Diagnosing Where Visitors Actually Drop Off

A redesign based on instinct is a guess wearing the clothes of a decision. Real diagnosis answers two separate questions before any change gets proposed: which stage of the path to conversion is actually losing the most people, and what's happening at that stage that's causing it. Three tools do most of the diagnostic work - funnel analysis, heatmaps, and session recordings - and each answers a different part of that picture, which is why skipping any one of them tends to produce a fix aimed at the wrong problem.

Funnel Analysis Answers "Where," Before Anything Else

Before diagnosing why a specific page underperforms, funnel analysis identifies which stage actually deserves that attention. Mapping the full path to conversion - landing page to product page to cart to checkout to confirmation, for instance - and measuring the drop-off rate at each individual step shows where the largest share of visitors is actually being lost, relative to the other stages. This matters because intuition about where the problem sits is frequently wrong: a page that looks unpolished might have a perfectly healthy conversion rate, while a clean, professional-looking step elsewhere in the funnel is quietly losing half its visitors.

Distinguishing a Leak From a Bottleneck Changes the Priority

Not every rough spot in a funnel deserves equal urgency. A funnel leak versus a funnel bottleneck is the useful distinction here: a leak is a stage where an unusually high share of visitors abandon outright compared to other stages - the clearest priority, since it's actively costing conversions. A bottleneck is a stage that's slow, effortful, or frustrating but doesn't necessarily lose people - a required form field that's tedious to fill out but still gets completed by most visitors. Both are worth addressing eventually, but a leak generally deserves attention first, since a bottleneck that doesn't lose people is a worse experience, not necessarily a worse conversion outcome.

Heatmaps Show Patterns Across Many Visitors at Once

Once a specific stage or page is identified as the priority, heatmaps aggregate behaviour - where clicks land, how far people scroll, where attention concentrates - across many visitors into one visual summary. This is useful for spotting patterns that wouldn't be obvious from a conversion number alone: a call-to-action getting far fewer clicks than its visual prominence would suggest, or a scroll-depth heatmap showing most visitors never reach content placed below a certain point on the page. Heatmaps are strongest at revealing what's happening in aggregate, across a large sample, rather than explaining any one visitor's specific experience.

Session Recordings Show the Specific Moment It Goes Wrong

Where a heatmap shows a pattern, a session recording shows the individual story behind it - one visitor's actual path through a page, including the exact moment they hesitate, scroll back and forth in confusion, or abandon a form partway through. This is where the "why" behind a heatmap pattern often becomes clear: a heatmap might show low engagement with a form field, and a handful of session recordings might reveal visitors repeatedly trying and failing to select a date because a date picker is broken on a specific browser. For a practical walkthrough of running this kind of diagnosis, see how to use Hotjar to diagnose where visitors drop off, and for a look at the platforms built specifically for this work, see our comparison of Hotjar versus Microsoft Clarity or the broader comparison of the top 3 session replay and heatmap tools.

Diagnosis Turns Into a Specific, Testable Hypothesis

The output of this diagnostic work isn't a list of vague observations - it's a specific hypothesis that connects an identified problem to a proposed fix: "visitors abandon the checkout step at a particular rate because the shipping cost isn't shown until the final page, so surfacing it earlier should reduce that specific drop-off." A hypothesis this specific is testable in a way that "the checkout feels clunky" never is, and it's exactly what the next lesson in this playbook builds on.

What's Next in This Playbook

With a clear picture of where and why visitors are actually dropping off, the next lesson covers how to turn that diagnosis into a properly structured, testable hypothesis - the step that separates a real A/B test from a guess with a p-value attached to it.