User Flow and Journey Map

In a 2016 survey, Nielsen Norman Group asked 48 UX practitioners when journey mapping fails¹. The most common answer (36%) was a map whose purpose was not clearly defined upfront. Without that focus, participants said, maps become too general and unfocused.

Write the scope as one sentence before you open any data: who, from where, to what outcome, within what window. For example: new accounts, from signup to sharing a first report, within 90 days. Make the start and the outcome events you can count, since every number on the map depends on them.

Keep to one kind of user per map, because a new customer and an administrator take different journeys. Give people time to finish, too. If the median account takes 7 days to reach the outcome, a 7-day window misses about half the people who eventually get there.

The second most common answer in that survey (25%) was a map not based in reality: built from preconceptions and assumptions, then never checked against research. Collect the evidence first, and date every piece of it.

Ask whoever runs your analytics for event data by account, not a dashboard summary. If your interviews are still raw transcripts, a user research synthesis turns them into themes backed by each person's exact words.

Start from the event stream, not the design file. List what the accounts that reached the outcome did, in order, and mark each step as designed or not. Keep steps apart when they lose people for different reasons, such as choosing a data source and authorizing it.

Waits need their own row. The usability researcher Jakob Nielsen puts the limit for keeping a user's attention on a task at about 10 seconds². For longer delays, he writes, people want to do something else while they wait. A sync that takes 30 minutes is a step people can walk away from.

An empty result needs its own row too. Nielsen Norman Group notes that a totally empty screen leaves people unsure whether the system is working³, or whether they set something up wrong.

Group the accounts that reached the outcome by their exact sequence of steps, count each distinct route and rank them. Then find the designed path in the ranking and state its share of everyone who finished; even in first place, it may be a minority.

Build the routes from raw events rather than a path report. Google Analytics' path exploration, for example, shows the top five nodes in each step, up to 20 if you expand it, and groups the rest into an "Others" node⁴. Sessions add a second blind spot. Google Analytics ends a session after 30 minutes of inactivity⁵ by default, so in a session-based funnel a longer wait looks like an exit.

1. 1

   Export by account

   Pull every event with an account ID and a timestamp, for the whole window.

2. 2

   Clean the step names

   Strip record IDs from page paths, so /reports/4821 and /reports/977 become one step.

3. 3

   Stitch across sessions

   Order each account's events by time across the whole window, ignoring session breaks.

4. 4

   Keep the repeats

   A step someone repeats before passing is failing for them, so don't deduplicate it away.

Thoughts and feelings are the easiest parts of a map to make up. Nielsen Norman Group's primer says the thoughts should ideally be customer verbatims from research⁶; hold the feelings to the same standard. For each one, attach a recording, ticket or quote you can place at that step. If you can't, keep the claim, label it an assumption and count how many you have.

Date the evidence, because interviews from before a redesign describe a screen that no longer exists. Then measure effort apart from loss. A step people struggle through but finish is a cost; a step they quietly leave is a wall, and the two need different fixes.

Rank findings by how many people each step loses, then match the fix to the kind of step. A wait needs to get faster, stop blocking or give people something worth doing. A silent empty result needs a message saying what happened. A screen people keep retrying is usually failing for them, so fix what breaks before you touch the wording.

The practitioner survey's third most common answer (21%) was a map that nobody shares or uses, so give each finding an owner. Before anyone builds a fix, watch people try the step and write a usability test report on what stopped them. Then score the fixes with a RICE prioritization template.

The template on this page lays this out in three sheets and four documents, filled in with the Kingsmoor example. Open it with Edit with AI and River rebuilds the routes from your event export and finds the step that loses the most people. Then it offers to audit your current map.