Case Study · Mixed Methods × Experimentation Strategy
Building the missing half of optimization
How qualitative research, behavioral analytics, and A/B testing were connected into a repeatable optimization system for a high-traffic insurance acquisition funnel.
Back-button clicks increased in the two-step lead form — a behavioral signal that users were confused by the conversion flow.
+6%
Overall B2C lead conversions increased after the one-step form experiment reduced flow friction.
+20%
Desktop conversion lift from the same lead-form optimization, validated at 99.98% confidence.
3.5K
Incremental lead growth over two months, plus additional future opportunities identified through mixed-method research.
The analytics team could see exactly what was happening. They couldn't see why. Back-button clicks were rising, form drop-off was high, and more than tens of thousands of daily visitors were moving through a funnel where small moments of confusion could become meaningful business loss.
01 · The business challengeAnalytics found the leak, but not the cause
The client's optimization team had a mature quantitative practice: web analytics, conversion tracking, and A/B testing. But the team was mostly optimizing from transactional behavior. They knew where the lead funnel was struggling, but lacked the qualitative evidence needed to understand the user experience behind the numbers.
The immediate signal was a two-step lead form. Back-button clicks had increased by more than 100% year over year, suggesting users were uncertain about what would happen next, what information was required, or whether they were ready to request a quote.
The current journey showed multiple places where confidence dropped
Anonymized artifact
The problem was not isolated to the form. The journey revealed confidence drops across messaging, product understanding, estimator comprehension, and contact expectations — all of which shaped whether users felt ready to become leads.
02 · The reframeNot just usability testing — a mixed-method optimization engine
A one-off usability study would have identified issues, but it would not have changed how the organization made optimization decisions. The bigger opportunity was to connect qualitative evidence to the existing quantitative practice so every future experiment had a stronger behavioral rationale.
The work shifted from “run a usability test” to “build a repeatable way to turn customer behavior into better experiment decisions.”
Original ask
Use usability testing to gather qualitative feedback on high-friction pages and conversion paths.
Research reframe
Pair analytics, usability testing, competitive benchmarking, and A/B testing into one optimization loop.
Strategic shift
Research became an input to prioritization and experimentation, not a validation activity after decisions were already made.
03 · Research foundationHow the optimization program was grounded
The process combined self-site usability testing, competitive studies, behavioral analytics, experimentation planning, and a standardized metric framework. The intent was to create a repeatable cadence: diagnose friction, translate findings into testable hypotheses, prioritize by conversion impact, and validate with live experiments.
4
Task-based studies
Self-site and competitive usability studies focused on the pages most important to conversion.
3
Evidence streams
Usability behavior, web analytics, and A/B testing were connected into a single decision process.
1
SUM framework
Effectiveness, efficiency, and satisfaction were standardized into a single usability score.
10+
Optimization ideas
Research findings became a prioritized testing roadmap across form, messaging, calculator, and CTA patterns.
The research plan created a repeatable study-to-test pipeline
Planning artifacts
Study planning
Single usability metric
The method was intentionally operationalized. Each study was designed to produce comparable evidence, prioritized insights, and experiment-ready hypotheses rather than a static research report.
04 · What we learnedFriction was not one problem
The research showed that users were not simply abandoning because the form was too long. Confidence eroded across the experience: confusing messaging, unclear cost information, inconsistent contact expectations, and benefits content that looked useful but required too much interpretation.
Insight 01
Messaging created confusion before the form
Users struggled to interpret campaign headlines and value propositions, which weakened confidence before they reached the conversion path.
Insight 02
The lead form created uncertainty
The two-step flow introduced ambiguity around call timing, contact expectations, and what would happen after submission.
Insight 03
The estimator looked helpful but was hard to trust
Users wanted cost clarity, but struggled to interpret product selections, numbers, and what monthly premium or coverage really meant.
Insight 04
Users wanted more control over contact
Research surfaced a strong preference for alternative contact paths, including live chat, especially when users were not ready for a phone call.
Patterns were translated into prioritized optimization opportunities
Research synthesis
The synthesis connected observation to action. Findings were organized by page, task, user sentiment, open-ended feedback, and conversion relevance so issues could move directly into the testing backlog.
05 · Research → ExperimentsTurning behavioral evidence into a testing roadmap
Each insight was translated into a measurable experiment or future optimization idea. The goal was not to fix every issue at once; it was to prioritize the changes most likely to move conversion while improving the experience.
Research insight
Experiment or product decision
Users were confused by the two-step lead form and call timing expectations.
Test a one-step lead form and consolidate contact preferences into a clearer single-screen flow.
Users found the homepage cost calculator helpful.
Extend the cost-calculator pattern to product pages and benefits-estimator experiences.
Benefits estimator numbers were difficult to interpret.
Reduce carousel complexity and clarify premium, coverage, and payout information.
Users wanted alternatives to phone and form-based lead capture.
Frame live chat as a future feature opportunity and estimate lead recovery potential.
Two examples where research shaped experiment direction
Anonymized concepts
Lead form friction
Benefits estimator confusion
The strongest experiments were not opinion-led. They came from points where behavioral analytics, task performance, and user explanation all pointed to the same breakdown.
06 · ResultsThe one-step form experiment proved the model
The clearest validation came from the lead-form experiment. Research showed that the two-step pattern was creating uncertainty; the experiment tested whether reducing friction and consolidating the form would improve conversion without sacrificing lead quality.
Reduced friction. Increased conversion.
The one-step lead form transformed a research-backed usability finding into measurable business impact. The result gave the team confidence that qualitative evidence could improve experiment selection, not just explain results after the fact.
+6%
Overall B2C lead conversion lift.
+20%
Desktop conversion lift at 99.98% confidence.
+5%
Click-to-call increase at 97% confidence.
3.5K
Incremental lead growth over two months.
Experimentation and analytics closed the loop
Anonymized dashboards
Testing roadmap
Website analytics
The output was a system, not a study. The research created a pipeline from usability evidence to A/B testing ideas, measurable results, and future optimization priorities.
07 · Beyond the first winResearch created future revenue opportunities
The work also identified future optimization opportunities beyond the successful form test. One of the most valuable was live chat: usability testing showed a meaningful share of users preferred chat over phone or form-based contact, especially when they were still evaluating coverage and costs.
Research signal
Users wanted information before committing
Cost, coverage, benefits, and contact expectations needed to be clearer before users felt ready to request a quote.
→
Future opportunity
Improve informational support upstream
Add monthly premium guidance, clarify estimator logic, and support users before the lead form.
Research signal
Phone calls were not always the preferred path
47% of users from testing preferred live chat as a communication method over phone or forms.
→
Business opportunity
Recover leads lost from callback friction
Live chat was estimated to close a portion of leads lost through the “Call Me Later” path, representing a seven-figure annualized opportunity.
08 · What changedOptimization became more evidence-driven
The most durable outcome was not a single lift metric. It was proving that conversion optimization becomes stronger when analytics and experimentation are paired with human behavior evidence from the beginning.
Decision impact
Test ideas became better grounded. Experiments were prioritized based on observed behavior, user explanation, conversion relevance, and business impact.
Product impact
Lead generation became easier to complete. The one-step form reduced friction and improved conversion across key segments.
Business impact
Research identified both immediate and future value. The work produced validated lift, incremental leads, and a quantified future opportunity tied to live chat.
Practice impact
Qual and quant became one system. Usability testing explained the “why,” analytics located the “where,” and experimentation validated the “what changed.”
09 · ReflectionThe real contribution was the bridge
This work reinforced a simple pattern: analytics are powerful at locating behavior, but they rarely explain it. Usability research can explain behavior, but it becomes more influential when connected to measurable business outcomes.
The value of the work was building the bridge between the two — creating a process where research generated better hypotheses, experimentation validated them in production, and the organization had a stronger basis for deciding what to optimize next.
The work was not usability testing, and it was not A/B testing. It was the system that connected them.
