River
Y CombinatorBacked by Y Combinator
Log InStart Free
FREE TEMPLATE

User Flow and Journey Map

Steps taken from what the product actually forces, routes ranked by how many people walked them, and every claim on the emotion row evidenced.

Edit with AIDownload Word and CSV

Free download  ·  No account needed

What you get

Pack of 7 files: 4 documents, 3 spreadsheets

User Flow and Journey Map
  • How the Journey Is DerivedDocument
  • Path MapDocument
  • Journey NarrativeDocument
  • Friction Findings and What They CostDocument
  • Step RegisterSpreadsheet
  • Path InventorySpreadsheet
  • Emotion and Evidence MapSpreadsheet

A journey map is five or six boxes across the top of a grid, one path, and an emotion line drawn by whoever held the pen. The stages come from how the product was designed, so they hold the steps somebody drew and none of the ones the product forces afterwards. At the fictional Kingsmoor the map had five boxes. The product made people pass through nine, and the largest loss in the whole journey was on one of the four that were missing.

Three kinds of step go missing every time. A wait is a step, and nobody designs a progress bar. A silent failure is a step, because a report that renders empty is not the same screen as one that renders. A prerequisite is a step, because sharing needs somebody to share with. Kingsmoor lost 1,764 accounts, 18.7% of everybody who signed up, waiting a median of thirty minutes for a first sync.

So this pack derives the steps from event data and ranks every route by how many people took it. Kingsmoor's 2,806 successful accounts took 214 distinct routes, and the designed path was the most common of them at 15.4%. Every claim on the emotion row carries a recording, a ticket or a quote joined to that step, or it is counted as an assumption. It reads the product the way funnel and retention analysis does, assumes the instrumentation behind it has been checked, and hands the steps nobody drew to the handoff register.

Five boxes on the map, nine steps in the product, 214 routes through it

Steps taken from the event stream instead of the flow diagram, routes ranked by volume, and every claim on the emotion row checked against the accounts that made it.

User Flow and Journey Map
RiverStep Register

Step Register

Nine steps, five of them drawn, and the biggest loss is on one that was not

Illustrative, for a fictional B2B product called Kingsmoor. Signup to a report shared with a colleague: 9,412 accounts entered in a ninety-day window and 2,806 reached the outcome, which is 29.8%. Rows come from the event stream rather than from the flow diagram.

StepOn the mapKindReachedLost hereMedian timeRepeated
Sign upYesScreen9,41201m 11s3%
Create workspaceYesScreen8,1091,30344s2%
Choose a data sourceYesScreen6,6341,4752m 18s19%
Authorise the connectionYesScreen5,2181,4163m 34s31%
First sync completesNOWait3,4541,76430m 40s44%
Pick fields for the reportYesScreen3,1872676m 36s58%
Report renders with data in itNOSystem outcome2,9712161m 2s27%
Invite a colleagueNOScreen2,884871m 49s11%
Share the reportYesScreen2,8067838s4%

The workshop drew its emotional low at the field picker, which loses 267 accounts and has the highest effort of any screen. The sync wait loses 1,764, is 6.6 times larger, and had no box on the map because it is not a screen. Effort and loss stay in separate columns for exactly this reason.

RiverPath Inventory

Path Inventory

214 distinct routes, and the designed one carries 15.4%

Every sequence of steps that reached the outcome, deduplicated and ranked by accounts. Rebuilt from raw per-account events rather than from a path exploration report, which groups everything past its top twenty nodes into a single aggregate.

RankStepsAccountsShareRoute
1943115.4%The designed path, no repeat and no detour
21440214.3%Report renders empty, back to the source, reconnect wider, through again
31338813.8%Invites a colleague first, then exports to a spreadsheet and returns
41729710.6%Three data sources tried before one syncs
5122418.6%Builds from an uploaded sample, connects the real source later
6 to 2149 to 411,04737.3%209 routes, none above 4%

Five boxes on the map, against a median journey of 13 steps, a mean of 14.2 and a ninetieth percentile of 24. And 1,151 accounts, 41.0% of everybody who succeeded, exported to a spreadsheet and came back mid-journey. That detour is in three of the top five routes and on no map anywhere.

RiverEvidence Map

Emotion and Evidence Map

34 cells, 6 with evidence, and 2 of those still current

Every experiential claim from the existing map, checked for evidence joined to that step by account and timestamp. A room agreeing is not evidence, which is why nobody had ever been able to say which cell was wrong.

StepClaimStatusEvidenceDatedWhat the paths say
Choose a data sourceUsers feel confident, the list is familiarAssumptionNoneWorkshop612 searched for a connector that does not exist
Choose a data sourceCannot find their warehouseEvidenced47 tickets90 daysSecond largest fixable loss
Authorise the connectionFrustration with permissionsEvidenced9 recordings30 days31% retry, so it is failing not confusing
First sync completesNot on the mapMissingNoneNever1,764 lost, 18.7% of entrants
Pick fields for the reportThe emotional low of the journeyStale6 interviews14 months267 lost, and the screen has been rebuilt since
Report renders with dataNot on the mapMissingNoneNever216 lost, and the empty render is silent
Invite a colleagueNot on the mapMissingNoneNever388 accounts do this step first, not last
Between sync and fieldsNot on the mapMissingNoneNever1,151 export to a spreadsheet and return

Twenty-eight of thirty-four cells were assumption, which is 82%. Assumption is a legitimate status here and unlabelled assumption is not, because the room was most confident about the cell pointing at the step that loses 2.8% of entrants.

What is in the pack

01

Step Register

Every step the product actually makes people pass through, marked designed or not, with accounts lost, median time on step, repeat rate and back-navigations. Kingsmoor's had nine rows against a map with five boxes, and four of the nine had never been drawn by anybody.

02

Path Inventory

Every distinct route from entry to outcome, ranked by accounts. Kingsmoor had 214 of them, the top five covered 62.7%, and 1,047 accounts sat in a tail of 209 routes that a path exploration report collapses into a single grey node.

03

Emotion and Evidence Map

Every experiential claim with the recording, ticket or quote joined to that step by account and timestamp. Kingsmoor's existing map had thirty-four filled cells, six cited evidence, and four of those six predated a rebuild of the screen they described.

04

Effort measured separately from loss

A step people repeat and then pass is failing for them. A step people leave from is a wall. Kingsmoor's field picker had the highest effort of any screen, six minutes and thirty-six seconds with 58% returning, and lost 267 accounts against the sync wait's 1,764.

05

Path Map and Journey Narrative

The volumes on every route and the story behind them, written so each section stands up without the interpretation. The detour that appears in most successful journeys and on no map gets its own section, because it is usually the most interesting thing in the analysis.

06

Friction Findings and What They Cost

One entry per finding, ranked by accounts lost, each classified honestly. A wait is not a copy problem, and a silent empty result is not a layout problem, so a document that proposes wording changes for either is a quarter spent on the wrong thing. Where the fix genuinely is the sentence, that is copy work with a standard of its own.

How it works

  1. 1

    Send the events, not the report

    Path data at account level, plus whatever research and recordings exist with their dates, and the map you already have. The map is there to be checked rather than followed.

  2. 2

    Enumerate the real steps

    Waits, silent failures and prerequisites included. Kingsmoor's five-box map turned into nine steps, and the biggest loss in the product was on one of the four nobody drew.

  3. 3

    Rank every route

    Deduplicate the sequences and count accounts per route. The designed path is often rank one and still a minority of a minority, which is the finding rather than a reprieve.

  4. 4

    Evidence the emotion row

    Every claim gets a recording, a ticket or a quote joined by account and timestamp, or it is marked as an assumption and counted. The count is the map's honest confidence level.

Frequently asked questions

Why not use the path exploration report in our analytics tool?

Because it hides the tail. Google's own documentation says the graph shows the top five nodes per step, expandable to twenty, and everything past that is grouped into an Others node. With 214 distinct routes, most of Kingsmoor's inventory lives inside that one grey box.

How does a thirty-minute wait become invisible?

A session ends after inactivity, and the default is thirty minutes. Kingsmoor's sync had a median wait of thirty minutes and forty seconds, so a session-scoped funnel reported the largest loss in the product as a clean exit. Stitch by account across the whole window instead.

Our page paths have record IDs in them. Does that matter?

Yes, and it breaks the analysis before it starts. Google treats a dimension with more than 500 unique values in a day as high cardinality and condenses rows past the report limit into an other row. Normalise paths to templates first.

Does this replace research?

No, it aims it. The evidence map shows which steps have research, which have stale research, and which have none. Kingsmoor's biggest loss had no research at all, because nobody schedules a session on a progress bar, and that blank was the finding.

The workshop said our hardest screen is the field picker. Were they wrong?

About the experience, no. It genuinely is the hardest screen: six minutes and thirty-six seconds median, 58% returning to it. They were wrong about the consequence. It loses 267 accounts and the sync wait loses 1,764, which is 6.6 times more.

What if most of our users take one path?

Then the count says so and you have lost an afternoon. In practice a designed path carrying more than a third of successes is rare, and the number that changes the conversation is not the share but the count of distinct routes people found instead.

Sources

  1. Google Help[GA4] Path exploration (opens in a new tab)
  2. Google HelpAbout Analytics sessions (opens in a new tab)
  3. Google Help[GA4] Cardinality (opens in a new tab)

Written by

Chandler Supple

Co-founder & CEO, River

Chandler has spent the last six years working in AI, building tools for millions of users and Fortune 100 companies. In 2025, he co-founded River to build a new kind of operating system for AI agents.

LinkedInGitHubX

Reviewed by

Brian Joseff, Co-founder, River

LinkedInX

Related

Funnel and retention analysis

The same event data read as a funnel, when you already know which steps matter.

Design handoff pack

Where the steps nobody drew get frames, copy, or a named default before anybody builds them.

Product analytics instrumentation audit

Run this first if you are not sure the events behind the path data are trustworthy.

More free templates

Accessibility Audit Remediation Plan

Free remediation plan that regroups your axe or WAVE export by the component that rendered each selector, then ranks the fixes by what they actually block.

Design Critique Framework Template

Four documents and three sheets that hold design feedback to a stated requirement, and count the critique that keeps coming back across reviews.

Design System Documentation Template

Free design system documentation template that puts four usage counts on every component, so the parts of the system nobody imports stop looking available.

Design to Development Handoff Checklist

Free design handoff template that derives every state your screens actually need from the API contract and the role model, then reports coverage as a fraction.

Find out how many routes your product really has

Send the events and the map you already have. The first thing back is the step count, and it is usually larger than the number of boxes anybody drew.

Map the real routes