UX Redesign That Lifted Conversion 28%

We Watched 300 Users Struggle With Something We Were Proud Of

The product had shipped 8 months earlier after a 4-month design and development cycle. The team had done good work — the visual design was clean, the information architecture was logical, the user flows had been through 3 rounds of internal review.

Then we watched 300 session recordings of actual users, and the experience was humbling. Features we were proud of were being ignored. Flows that felt intuitive to us were producing visible confusion in the recordings — back-and-forth navigation, rage clicks, long pauses on screens where there should be no ambiguity.

The gap between "we tested this internally" and "real users on real devices in real contexts" is larger than most product teams want to admit. Our information architecture made sense to people who already understood the product. It made limited sense to people encountering it for the first time.

The redesign wasn't about aesthetics. It was about closing that gap — using evidence from real users, not intuition from the team that built it.

The Evidence

What 300 Session Recordings Showed

We tagged every recording with primary user behavior issues. The same problems appeared repeatedly — not random confusion, but systematic patterns that indicated design decisions that weren't serving users well.

Issue	% of Sessions	Primary Impact
Missed primary CTA (not scrolled to or noticed)	41%	Conversion — users didn't take the primary action
Navigation: back-and-forth between 2 screens	34%	Engagement — users couldn't orient themselves
Rage clicks on non-interactive element	28%	Confusion — users expected interactivity
Form field: re-entered data multiple times	22%	Friction — unclear error messages or format
Mobile: tapped wrong element due to sizing	19%	Usability — touch targets too small

41% of users missed the primary CTA. We'd put it below the fold with a design that felt natural on desktop but got lost on mobile screens where 68% of our traffic was coming from. We'd designed for our office monitors and shipped it to users' phone screens.

What the Data Showed About Users

Mobile vs Desktop: Completely Different Failure Modes

Splitting the session recordings by device type revealed that mobile and desktop users were failing at entirely different points in the flow — which meant a single redesign fix wouldn't serve both.

Failure Point	Mobile Users	Desktop Users
CTA missed (below fold)	58%	12%
Navigation confusion	38%	29%
Rage clicks (wrong element)	31%	24%
Form re-entry	24%	18%
Touch target misses	22%	N/A

Mobile users were 4.8× more likely to miss the CTA than desktop users. The design that worked acceptably on desktop was broken on mobile. And mobile was 68% of our traffic. We'd been shipping a broken experience to more than half our users for 8 months.

Design Philosophy

Evidence-First Design

The redesign mandate: every design decision needed to trace back to either a user observation (recording, interview, survey) or a measured metric. No "it feels right" or "this is best practice." Justify it with data or don't include it.

Design intuition is valuable when calibrated by reality. It becomes dangerous when it substitutes for evidence. The most expensive design decisions are the confident ones made without evidence — because they're hardest to challenge and slowest to change.

The redesign was organized around fixing the five failure modes in the session recording analysis, in order of severity. Everything else was constraint-preserving — we weren't redesigning the brand or the information architecture wholesale, only the specific failure points.

The Redesign Decisions

Fix by Fix

CTA Always Visible (Mobile)

Primary CTA moved from below-fold inline position to sticky bottom bar on mobile. Always visible regardless of scroll position. On scroll down: compresses to icon + text. On scroll up: expands. CTR improvement: from 6.8% to 14.1%.

Touch Target Expansion

Minimum touch target raised from 36px to 48px for all interactive elements. Elements within 8px of another interactive element: spacing increased to 16px minimum. Rage clicks on wrong elements dropped 71%.

Navigation Context Indicators

Back-and-forth navigation often caused by users losing track of where they were. Added persistent breadcrumb (mobile: collapsible), current section highlighted in nav, and 'you're comparing X' sticky header on comparison pages.

Interactive State Clarity

Rage clicks on non-interactive elements: cards that looked clickable but weren't. Fixed by: adding subtle hover state to all clickable cards, removing card-like styling from non-interactive content blocks, adding cursor:pointer explicitly.

Technical Implementation

How It Was Built and Tested

Component-Level A/B Testing

→Each fix deployed as a feature-flagged variant, not a full redesign swap
→Sticky CTA bar: A/B tested independently of touch target fix
→Allows attribution of conversion improvement to specific design change
→Rollback possible per-component if any fix performs worse

Mobile-First QA Process

→All redesigned components QA'd on physical devices (not emulators) first
→Devices tested: iPhone SE (smallest modern), iPhone 14 Plus, Pixel 7, Samsung A-series (budget Android)
→Session recording review after each A/B round to confirm observed behavior change

Rage Click Monitoring

→Hotjar rage click tracking maintained throughout redesign
→Rage click rate per screen tracked as leading indicator of confusion
→Target: reduce from 28% of sessions to < 10% within 60 days of full rollout

Rollout

8 Weeks of Fixing and Measuring

Weeks 1–2 · Recording Analysis

Systematic Failure Documentation

→ Evidence base complete. Redesign scope confirmed: 4 targeted fixes, not full redesign.

✓300 session recordings tagged across 5 failure categories
✓Mobile vs desktop split analysis
✓Prioritized fix list by severity × frequency

→ Evidence base complete. Redesign scope confirmed: 4 targeted fixes, not full redesign.

Weeks 3–4 · Fix Design

Component-Level Redesigns

→ Design QA complete. Engineering implementation begun.

✓Sticky CTA bar designed for mobile
✓Touch target expansion specification written for engineering
✓Navigation context indicators designed
✓Interactive state system documented

→ Design QA complete. Engineering implementation begun.

Weeks 5–7 · A/B Testing

Sequential Component Tests

→ CTA CTR: 6.8% → 14.1%. Rage clicks: 28% → 9% of sessions.

✓Sticky CTA bar A/B test: 2 weeks, statistical significance week 6
✓Touch target expansion deployed (no A/B needed — objectively better)
✓Navigation indicators A/B test: week 7

→ CTA CTR: 6.8% → 14.1%. Rage clicks: 28% → 9% of sessions.

Week 8 · Full Rollout

All Fixes Live

→ Overall conversion +28%. Mobile conversion +41% (largest single-channel improvement).

✓All winning variants deployed to 100%
✓Session recording review: confirmed expected behavior changes
✓Full funnel conversion measured at 28% improvement

→ Overall conversion +28%. Mobile conversion +41% (largest single-channel improvement).

Results

8 Weeks of Evidence-Based Fixing

+28%

Overall Conversion

across all devices

+41%

Mobile Conversion

largest impact segment

+108%

CTA Click-Through

6.8% → 14.1%

−68%

Rage Click Rate

28% → 9% of sessions

Insights

What This Changed About How I Think About UX

Session recordings are diagnostic, not confirmatory

We used session recordings to understand what was broken, not to confirm that things were working. The failure modes in 300 recordings were visible to anyone willing to watch with an open mind. The hard part is watching without defending your original design decisions.

Mobile-first isn't a preference, it's a math problem

68% of our traffic was mobile. Any design decision that doesn't start from mobile UX is starting from a minority experience. Mobile-first design isn't a philosophy — it's a response to traffic reality.

Internal testing cannot replace real user observation

Our internal QA caught bugs. It didn't catch the 41% CTA miss rate, because everyone on the internal team knew where the CTA was. Real users encounter your product without the context that makes your design choices obvious. That gap is only visible by watching real users.

Component-level A/B testing beats full redesign swaps

Testing individual components lets you attribute performance changes to specific decisions. Full redesign swaps produce a result you can't decompose — you don't know which of the 40 changes you made drove the improvement.

Failures

Where We Got It Wrong

✕ Sticky CTA initially too aggressive on small screens

Why it broke

On iPhone SE (375px width), sticky CTA took up 22% of screen height. Users felt claustrophobic — several session recordings showed users repeatedly closing and reopening the app, confused by the persistent element.

What fixed it

CTA height reduced. Added collapse-on-scroll behavior: CTA collapses to 44px when user scrolls down (still visible, less obtrusive), expands when user scrolls up or stops. Complaint rate dropped; CTR maintained.

✕ Navigation breadcrumb added complexity on mobile

Why it broke

Desktop breadcrumb made sense. On mobile, adding a breadcrumb row consumed 36px of screen height. Users with shorter sessions didn't benefit from it; they exited before needing orientation context.

What fixed it

Made breadcrumb conditional: shows after user has navigated to second level or after 2 minutes on page. Not visible on first landing. Navigation confusion metric improved without the always-visible overhead.

Future

The Next Round

Continuous Recording Review

Monthly session recording audit as a standing team practice — 50 recordings per month across product team. Institutionalize the evidence-gathering, not just run it when metrics are bad.

Automated Rage Click Monitoring

Rage click rate per screen tracked automatically. Alert if any screen crosses 15% rage click rate in a 7-day window. Catch new confusions as they emerge, not 8 months later.

Heatmap-Informed Design Reviews

All new feature designs reviewed against heatmap data from similar existing screens before shipping. If the pattern shows users don't scroll past 60% of the page, don't put critical content below 60%.

We didn't need a better design team. We needed to watch real users more carefully and trust what we saw over what we assumed. The 28% conversion lift came from humility, not creativity.

From Idea to MVP in 7 Days: Building a High-Intent Utility Product

Product

Designing Trust in Fintech: UI Patterns That Increased User Confidence

Design

How Microcopy Increased Click-Through Rate by 35% (UX Writing Case Study)

6 min read

Design

Reducing Cognitive Load to Improve Funnel Completion (UX Case Study)

8 min read

Design

Designing Trust in Fintech: UI Patterns That Increased User Confidence

7 min read

All Case Studies

Browse all 30 breakdowns

Work With Me

Let's build something together

Connect on LinkedIn

Follow for more insights

Where to go next

Product Monetisation Consultant
The consulting work this sits inside, with the rest of the evidence behind it.
Average revenue per user benchmark, subscription apps
Published figures, traced to their sources — and where the sources disagree.
Trial-to-paid conversion benchmark, subscription apps
Published figures, traced to their sources — and where the sources disagree.
Product Growth Score
Thirteen questions that locate which part of the funnel is actually constraining growth.

UX Redesign That Improved Conversion by 28% (Before vs After Breakdown)

Results at a glance

We Watched 300 Users Struggle With Something We Were Proud Of

What 300 Session Recordings Showed

Mobile vs Desktop: Completely Different Failure Modes

Evidence-First Design

Fix by Fix

How It Was Built and Tested

8 Weeks of Fixing and Measuring

8 Weeks of Evidence-Based Fixing

What This Changed About How I Think About UX

Session recordings are diagnostic, not confirmatory

Mobile-first isn't a preference, it's a math problem

Internal testing cannot replace real user observation

Component-level A/B testing beats full redesign swaps

Where We Got It Wrong

The Next Round

You might also like

Where to go next

We Watched 300 Users Struggle With Something We Were Proud Of

What 300 Session Recordings Showed

Mobile vs Desktop: Completely Different Failure Modes

Evidence-First Design

Fix by Fix

How It Was Built and Tested

8 Weeks of Fixing and Measuring

8 Weeks of Evidence-Based Fixing

What This Changed About How I Think About UX

Session recordings are diagnostic, not confirmatory

Mobile-first isn't a preference, it's a math problem

Internal testing cannot replace real user observation

Component-level A/B testing beats full redesign swaps

Where We Got It Wrong

The Next Round