How to use User Journey Analysis
User journey analysis maps the messy reality of how shoppers move through your store — not the funnel you designed, but the loops, backtracks, and exits that actually happen. Here's how to read it and where to act.
User Journey Analysis
Mapping the real paths visitors take through a site — loops, detours, and exits — to find where intent breaks and conversion leaks.
User journey analysis is the practice of visualising the actual sequences of pages and events visitors follow on your store, not the linear funnel you wish they followed. Where funnel reports collapse traffic into stages, journey analysis preserves order, branching, and repetition — so you can see the PDP-to-cart-back-to-PDP loop, the search bar rescue after a failed category click, or the long tail of exits that never reach checkout.
It sits inside the broader practice of funnel analytics but answers a different question: not "how many made it through?" but "what did the ones who didn't actually do?" Sankey diagrams and path explorers are the standard visual format.
A funnel report tells you that 38% of cart viewers reach checkout. Journey analysis tells you that of the 62% who didn't, most looped back to the product page twice, opened the size guide, and exited from there — which is a completely different problem than "checkout friction."
That distinction matters because the interventions are different. Funnel gaps point you toward conversion-rate optimisation on a single step. Journey gaps point you toward content, navigation, or trust signals on a step you didn't even know was load-bearing.
What journey analysis reveals that funnels miss
Funnels are step-locked: they count whether a session hit each stage in order. Journey analysis is order-aware but step-flexible — it shows every transition between pages, weighted by frequency, so non-linear behaviour becomes visible instead of being averaged away.
The three patterns funnels reliably hide: backtracking (cart → PDP → cart again, usually a sign of price or shipping uncertainty), lateral browsing (PDP → PDP → PDP without ever adding to cart, a category-page or recommendations problem), and rescue paths (an exit-bound session that hits the search bar and recovers).
Each of these costs revenue in a different way, and each needs a different fix. A funnel report would lump them all into "PDP-to-cart drop-off" and send you optimising the add-to-cart button when the real issue is a missing shipping-threshold message.
Funnel vs journey, in one line
Funnels measure how many made it. Journeys measure what the ones who didn't were actually doing. You need both — funnels to size the problem, journeys to diagnose it.
How to read a Sankey diagram without getting lost
A Sankey shows page-to-page flow as ribbons whose thickness equals session volume. Read left-to-right: each column is a step deeper into the session, each node is a page or page group, each ribbon is a transition. The wider the ribbon, the more common the path.
Start by finding the two or three thickest ribbons leaving each major node — those are your dominant journeys, and they account for most of your revenue. Then look for surprising thick ribbons going "backwards" or to unexpected destinations. Those are your diagnostic signal.
Share of sessions following each journey pattern (apparel store, 90 days)
On a typical apparel store, the "clean" linear journey accounts for around 14% of sessions. The other 86% are doing something messier — and that's where the diagnostic and optimisation budget should go.
Common journey patterns and what they mean
Most journey insights collapse to a handful of recurring patterns. Recognising them by shape saves hours of staring at Sankeys — once you know the silhouette, you know roughly where to look for the fix.
The table below shows the four patterns we see most often across Shopify and WooCommerce stores in the €1M–€15M revenue band, with the typical share of sessions each represents and the intervention that usually moves the needle.
Frequent journey patterns, typical session share, and the usual fix
| Pattern | Typical session share | What it usually means | Where to intervene |
|---|---|---|---|
| Cart ↔ PDP backtrack | 8–14% | Price, shipping, or size uncertainty | Shipping threshold + size guide on PDP |
| Lateral PDP browsing | 20–30% | Weak category filtering or merchandising | Faceted filters, sort-by-bestseller default |
| Search rescue | 5–12% | Nav doesn't expose what users want | Promote search bar, fix zero-result queries |
| Checkout abandonment loop | 6–10% | Account creation or payment friction | Guest checkout, more payment methods |
| Help-page detour | 3–7% | Trust or policy concern | Surface returns/shipping copy on PDP |
Note that the highest-share patterns aren't always the highest-value to fix. A 25% lateral-browse rate that converts at 1.8% may be worth less than a 7% help-page detour that converts at 0.3% — because the latter is closer to purchase intent and the fix is usually a 30-minute copy change.
Turning journey insights into experiments
Journey analysis is diagnostic, not prescriptive. It tells you where attention is leaking; it doesn't tell you the winning variant. Every pattern you spot should become a hypothesis, and every hypothesis should become a ranked test in your experimentation backlog.
A reasonable workflow: extract the top three anomalous patterns each month, write each as an "if we change X, the journey should shift from A to B, lifting conversion by Y" hypothesis, then prioritise by reachable session volume × expected lift × implementation effort. The patterns that touch the most sessions and cost the least to test go first.
Don't optimise for the journey itself
The goal isn't a tidier Sankey — it's more revenue. A site can have messy-looking journeys and excellent conversion (curious shoppers browse a lot before buying), or clean journeys and poor conversion (decisive shoppers leaving fast). Always tie journey changes back to a conversion or revenue metric.
User journey analysis FAQ
Funnel analytics measures how many sessions pass each predefined step in order. Journey analysis preserves the full sequence of pages a session visited, including backtracks, loops, and detours. You use funnels to size where drop-off happens; you use journeys to understand why.
Roughly 20,000 sessions per month gives you enough volume for the top five or six journey patterns to stabilise. Below that, you'll see patterns but the long-tail ribbons in a Sankey become statistical noise and you should focus on session recordings instead.
Split them. Mobile and desktop journeys on a Shopify store are usually very different — mobile leans heavier on search and lateral browsing, desktop on direct category navigation. Aggregating hides the patterns that matter for each device's optimisation roadmap.
Four to six transitions is the sweet spot. Beyond that, the Sankey becomes unreadable and the volume per ribbon is too small to act on. If you need to investigate longer paths, segment to a specific entry point or goal and re-run.
Not really — journey analysis is session-scoped or user-scoped behavioural data, not channel attribution. You can combine it with attribution by segmenting journeys by acquisition source, but the tool answers behavioural questions, not media-mix ones.
Filter them aggressively before analysis. Bot sessions create unrealistic linear or single-page journeys at scale and will distort your dominant ribbons. Most analytics tools have a bot filter; turn it on, and additionally exclude sessions with implausible page-load speeds or zero scroll depth.
No — they complement each other. Journey analysis tells you which paths are common at scale; session recordings show you what's happening within a single instance of that path. Use journeys to find the pattern, then watch five recordings of it to understand the human behaviour.
GA4's exploration path report is a basic journey view — it shows page-to-page transitions but is limited in segmentation and event richness. Dedicated journey tooling adds event-level nodes (add-to-cart, search, coupon-apply), better filtering, and Sankeys that handle higher fan-out without breaking.
Monthly is enough for steady-state diagnosis. Run it more often around major site changes — a new theme, a navigation redesign, a Black Friday landing structure — because those are exactly the moments dominant journeys shift and old assumptions break.
Yes, and you should. Logged-in users typically have much shorter, more decisive journeys (returning shoppers know what they want), while anonymous traffic does the heavy browsing. Mixing them obscures both stories, and the optimisations for each segment are different.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.