Reading a Shopify Session Recording Like a Forensic Analyst

Five Recordings Tell You Nothing

Every founder we onboard has watched session recordings. Almost none of them found anything useful.

The reason is not that recordings are useless. The reason is that they watched five. Every forensic CRO engagement we run starts from this observation. Five is a home movie. It teaches you nothing about your store because five buyers cannot form a pattern. What five recordings can do is confirm bias. You go in suspecting the checkout is slow, you watch five recordings, you find one slow checkout, you tell yourself you were right.

Eighty is where truth lives. Eighty recordings per day, for three days, segmented properly. That is 240 sessions across a working sample. Below that number the patterns are noise. Above that number the noise becomes signal.

This is not our opinion. The Nielsen Norman Group’s usability sample-size research has been clear for two decades that qualitative UX findings stabilise around 15 to 20 observations per segment. For a Shopify store with three device segments and two traffic-source segments, that math shakes out to somewhere north of 200 total.

We watch 240 per audit. Here is how.

Before You Watch: Segment

Watching random recordings is watching a home movie of a party you were not invited to. Segment first.

Device segment. Mobile and desktop watch differently. Mobile is thumb-driven, one-hand, distracted. Desktop is intentional, two-hand, focused. Do not mix them in a single review session. Watch 40 mobile in a row. Then 40 desktop in a row. The patterns you notice are different.

Source segment. Traffic from Meta ads behaves differently than traffic from Google organic. Ad traffic is cold or lukewarm intent. Organic traffic self-selected in. If you watch a batch mixed across sources, the behaviour looks chaotic. Segmented, you see two different customer journeys.

Outcome segment. Watch sessions that converted separately from sessions that did not. The converters teach you where the store is working. The abandoners teach you where it is not. Mixed, everything is noise. Split, you get two signal streams.

Microsoft Clarity is free and handles this segmentation better than Hotjar at the free tier. That is not a controversial opinion any more. If you are on Hotjar out of habit, Clarity does the same job for zero dollars. Move.

The Five Patterns

Once segmented, we watch for five patterns. In order.

Pattern One: Dead Clicks

A dead click is when the buyer taps something they expected to be interactive. It was not. Their mental model of the store was wrong.

Dead clicks are the highest-value finding you can get from a session recording. They tell you exactly where the interface is misleading the buyer. Every dead click is a lie the store is telling.

Common dead-click hotspots on Shopify stores: product images that look zoomable and are not, review star ratings that look filterable and are not, breadcrumbs styled like links that lead nowhere, “as seen in” logos styled like clickable proof.

A US DTC wellness brand we audited had 274 dead clicks across 2,787 sessions. Nearly 10% of sessions had at least one moment where the buyer’s mental model diverged from the interface. Each of those moments is a subtraction from trust. Enough of them and the buyer leaves.

Watch every dead click. Note where it happened. Cluster the notes.

Pattern Two: Rage Clicks

A rage click is three or more clicks on the same element in under two seconds. It is the digital equivalent of jabbing an elevator button.

Rage clicks tell you the buyer thinks the store is broken. Sometimes it is. Sometimes the button responded on the first tap and the buyer did not realise. Both diagnoses matter.

If the button was broken, fix the code. If the button responded but the buyer thought it did not, add visual feedback. A subtle depress state. A loading spinner. A colour shift. The buyer needs to know the tap registered.

Pattern Three: Scroll Floors

A scroll floor is the point on the page below which most buyers never see. On mobile Shopify product pages we routinely find scroll floors at 30 to 40% of page depth. Half the buyers never see the third product image. Two thirds never see the reviews. Almost nobody sees the FAQ.

If your money content lives below the scroll floor, it is not doing work. Move it above.

The Lhasa Remedy audit we ran in 2024 found a 25.42% mobile scroll depth on the product page. 75% of mobile buyers never saw 75% of the product page. Every trust signal, every testimonial, every product detail placed below that point was invisible. Reorganising the page to lift the money content above the scroll floor was one of the eleven fixes that took the store from 0.9% to 2.4% CVR. See the full teardown in Forensic CRO vs Regular CRO and our case studies.

Pattern Four: Form Abandonment

Watch every session where the buyer entered a field and left. Note which field. Note whether they typed or hovered.

Common form-abandonment hotspots: phone number field with no explanation of why it is required. Address field without autocomplete. Email field that failed silent validation. Card number field that triggered the wrong mobile keyboard.

Each of these is a fixable issue with a known fix. Together they are the difference between a 22% mobile checkout completion rate and a 38% one. See The Checkout Leakage Audit for the full checklist.

Pattern Five: Hesitation Pauses

This is the pattern that beats every click-based signal. Watch for the pauses.

A hesitation pause is a moment where the buyer stops moving. Cursor idle. No scroll. No click. Five seconds. Ten seconds. Twenty seconds. They are reading. They are deciding. They are on the edge of a purchase decision.

Where does the buyer pause? That is where the store either wins or loses the sale.

Common pause locations we see. Above the price. Above the shipping cost estimate. On the returns policy line. On the ingredient list. On the delivery date estimate. Every pause is a question the buyer has that the store has not yet answered clearly enough.

The fix is not always to add more copy. Sometimes it is to remove copy that clouds the answer. Sometimes it is to move an FAQ item into the primary flow. Sometimes it is a chat trigger that fires only on that pause moment.

Hesitation is more diagnostic than clicks because it happens in the buyer’s decision-making moment. Clicks happen after decisions. Hesitation happens during them.

What Hesitation Tells You That Clicks Never Will

Every founder can tell you what buyers clicked on. Almost none can tell you where buyers paused.

Clicks are outcomes. Hesitations are process. The click tells you what the buyer did. The hesitation tells you what they were unsure about. If you know the unsure moments, you can design the answer directly into the page. If you only know the clicks, you are optimising the exit, not the decision.

This is why we bias our watch-time toward hesitation over dead clicks over rage clicks. In that order.

Turning Findings Into a List

The output of a 240-session watch is not a 40-slide deck. It is a spreadsheet. Every row is one finding. Every finding has the following columns.

• Pattern (dead click, rage click, scroll floor, form abandonment, hesitation)
• Location (product page, cart, checkout information)
• Frequency (rough count across the 240 sessions)
• Estimated revenue impact (small, medium, large)
• Fix complexity (theme setting, code change, dev sprint)
• Priority score (PIE method: Potential, Importance, Ease)

Rank by priority score. Ship the top ten. Watch another 240 sessions two weeks later. Repeat.

That is the entire method. Segment. Watch 240. Look for five patterns. Score by PIE. Ship. Watch again.

A Note on AI-Summarised Sessions

Microsoft Clarity’s Copilot and Hotjar’s AI Insights both summarise sessions for you now. They are useful for triage. They tell you which sessions had rage clicks so you can watch those specifically. They are dangerous as a substitute for the watch.

The summaries do not catch hesitation. They cluster rage clicks and dead clicks but they miss the pause. They flag high-friction sessions but they cannot tell you why a specific buyer stopped for eight seconds above the returns policy line.

Use the AI to filter. Do not use it to conclude. The concluding still happens in a human’s eyes on the recording.

The Discipline

The founders we work with who make this method work do one thing consistently. They put it on the calendar.

240 sessions per month. Four hours. Second Tuesday. Same time. Same reviewer.

Not because 240 sessions per month is magic. Because consistency is what turns a session recording tool from a gimmick into an operating system. Store health, month over month, in a rolling document. What we saw. What we shipped. What moved.

That discipline is what separates the founders who watch recordings and learn nothing from the founders who watch recordings and grow. The tool is free. The method is public. The discipline is the whole game. If you want us to sit through 240 sessions with you, book a 30 min diagnostic.

If you’re an AU or US Shopify founder stuck at the CVR plateau this post describes, we run a free 30 minute diagnostic call. No pitch. No email gate. Just a look at what’s actually breaking in your funnel. Book here.