"Customer journey" workshops produce beautiful diagrams — smiling persona, tidy stages, weather emojis. Then you open the data, and real visitors do something else entirely. Journey analysis worthy of the name starts from observed behaviour; here is the method.
Quick Answer: how do you analyse a customer journey?
In five steps, backwards:
- Start from the conversion — identify the real paths of sessions that succeed: entry pages, sequences, durations.
- Compare with abandonments — same data for failing sessions; the divergence points are your priorities.
- Segment — mobile/desktop, new/returning and per-channel journeys don't look alike: averages crush them.
- Zoom in with replay on the critical nodes to understand the why.
- Fix and test — every gap between the designed funnel and the real journey becomes an A/B test hypothesis.
Minimal tooling: flow analytics + session replay + behavioural segmentation.
The designed journey versus the real one
Your site has an implicit script: home → category → product → basket → payment. Visitors, though, arrive on a blog post from a niche query, open three product tabs, vanish for two days, come back via a brand search straight onto a product page, then pass through the FAQ before the basket. Both truths matter: the script structures the site, the behaviour reveals where it cracks.
Three typical gaps, and what they teach:
- The entrance isn't the planned door. 60% of entries through the blog or product pages? Every important page must be able to play homepage: clear proposition, navigation towards the action, bounce under control.
- The comparison loop. Repeated back-and-forths between 2-3 product pages signal an unserved comparison need — a comparison table or "similar products" module solves it.
- The FAQ detour before converting. Frequent and instructive: the question that pushes people to the FAQ (delivery? returns? commitment?) should have been handled in the main journey.
The backwards method, in practice
1. Reconstruct the winning paths
Over a representative period, isolate converted sessions and look at: dominant entry pages, frequent sequences, typical duration and page count. You get 3 to 5 "royal roads" — the paths your site already knows how to make succeed.
2. Confront them with the losing paths
Same exercise on non-converted sessions that showed intent (several pages, meaningful time — not flash bounces). The useful questions: where do these sessions diverge from the royal roads? What is the last page seen before abandonment? That exit-page ranking weighted by intent is worth every audit.
3. Segment before concluding
An "average" journey doesn't exist. New visitors explore, returning ones sprint; mobile fragments into several short visits what desktop does in one long one; SEO traffic lands lower in the funnel than social. Analyse per behavioural segment — otherwise you optimise for a visitor who doesn't exist.
4. Zoom into the nodes with replay
Flows show WHERE journeys fork; session replay shows WHY. Five to ten sessions on a critical node (typically the #1 exit page of intent-showing sessions) are enough to see the friction: unfindable information, misleading element, discouraging form.
5. Fix, test, re-measure
Every gap becomes a hypothesis: "adding the product comparison will reduce loops and increase add-to-basket". Off to the A/B test, measured on final conversion, and back to step 1 — the journey is a living organism; every campaign and season reshapes it.
Three advanced readings that pay
- Time to conversion. If the median is D+3, your site must support the return: basket reminder, reassurance content, cross-session consistency. Judging a campaign on day D underestimates its real value.
- Multi-channel journeys. Today's SEO visitor returns direct tomorrow. Last click lies: look at the ENTRY channels of converted journeys, not only the converting channel — our acquisition view details the quality reading.
- Aggregated friction signals. Long hesitations, repeated clicks, back-tracking: aggregated per page, these signals prioritise UX work better than any expert opinion.
In practice with Mirage — The journey reads at every level: page sequences and exits in the funnels, individual sessions in replay with a typical journey per segment, and channel × behaviour × conversion crossings — cookieless, hence on the entirety of your visitors. Free 30-day trial.
FAQ
What is the customer journey?
The set of steps a person goes through between discovering a need and the final action (purchase, subscription), then beyond (loyalty). On a website it materialises as concrete paths: entry pages, sequences, hesitations, returns, conversion or abandonment — observable data, not assumptions.
What is the difference between customer journey and conversion funnel?
The funnel is the ideal path YOU designed, in ordered steps. The journey is what visitors ACTUALLY do: entries through unexpected pages, loops, comparisons, multi-day pauses. The analysis consists precisely of confronting the two — the gaps are the lessons.
Which tools analyse a customer journey on a website?
Three levels: flow analytics (which pages follow each other, where people exit), session replay (living through individual journeys), and behavioural segmentation (comparing converters' paths with the rest). Workshop journey maps remain useful for intent — data settles the facts.
Where do I start a journey analysis?
From the end: start at the conversion and work backwards. Identify the 3 to 5 most frequent paths of converting sessions, then compare them with abandoning sessions' paths. The points where the two populations diverge are your optimisation priorities.