CRO Case Study | Ana Zamfirache
Case Studies  /  E-commerce · CRO · UX

How data-driven CRO lifted
Annie Selke's conversion rate
from 2.5% to 3.4%.

A five-phase Conversion Rate Optimisation initiative combining heuristic evaluation, A/B testing, and personalisation to reduce cart abandonment, improve navigation, and drive measurable revenue growth.

RoleProduct Manager · CRO Lead
ClientAnnie Selke (Home Décor E-commerce)
ScopeResearch · Hypotheses · A/B Testing · Implementation · Monitoring
MethodsHeuristic Eval · Analytics · User Interviews · A/B + MVT
+36%
Conversion rate
improvement (2.5→3.4%)
−21%
Cart abandonment
rate reduction
+12%
Average Order Value
via personalisation
5
Phases: Research →
Monitor & Iterate
01 — The Problem

A high-traffic site
with a leaky funnel.

Annie Selke's e-commerce platform had the traffic — but users were bouncing, abandoning carts, and failing to convert. The homepage attracted visitors who never navigated further. The checkout process had too much friction. Trust signals were absent at the moments that mattered most.

Baseline Metrics — Before
  • Bounce rate on product pages: 42%
  • Average cart abandonment rate: 67%
  • Conversion rate: 2.5%
  • Users landed on homepage but exited without navigating further
  • Checkout required 5 steps — no guest option
My Role

Led the full CRO initiative end-to-end: designed the research methodology, defined hypotheses, ran A/B and multivariate tests, implemented changes cross-functionally, and set up ongoing monitoring. Collaborated with design, engineering, and marketing to align every optimisation with business goals.

Core Hypothesis

The drop-off wasn't a traffic problem or a product problem. It was a trust and friction problem — users didn't feel confident enough to buy, and the path to checkout had too many decision points. Removing friction and adding well-placed trust signals at critical moments would unlock latent conversion.

02 — Five-Phase Process

From diagnosis
to measurable outcome.

A structured five-phase approach — each phase building on the last. No guessing, no random changes. Every intervention was grounded in data and validated through testing before full deployment.

1

Research & Analysis

Heuristic evaluation of existing UX, deep analytics review (bounce rates, abandonment funnels, session recordings), competitor analysis across 3 direct competitors, 15 user interviews + 200-respondent survey.

Research
2

Hypothesis Creation

Translated research findings into 3 testable hypotheses — each mapped to a specific friction point and a measurable KPI target. Every hypothesis had a clear success condition before testing began.

Hypotheses
3

Experimentation — A/B + Multivariate Testing

A/B tested navigation simplification and trust badge placement. Ran multivariate tests on CTA copy and colour combinations. Deployed recommendation engine personalisation for AOV impact.

Testing
4

Optimisation & Implementation

Rolled out winning variants at scale: redesigned navigation, reduced checkout steps, added trust signals, improved product pages, implemented SEO enhancements and content strategy changes.

Implementation
5

Monitoring & Iteration

Set up automated abandoned cart email campaigns, initiated a loyalty program, tracked all KPIs against baseline, and established a continuous improvement cycle for ongoing gains.

Monitoring
03 — Research Findings

What the data
actually revealed.

Three research streams converged on the same conclusion: users wanted to buy, but the experience was getting in the way. The issues weren't aesthetic — they were structural.

Heuristic Evaluation
  • Navigation was cluttered — users couldn't find specific product categories
  • Checkout required too many steps, driving high cart abandonment
  • Trust signals (reviews, certifications) absent from product pages
Competitor Analysis
  • Competitors prominently featured "free returns" and "secure checkout" trust cues
  • Product pages used larger, higher-quality images with zoom functionality
  • Cleaner navigation with fewer top-level categories
User Research — 15 Interviews + 200 Survey Respondents

Users consistently described the navigation as "confusing" and "hard to browse." The strongest unmet desire: personalised recommendations and faster access to bestsellers. Trust was a recurring theme — users mentioned hesitating at checkout because the site "didn't feel secure enough."

Hypothesis 1
Navigation Simplification
  • Reduce to 6 primary categories
  • Expected: decrease bounce rates
  • Validated via A/B test
Hypothesis 2
Trust Signals on Product + Checkout
  • Add trust badges at decision points
  • Expected: +15% completed transactions
  • Validated via A/B test
Hypothesis 3
Personalised Recommendations
  • "Top Picks for You" based on behaviour
  • Expected: +10% Average Order Value
  • Result: +12% AOV achieved
04 — A/B Testing Results

Every change
earned its place.

No assumption was deployed without testing. Each variant was validated against a clear success metric before being rolled out at scale. Here's what the data showed.

Test
Variant A (Control)
Variant B (Winner)
Navigation Simplification
Reduced to 6 top-level categories
Bounce rate
42%
Bounce rate
34% ↓ 8pp
Trust Badges on Product Pages
"Free Shipping Over $50" + security signals
Conversion rate
2.5%
Conversion rate
3.1% ↑ 0.6pp
Multivariate: CTA Colour + Copy
Best performer identified across 4 combinations
Winner
Winner
Green "Add to Cart" + "Secure Checkout Guaranteed"
Personalisation Engine
"Top Picks for You" recommendation module
AOV (baseline)
Baseline
AOV improvement
+12% ↑
05 — Implementation

Five changes.
All measurable.

Winning test variants were deployed as permanent changes across the site. Each change targeted a specific friction point identified in Phase 1 — nothing was cosmetic.

Change 01 Navigation Redesign ↓ Bounce rate 42% → 30%

Reduced cognitive load at the top of the funnel

  • Reduced categories to 6 clear options: Bedding, Rugs, Furniture, Décor, Sale, Inspiration
  • Introduced sticky navigation bar for persistent access as users scroll
  • Removed cluttered sub-menus that were causing decision fatigue
Change 02 Checkout Streamlining ↓ Cart abandonment −21%

Removed the biggest barrier to purchase

  • Enabled guest checkout — no account required to complete a purchase
  • Reduced checkout steps from 5 to 3
  • Added inline trust signals at each checkout step
Change 03 Product Page Overhaul ↑ Conversion 2.5% → 3.4%

Built confidence at the point of decision

  • Added high-quality images with zoom capabilities
  • Highlighted trust signals: customer reviews, secure payment icons, stock availability
  • Integrated personalised "Top Picks for You" recommendation module
Change 04 SEO Enhancements ↑ Organic discoverability

Improved findability and structured data

  • Rewrote product descriptions with targeted keywords
  • Added schema markup for reviews and FAQs to improve rich snippets
Change 05 Post-Purchase & Retention +$10K/mo recovered · +15% repeats

Extended the funnel beyond checkout

  • Automated abandoned cart email sequences — recovering ~$10,000/month in sales
  • Launched loyalty program driving a 15% increase in repeat purchase rate
  • Published content strategy: buying guides and product video tutorials
06 — What This Taught Me

CRO is a systems
problem, not a design problem.

The biggest gains came from removing friction, not from adding features. Every intervention that moved the needle was rooted in understanding why users weren't converting — not in guessing what might look better.

Trust is a conversion lever, not a branding decision. Adding "Free Shipping Over $50" and "Secure Checkout" at the right moments moved conversion more than any design change. Users need reassurance exactly where doubt emerges — not in the header.

Checkout friction is silent revenue loss. Reducing steps from 5 to 3 and enabling guest checkout had immediate, measurable impact on cart abandonment. Every extra step is a decision point where users can talk themselves out of buying.

Personalisation works — but only after you fix the basics. The recommendation engine delivered +12% AOV, but it would have been invisible if the checkout was still broken. Optimisation has an order of operations: remove friction first, add delight second.

The post-purchase funnel is undervalued. Automated abandoned cart emails recovered $10K/month that the business was already earning but losing. The loyalty program turned one-time buyers into repeat customers. The biggest CRO wins aren't always at the top of the funnel.

Hypotheses without success metrics are just opinions. Every test had a pre-defined success condition. This discipline is what separates CRO that compounds over time from A/B testing that goes nowhere.

Summary

From leaky funnel to
measurable revenue engine.

Five structured phases. Fifteen user interviews. Four tested variants. Five implemented changes. One significantly healthier conversion funnel.

+36%
Conversion rate improvement
2.5% → 3.4%
−21%
Cart abandonment
rate reduction
+12%
Average Order Value
via personalisation
$10K
Monthly revenue recovered
via cart email automation

This project demonstrated that conversion rate optimisation is not a one-time sprint — it's a continuous system. The monitoring and iteration phase set up feedback loops that continued delivering gains long after initial implementation. The real value wasn't the 3.4% conversion rate. It was the infrastructure to keep improving it.

Want to optimise your
digital funnel?

Whether it's e-commerce CRO, product adoption, or AI-driven automation — let's find the friction and fix it.

Connect on LinkedIn →
Previous
Previous

Purchasing App for Public Sector

Next
Next

Service Design Project