when |
|---|---|---|
| Backstage activity | Which internal work produces the customer-facing delay or error | The point of the map is connecting operations to experience |
| Systems and applications | Which platform a step depends on, and where two of them meet | Handoffs between systems are a known source of failure |
| Owner or team | Who has to change something, per step | The map will be used to assign work, not only to align |
| Metrics per step | How often this happens, how long it takes, what it costs | A number already exists and refreshes on a known cadence |
| Evidence and quotes | What the claim in the row above it is based on | The map will be challenged by people who were not in the research |
| Ideas and proposed changes | What has been suggested, and where | Suggestions are being scored, not just collected |
The metrics row is the one that most often goes wrong. Drawn in a workshop, it fills with estimates, and estimates in a numeric row read like measurements to anyone who arrives later.
How many lanes, and in what order
Two questions decide whether a map with the right rows is also a map people can read: how many rows it carries, and what sequence they sit in. Both tend to get settled by taste, and both have better answers available.
The range that stays readable
Five to eight lanes is the range where a map still works as a shared picture. The constraint is vertical space, not aesthetics: once a map has to be scrolled or zoomed to be discussed, it stops being something a room can point at. The failures at each end differ. With three or four rows, the map describes a sequence and explains nothing, so the review produces agreement and no decisions. With twelve, it becomes a document rather than a picture, read alone at desks, and the conversation it existed to create never happens.
How much of the journey you cover changes the number. Detail is inversely proportional to the timespan a map covers, a principle that runs through foundational service design work, and it applies to rows as much as to steps. A map spanning two years of a customer relationship supports fewer lanes than a map of a forty-minute checkout, because at that scale less can honestly be true of every step. The check is practical: project the map at the size it will be reviewed at, and see whether someone can find one specific cell while a colleague is still talking.
Order is an argument, not a preference
Rows read top to bottom as a chain, and readers follow that chain whether or not you intended one. So place what was observed above what was interpreted, and interpretation above what you intend to do. Invert that and the map argues badly: an opportunities row above the pain points row asks a reader to accept a proposed change before seeing the evidence for it, and reviews of maps built that way turn into debates about whose idea is better rather than about what customers experienced.
Two kinds of row sit outside that chain for a reason. A stage row labels the columns rather than describing the experience, and reference material gets consulted when someone challenges a claim rather than read as part of the narrative.
Adjacency matters more than most teams expect. Rows that get compared against each other should touch. If the argument of the map is that satisfaction collapses exactly where a handoff happens, the emotion row and the backstage row need to be neighbours, because comparison across four intervening rows does not happen visually.
What an empty cell is telling you
Blank cells are diagnostic, and they read two ways depending on how many there are. A row empty in a few places usually marks a research gap, which makes those cells a shortlist for the next round of interviews. A row empty in most places is a lane that does not belong here. As a working rule: if a lane is populated in fewer than half its columns and the gaps are not deliberate, it is either at the wrong level of detail or it belongs to a different map. A systems row on a lifecycle map usually fails this test, because most lifecycle steps do not correspond to a single system.
Resist filling blanks to make the map look finished. A cell completed for symmetry reads identically to one completed from evidence unless the map marks which cells are assumed.
Add, reorder or hide lanes as the map's job changes, without rebuilding it from scratch.
The lane stack decides what the map can be used for
The same steps with a different set of rows underneath them is a different instrument. What a map can be used for is decided by its stack, not by the journey it describes. A grocery retailer's click-and-collect experience, roughly ten steps from placing the order to unpacking the bags, produces three genuinely different maps depending on what sits below the step row.
- Lanes: actions, thinking, feeling, quotes, pain points
- Supports: agreeing what the current experience is and where it hurts
- Cannot answer: which team has to change anything
- Lanes: actions, touchpoints, feeling, pain points, proposed change, owner
- Supports: deciding what to build and who builds it
- Cannot answer: how common any of these problems are
- Lanes: actions, channels, backstage activity, systems, metrics, failure points
- Supports: finding where substitutions and late handovers are generated
- Cannot answer: why the customer reacted the way they did
The third one is worth sitting with. An operations stack can show that 18 percent of orders get a substitution and that most are decided during picking, without holding a line of the customer's reasoning, because none of its rows are built to carry that. It is an excellent diagnostic and a poor map to hand anyone who wants to understand why customers behave the way they do.
So before adding any row, name the decision it will inform and the person who makes that decision. A lane whose reader cannot be named is decoration with a maintenance bill attached.
The same test has a second edge. When a proposed lane would need a different level of detail from the rest of the map to be true, or would only interest people who never open the rest of the map, the answer is a separate map rather than another row. That is where a journey map hierarchy does the work instead, with maps at different altitudes carrying different stacks by design.
Lanes should keep changing after the workshop
Lanes accumulate because adding one in a workshop feels like progress and removing one feels like throwing away work, so the row set grows by one per review until the map has eleven rows, four of which were last touched during the session that created them. The deletion test is simple enough to run in a review: read a row across the whole map and ask which decision in the last quarter would have gone differently without it. If the answer is none, that row is documentation rather than instrumentation, and it is costing someone maintenance time to stay plausible.
Deleting is not the only option, and it is often the wrong one. Keeping a populated row available but hidden for a given audience is different from removing it, and a good share of the too-many-lanes problem is really an audience problem: an executive review and a delivery stand-up can read the same map with different rows visible. In Smaply, lanes are part of the map's structure alongside stages, steps and cards, so the row set stays a decision you can revisit rather than one frozen at the moment of drawing.
Consistency matters as soon as an organization has more than a handful of maps. Mandate a minimum spine that every map carries, for example actions, feeling, touchpoints and pain points, and let each team add what their journey needs on top. Two maps that share four rows can be compared and rolled up, whatever else they carry. Choosing rows is one half of deciding which journey map elements a map holds, and it is the half that decides whether those elements can be read together.
The dealership map ended up with an owner row under the pain points, and lost the ideas row it had carried since the first workshop, which by then held thirty-one suggestions in no particular order. The next review ran shorter and produced two named owners for the courtesy-car handover instead of a fresh list of things somebody might try.
Stages, steps, lanes and cards give every journey the same skeleton, so maps can be compared and rolled up.
Frequently asked questions
How many lanes should a journey map have?
Five to eight for a map that will be reviewed by a group. The real limit is vertical space: once the map has to be scrolled or zoomed to be discussed, people read it alone instead of together.
What order should journey map lanes be in?
Observed information at the top, interpretation beneath it, intended action at the bottom: actions and touchpoints, then thinking, feeling and pain points, then opportunities and owners. A stage row sits above everything as a column header, and reference lanes like evidence and attachments sit at the bottom.
Are lanes and layers the same thing?
In practice, yes. Some tools say layers where others say lanes, and both mean a row holding one type of information. Neither matches a swimlane in a process diagram, which holds one actor's activity rather than one dimension of the experience.
Do I need a backstage lane if I am not making a service blueprint?
Only when connecting internal work to customer-facing symptoms is part of the map's job. If the map exists to agree on what customers experience, a backstage row adds cells without changing a decision. If it exists to fix a delay, that row often holds the cause.
Should every map in our organization use the same lanes?
Mandate a small common spine and leave the rest to the teams. Four shared rows are enough to compare maps or roll them up, while a full standard stack forced onto every journey produces rows filled to satisfy a template.
Should every lane have something in every column?
No, and a fully populated row is not automatically a good sign. Scattered blanks usually mark research gaps worth closing. A row blank in most columns is either at the wrong level of detail for this map or belongs on a different one.
