Websites & CRO
Heatmaps & session recordings: finding what to test
Analytics tells you where homeowners drop off your remodeling site. Heatmaps, recordings, and on-page surveys tell you why. Watch what visitors actually do and you stop guessing at test ideas, which is why pre-qualified hypotheses win 25 to 30 percent of the time instead of under 20.
A/B testing answers which version wins. It cannot tell you what to test in the first place. That comes from watching real behavior: click maps that expose dead clicks, scroll maps that find false bottoms, recordings that catch rage clicks, and on-page surveys that ask a homeowner why they left before requesting an estimate. This is qualitative research, and it is the source of strong hypotheses. Teams that pre-qualify test ideas with heatmap and analytics data win 25 to 30 percent of the time; programs running on hunches win fewer than one in five. With 88 percent of users saying they will not return after a bad experience, the friction you cannot see is the friction costing you the most.
Start with click maps and dead clicks
A click map aggregates every tap and click on a page into a heat overlay. The first thing it exposes is the dead click: a tap on an element that looks interactive but does nothing. Bold text, a product image, a static icon, a heading that reads like a button. Each dead click is a small broken promise, and clusters of them mark exactly where your interface misleads people. Microsoft Clarity and FullStory both flag these automatically, so you do not have to eyeball the map.
The fix is usually cheap and the signal is unambiguous. In one Clarity case, recordings showed clicks on a search box being silently dropped, affecting 4 percent of everyone who clicked it. That is a measurable conversion leak hiding behind a feature people assumed worked. Pull your top landing pages, sort by clicks, and make every heavily clicked element either functional or visibly inert. WellBuilt runs this audit first because it converts vague underperformance into a concrete list of things to fix.
Read scroll maps for false bottoms
Scroll maps show how far down the page people travel and where attention dies. Eyetracking from Nielsen Norman Group found about 80 percent of viewing time happens above the fold, so anything below the initial screen is read by a shrinking minority. If your form, price, or primary call to action sits where only a fraction of visitors reach it, the scroll map will show the drop-off as a cold band.
The map also catches false bottoms. A full-width image, a heavy whitespace gap, or a section that visually resembles a footer can convince users the page has ended when more content waits below. Scroll depth flatlines at that line and never recovers. The fix is layout, not copy: break the visual seal, add a peeking next section, or move the decisive element up. Because attention concentrates so sharply at the top, every block you promote past the fold compounds.
Watch recordings for rage clicks and friction
Session recordings replay individual visits so you see hesitation, backtracking, and frustration that no chart captures. The headline signal is the rage click: five or more rapid clicks on the same spot within a fraction of a second, the digital equivalent of jabbing a stuck button. FullStory tracks these alongside dead clicks, error clicks, and thrashed cursors, and the volume spikes under stress; rage clicks rose about 16 percent across the Thanksgiving-to-Cyber-Monday window in their data. They are reliable markers of broken or confusing UI.
You do not need to watch thousands. The 5-user rule from Jakob Nielsen holds that five sessions surface roughly 85 percent of major usability problems, and reviewing about 25 replays a week catches a large share of issues analytics never shows. The discipline is pattern-counting, not anecdote-hunting: if 2 of 100 users do something odd, ignore it; if 30 of 100 do, you have found your next test.
Frustration signals worth filtering recordings by:
- Rage clicks: repeated rapid taps on one element, signaling broken or unresponsive UI
- Dead clicks: taps on non-interactive elements that users expected to do something
- Error clicks: interactions that triggered a JavaScript or form error
- Excessive scrolling: far more vertical movement than the page average, hinting at confusion or poor relevance
- U-turns: visitors who land, scan, and bounce without engaging the primary action
Analytics shows you where visitors fall off. Heatmaps, recordings, and surveys are the only things that tell you why.
Find where forms lose people
Form analytics tracks each field individually: time spent, refills, blank submissions, and the exact field where people quit. This matters because length kills completion. Baymard found the average ecommerce checkout shows 14.88 form fields against an ideal closer to 7, and reducing a lead form from 11 fields to 4 has been shown to lift conversions by 120 percent. Across millions of tracked sessions, Zuko puts average form completion at 51.7 percent, meaning roughly half of starters never finish.
Field-level data tells you which field to cut. A phone number marked required, a "project budget" dropdown that scares off a homeowner not ready to commit, an address field that fights autofill. Each one shows up as a spike in abandonment at a single step. Pair that with the rest of your funnel: Baymard reports 18 percent of US shoppers have abandoned an order specifically because checkout was too long or complicated, and fixing documented usability issues can raise conversion by up to 35 percent. The estimate-request form is often the single highest-leverage place to look.
Ask leaving visitors why
Heatmaps and recordings show what happened; they cannot read intent. On-page surveys close that gap by asking directly. An exit-intent poll fires as the cursor moves to leave: "What stopped you from completing this today?" with options like price, shipping, needed more information, or just browsing. Because the user has already decided to go, the answer is unusually honest, the most candid feedback you will collect. Response rates on website surveys range widely, from roughly 5 percent to nearly 60 percent depending on timing and effort.
Keep it to one question and trigger it where the value is highest: your pricing or cost-guide pages, the estimate-request form, and high-exit service landing pages. The replies turn a vague drop-off into a named objection you can test against. One remodeler ran an exit survey on a kitchen-remodel cost page, learned visitors wanted a ballpark price range before filling out a form, added a clear range, and lifted estimate requests in three months. That is a hypothesis you would never have guessed from a funnel chart alone.
Pair the qualitative with the numbers
Quantitative analytics tells you where conversion breaks; qualitative research tells you why. Neither is enough alone, and the strongest programs triangulate. Your funnel flags a 50 percent drop at the shipping step. A recording shows users hesitating over an unexpected fee. A scroll map confirms the fee sits below where most people look. An exit survey names "surprise shipping cost" outright. Four sources, one conclusion, and now a hypothesis worth the test slot.
This is why research quality predicts test outcomes. Pre-qualifying ideas with behavioral data pushes win rates to 25 to 30 percent, while programs guessing from opinion sit below 20. The workflow WellBuilt runs is consistent: find the leak in analytics, investigate it with heatmaps and recordings, confirm intent with a survey, then write a hypothesis specific enough to test. The qualitative layer is where the ideas come from. Mobile makes it urgent: with mobile converting near 1.82 percent against desktop's 3.90, the small-screen friction these tools expose is where most of the lost revenue hides.
Key takeaways
- Run a click-map audit on top landing pages first; make every heavily clicked element either functional or visibly non-interactive to kill dead clicks.
- Use scroll maps to confirm your form, price, and primary CTA sit above the fold, where 80 percent of viewing time happens, and break any false bottom that stops scrolling early.
- Filter recordings by frustration signals and count patterns, not anecdotes; about 25 replays a week and the 5-user rule surface most major issues without drowning you in footage.
- Track forms field by field, cut the field driving abandonment, and remember dropping a lead form from 11 to 4 fields has lifted conversions 120 percent.
- Add a one-question exit-intent survey on pricing and checkout pages to capture the honest objection, then turn the named reason into a testable hypothesis.
SourcesNielsen Norman Group, Scrolling and Attention eyetracking research, 2010 and 2018 · Nielsen Norman Group, Why You Only Need to Test with 5 Users, 2000 · Baymard Institute, Checkout Usability and cart abandonment benchmarks, 2024 · Zuko Analytics, Form completion and abandonment benchmarks, 2024 · FullStory, Frustration Signals and Cyber Monday rage-click research, 2023-2024 · Microsoft Clarity, rage clicks, dead clicks, and scroll map documentation, 2024 · Convert.com, triangulating qualitative and quantitative data for hypotheses, 2024 · DRIP / industry A/B testing win-rate analysis, 2026 · Survicate, website and exit-intent survey response-rate benchmarks, 2024
Questions, answered straight.
Do heatmaps and recordings replace A/B testing?
No, they feed it. Heatmaps, recordings, and surveys are research tools that tell you what is wrong and generate hypotheses; A/B testing validates whether a fix actually improves conversion. Use the qualitative tools to decide what to test, then test it properly. Hypotheses pre-qualified this way win 25 to 30 percent of the time versus under 20 for guesses.
How many session recordings do I need to watch?
Fewer than you think. Jakob Nielsen's research shows five sessions reveal about 85 percent of major usability problems, and reviewing roughly 25 replays a week catches most issues analytics misses. Focus on quantity of evidence per pattern, not raw volume: if many users hit the same snag, it is real; if one does, ignore it. Filter by frustration signals to skip the uneventful sessions.
Which tool should I use: Hotjar, Microsoft Clarity, or FullStory?
Microsoft Clarity is free with unlimited recordings and heatmaps plus rage- and dead-click detection, which makes it the easiest place to start. Hotjar adds polished on-page surveys and feedback widgets in one suite. FullStory leans enterprise with deep frustration analytics and search. Begin with Clarity to prove value, then add Hotjar's surveys when you need the why behind the behavior.
What is a dead click and why does it matter?
A dead click is a tap on something that looks interactive but does nothing, like a bold heading or a static image users mistake for a button. It signals a misleading interface and a frustrated visitor. Clarity and FullStory flag them automatically. Audit clusters of dead clicks on key pages and either make the element functional or restyle it so it no longer invites a click.
Websites & CRO
Want this run for you, not just read about?
Turn the traffic you already pay for into qualified leads with pages built to convert.