Reading a Heatmap: Signal or Noise?
Objective: Given a synthetic set of heatmap and scroll-map findings from a pricing page, apply the lesson's pattern-over-single-session rule to decide which findings are real problems worth fixing and which are noise or a single user's quirk.
You're auditing Casper's pricing page heatmap data after a CRO teammate flags 'something's wrong with this page' but hasn't isolated what.
Separate the findings that show a repeated pattern across many sessions from the ones that are a single visitor's quirk, then rank the real problems by fix priority.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free tier includes heatmaps on a limited number of daily sessions, enough for a single-page audit
Keeps the audit trail simple and shareable with a team
The process
Specimens to review
For each of the 6 findings, decide: real problem worth fixing, or noise/single-session quirk? Rank the real problems by fix priority.
Casper pricing page heatmap + scroll map findings (synthetic, 2-week sample, 3,400 sessions): 1. Scroll map: only 42% of visitors scroll past the mattress comparison table before leaving. 2. Click map: a red click cluster sits on the word 'financing' in a paragraph, it is not a link (appears in 61% of recorded sessions that reach that section). 3. Session recording #1,204: one visitor rage-clicked the 'Compare' tab 6 times, then left. 4. Click map: 38% of sessions click the FAQ accordion for 'Is there a trial period?' before any other element on the page. 5. Scroll map: 91% of visitors reach the main CTA button, it sits above the fold. 6. Session recording #2,890: one visitor's cursor moved erratically across the page for 45 seconds with no clicks (their internet connection likely lagged).
Specimen: synthetic, realistic
Final deliverable
A ranked list of the real, pattern-backed problems on the page, with the single-session quirks explicitly excluded and why.
See a reference example
Intertop pricing-page heatmap audit Real problems (ranked by priority): 1. Non-clickable price-match badge, 58% click rate, add a real link 2. Shipping-cost FAQ opened first by 41% of sessions, move above fold Excluded as noise: - 1 session with a stuck cursor (likely a device issue, not a pattern) - 1 rage-click on the size guide (n=1, watch more sessions before acting)
Success criteria
You're done when you can:
- Correctly separates the 3 pattern-backed findings from the 3 single-session/healthy-metric distractors
- Ranks the real problems using session-count as the priority signal, not their order in the list
- States a specific reason a distractor was excluded (n=1, healthy number, likely connectivity) rather than just labeling it noise