How to use Path Analysis
Path analysis maps the actual sequences shoppers take through your store — and reveals which pages predict a sale and which quietly kill it. Here's how to run one end-to-end.
Path Analysis
Path analysis is the quantitative study of the page sequences users follow before converting or abandoning.
Path analysis reconstructs the actual routes shoppers take through your store — not the idealised funnel a strategist drew on a whiteboard, but the messy, branching reality. It counts how often each sequence occurs, weights it by outcome (purchase, signup, exit), and surfaces the patterns that predict conversion or churn.
Unlike funnel analytics, which forces journeys into pre-defined stages, path analysis is exploratory. You start from a node — usually a landing page, a product page, or the checkout — and let the data show what really happens next. The output is a list of high-frequency or high-impact paths, often counter-intuitive: a FAQ visit that doubles checkout conversion, a size-guide loop that signals abandonment, a category re-entry that flags indecision.
Most teams instrument a funnel — Home → PLP → PDP → Cart → Checkout — and stop there. The funnel tells you where drop-off happens. It does not tell you why one shopper bounced after the PDP while another circled three times and bought.
Path analysis fills that gap. It treats every session as an ordered sequence of events and asks which sequences correlate with the outcome you care about. The answers are often surprising: shoppers who view shipping policy mid-session convert at a much higher rate, while shoppers who return to a category page from the cart almost never come back.
What path analysis actually measures
At its core, path analysis works on a directed graph. Nodes are pages or events (PDP viewed, size guide opened, cart added). Edges are the transitions between them, weighted by how many sessions made that jump.
From that graph, three outputs matter. Forward paths show what users do AFTER a chosen step — useful for diagnosing a leaky PDP. Reverse paths show what users did BEFORE converting — useful for finding hidden assist pages. Loops and re-entries show indecision or confusion — useful for spotting UX friction the heatmap missed.
The unit of analysis is the session, not the pageview. A session that touches PDP → Reviews → PDP → Size Guide → Cart is one data point with a five-step sequence, not five separate hits. This is what separates path analysis from raw clickstream reporting.
Path analysis vs. funnel analytics
Funnel analytics is confirmatory — you define the stages and measure conversion between them. Path analysis is exploratory — you let the data tell you which stages actually exist. Use funnel analytics to monitor a known journey; use path analysis to discover the journey you didn't know you had.
How to run a path analysis
Pick a starting node and an outcome. A common setup on a Shopify apparel store: start node is any product page view, outcome is purchase completion within the same session. Pull every session that touched the start node over a 30-day window — you want at least 5,000 sessions for stable patterns, ideally 20,000+.
Then split the sessions into two cohorts: converters and non-converters. Rank the top 20 sequences in each cohort by frequency. The interesting paths are the ones whose share differs sharply between cohorts — a sequence that's 18% of converter sessions but 3% of non-converter sessions is a signal worth investigating.
Conversion rate by post-PDP path (apparel store, 30-day window)
The pattern above is typical. Shoppers who touch shipping or reviews mid-session convert at roughly twice the baseline — they're resolving objections. Shoppers who loop back from the size guide to the PDP without progressing are almost always lost. That single insight is enough to brief a size-guide redesign experiment.
Patterns worth looking for
Across hundreds of e-commerce stores, a handful of patterns repeat. The assist page — usually FAQ, shipping, or reviews — predicts conversion when visited mid-session but predicts exit when visited from the cart. The loop, where users bounce between two pages, almost always means a missing piece of information. The category re-entry from cart signals price hesitation.
The trick is to weight by outcome, not just frequency. The most COMMON path on most stores is Home → Exit, which is useless. The most VALUABLE paths to find are the ones where conversion rate diverges sharply from baseline — high or low.
Typical path patterns and what they usually mean (apparel, beauty, electronics)
| Path pattern | Conversion lift vs. baseline | Typical interpretation |
|---|---|---|
| PDP → Reviews → Cart | +60% to +110% | Social proof resolved an objection |
| PDP → Shipping policy → Cart | +80% to +150% | Free-shipping threshold or speed was the blocker |
| Cart → PLP → Cart | -40% to -60% | Price hesitation or comparison shopping |
| PDP → Size guide → PDP → Exit | -70% to -85% | Sizing uncertainty unresolved |
| Search → PDP → Cart | +40% to +90% | High-intent shopper with known SKU |
| Home → PLP → PLP → PLP → Exit | -50% to -70% | Discovery failed, navigation or filters weak |
The size-guide loop is the one most teams underestimate. On a typical apparel store, 8–14% of PDP sessions touch the size guide, and of those, roughly half exit without adding to cart. That's often 3–5% of total revenue sitting in a single fixable UX problem.
Turning paths into experiments
A path is a finding, not a fix. Once you've identified a pattern, write the hypothesis explicitly: 'Because converters disproportionately view the shipping policy mid-session, surfacing shipping terms on the PDP will lift PDP-to-cart conversion by 8–12%.' Then run the test.
Good candidates for experimentation usually come from negative patterns — loops, dead-end re-entries, abandonment paths. Turning the size-guide loop into a hypothesis is straightforward: pull the size chart into a PDP modal, or add fit recommendations based on the user's previous purchases. The path analysis told you where to look; the test tells you whether your fix works.
Correlation, not causation
A path that correlates with conversion does not necessarily cause it. Shoppers who view the FAQ may convert more because they're more engaged to begin with — not because the FAQ persuaded them. Always validate path insights with a controlled A/B test before declaring a winner.
Frequently asked questions
Funnel analytics measures conversion through a predefined sequence of steps you already believe in. Path analysis discovers which sequences actually occur and which correlate with outcomes. Funnels confirm; paths explore. Most stores need both — funnels for monitoring, paths for diagnosis.
At least 5,000 sessions touching your start node for stable top-20 sequences, and 20,000+ before you can confidently compare converter and non-converter cohorts. Below that, rare paths look like noise. If you're a smaller store, widen the time window to 60 or 90 days instead of forcing a 30-day view.
Both, for different questions. Forward paths from a leaky page show where users escape to — useful for fixing drop-off. Reverse paths from the order confirmation show which pages converters touched before buying — useful for identifying assist content like reviews, FAQs, and shipping policies.
GA4 has a Path Exploration report under the Explore section. It works for shallow analyses — two or three steps from a start node. For deeper sequences, cohort comparison, or weighting by revenue, you'll typically need a dedicated analytics tool or to export GA4 events to BigQuery and query them directly.
A session is a single continuous visit, usually defined as activity with no more than 30 minutes between events. Path analysis treats one session as one ordered sequence — so a shopper who returns the next day starts a new path. Cross-session journey analysis is a related but separate technique.
Track them as events, not as separate pages. A typical apparel session might be PDP → filter:size=M → PLP → PDP, where the filter is an event on the PLP. Including these interactions makes sequences far more informative than a page-only view, which collapses meaningful behaviour into identical-looking PLP visits.
Ranking paths by raw frequency. The top paths are almost always trivial — Home → Exit, single-page sessions, direct PDP → Cart. The valuable paths are the ones where the converter share diverges sharply from the non-converter share. Always compare cohorts; never rank on volume alone.
It tells you WHERE and in what CONTEXT, not why. A size-guide loop tells you sizing is unresolved; it doesn't tell you whether the chart is unclear, the units are wrong, or fit reviews are missing. Pair path findings with session replays or surveys on the suspect pages to get the qualitative 'why'.
A full exploratory pass every quarter is usually enough — site changes, traffic-mix shifts, and seasonal behaviour all reshape paths over time. For specific projects, like a checkout redesign or a new product launch, run a focused path analysis before and after to measure how journeys actually changed.
Yes, and you should split them. Mobile and desktop paths differ sharply — mobile sessions are shorter, touch fewer pages, and rely more on search than navigation. Pooled paths hide device-specific friction, especially on checkout where mobile drop-off is usually 1.5–2x desktop.
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.