Jennie D. LambertGrowth Design · Product Design · Growth Hacking · CRO · UX · AI AgentsOpen to Roles
I design, test and scale what works, and build the AI agents that help.
CRO, UX and product design, A/B testing and AI agents, for healthcare, fintech and ecommerce.
Growth designer, product designer and CRO strategist with experience in science, healthcare, fintech and ecommerce. I find where people drop off and design the fix, then test and scale what wins. I design and build AI agents, apps and workflows, with guardrails.
Every story is told the same way: the problem, what we tried, the result, and why it matters.
👇 Pick a type of work
Showing all 6 case studies
Case 01 · HEALTHCARE · Paragard (CooperSurgical)
From launch to a testing roadmap for a hormone-free birth control brand
The problem
Paragard wanted more visitors to take high-value actions: talk to a nurse, sign up for emails, look up insurance costs or book a doctor visit.
The fix
We launched a flow where visitors choose how they want help: a nurse navigator, a short “Is Paragard right for me?” quiz, or other options.
The result
4×more engagements, plus a MedTech Breakthrough Award
The full story
We launched a flow where visitors choose how they want help: a nurse navigator, a short “Is Paragard right for me?” quiz, or other options. Heatmaps (pictures of where people click) then fed a ranked list of seven test ideas.
BeforeAfter
Before: the quiz button was the most-clicked item on the homepage (1,217 clicks)After: choose a path, choose how to talk to a nurse, confirm
Result details
4× more engagements and 2× more high-value actions (2,917 tracked engagements). 76% of people who started the quiz finished it.
Why it matters
The program was named Best Patient Education Solution at the 2023 MedTech Breakthrough Awards.
One button change, almost twice as many parents taking the first step
The problem
EndeavorRx is the first doctor-prescribed video game treatment for kids with ADHD. Its main button said “Get a Prescription.” For a parent still learning about a new kind of treatment, that was too big a first step, so few started.
The fix
We ran an A/B test (half of visitors see the old page, half see the new one) that changed only the button words to “Get Started.” Same page, same spot, same color, July 21 to August 11, 2022.
The result
+85.7%more parents started the telehealth flow
Telehealth flow start rate
Get a Prescription8.04%
Get Started14.92%
The full story
We ran an A/B test (half of visitors see the old page, half see the new one) that changed only the button words to “Get Started.” Same page, same spot, same color, July 21 to August 11, 2022. Then I redesigned the next screens so parents could talk to a doctor or a nurse.
BeforeAfter
Before: “Get a Prescription” — 8.04% started (1,045 of 13,004 sessions)After: “Get Started” — 14.92% started (1,923 of 12,887 sessions)
Result details
+85.7% more parents started the flow, with 99% confidence (very unlikely to be luck) across 25,891 sessions.
Why it matters
In healthcare, a button sets expectations. Asking for a smaller first commitment nearly doubled the start rate.
Turning a 3× pandemic traffic surge into revenue while products sold out
The problem
During the pandemic, traffic to Wahoo’s indoor training gear jumped about 3×. Products sold out minutes after restocking, many visitors left the homepage right away, and mobile shoppers were abandoning carts.
The fix
Four quick tests, each building on the last: put “Shop” first in the menu, add homepage recommendations pointing to in-stock products, remove distractions from mobile checkout, and test a new homepage in the US, EU and UK.
The result
+19%revenue from pointing shoppers to in-stock products
Revenue per user, fly-out cart test
No panel$9.17
Popular products$25.57
“By implementing CRO, personalization, and design enhancements, we achieved astounding results in just one month.”Mackenzie Housman, Director of Ecommerce, Wahoo Fitness
The full story
Four quick tests, each building on the last: put “Shop” first in the menu, add homepage recommendations pointing to in-stock products, remove distractions from mobile checkout, and test a new homepage in the US, EU and UK. Later, a fly-out panel of popular add-ons next to the mini cart.
BeforeAfter
Before: “Shop” moved from last to first (+3% revenue, +6% purchases)After: Homepage recommendations to in-stock products (+19% revenue, +6% purchases)
Result details
+19% revenue from recommendations, +15% mobile purchases from a simpler checkout, and the fly-out cart lifted purchases +146% and revenue per user +179%.
Why it matters
When demand outruns supply, the job is routing people to what you can actually sell. The client called the results “astounding” after one month.
Nine tests on a mortgage funnel: simpler first steps kept winning
The problem
Visitors arrived from TV, radio and search but weren’t starting or finishing “Goldie,” the online mortgage application.
The fix
Nine tests over a year.
The result
+33.8%more completed mortgage applications
Application completion rate
Old flow5.96%
Goldie 2.07.97%
The full story
Nine tests over a year. On the homepage, a lead form was replaced with two clear buttons: Refinance or Purchase. Then “Goldie 2.0” asked one simple question per screen, with a progress bar and a slider instead of typed numbers.
BeforeAfter
Before: a lead form in the homepage hero (2.99% completed)After: two buttons, Refinance or Purchase (7.24% completed)
Result details
Goldie 2.0 lifted completed applications +33.8% (5.96% to 7.97%), with 100% probability of beating the old flow, across 26,465 visitors.
Why it matters
Every extra field is a reason to quit. Losing tests mattered too: a celebrity video everyone liked lost, so it wasn’t shipped.
Fewer words and live social proof for a farm and ranch retailer
The problem
One website served very different shoppers: new and returning, pickup and ship-to-home. One experience for everyone was leaving money on the table.
The fix
Personalized tests (different versions for different shopper types).
The result
+177%revenue from returning shoppers after shortening one message
The full story
Personalized tests (different versions for different shopper types). For returning shoppers, the long free-shipping message was cut to a few words. Others added a “going fast” badge on popular items and an exit pop-up for new visitors.
BeforeAfter
Before: Mini cart opening for returning visitors (test CRO-168)After: Dashboard of extra revenue by experience, August 2023
Result details
+177% revenue from the shorter shipping copy (260,693 users). The “going fast” badge lifted purchases +108%; the exit pop-up lifted revenue +41%.
Why it matters
Returning shoppers already knew the offer, so less copy got them to the product. And a back-to-top button everyone liked cost 3.4% revenue across nearly a million users: a losing test is still a result.
Finding the leaks in a luxury boot store before testing
The problem
Cuadra sells handcrafted boots, many $300 to $500. In 30 days, 287,985 sessions produced only 1,375 orders, a 0.48% conversion rate (the share of visits that end in a purchase).
The fix
Research first: scrollmaps (how far people scroll), session recordings and funnel numbers.
The result
45%of homepage visitors ever reached the first products
Homepage visitors
Landed on homepage100%
Reached first products45%
The full story
Research first: scrollmaps (how far people scroll), session recordings and funnel numbers. Then four test designs, like a sticky mobile header with open search and a free-shipping note in the mini cart.
BeforeAfter
Before: the mobile header scrolls awayAfter: a sticky header with open search
Result details
Only 45% of visitors scrolled to the first row of boots, and a “We Do Not Ship to PO Boxes” pop-up covered the cart as soon as it opened. Each finding became a test.
Why it matters
With few buyers and high prices, picking the right tests matters more than running many. Research made the list short.
Since 2024 I lead AI ideation with clients and build AI-powered tools with Claude Code, Codex and Grok. For my own consultancy, I designed a team of AI agents with clear roles and approval guardrails. It roughly triples my output, and I review everything before it ships. These are tools I built for my own family and business.
AI 01 · AI PROJECT · Bella Rocks: Homework HQ
A rock-band homework app for kids
The problem
Homework is easier in small pieces. I wanted a fun app that breaks tasks into tiny first steps and short work sessions, and gives Mom a simple way to cheer Bella on.
The fix
I built a mobile-first app with AI coding.
The result
43geometry practice problems across 9 topics, with hints
The full story
I built a mobile-first app with AI coding. Bella's side has a homework “Setlist” that suggests tiny steps automatically, a 15 or 25-minute focus timer with a break, geometry practice with 3 hints per problem and 21 diagrams that works offline, a reading planner that splits pages into nightly chunks, and a band pet that grows as she earns points. Mom's side has a dashboard, one-tap cheers, shared chat and rewards she sets.
Focus sprint: one tiny step, a 15-minute timer, and points for effortGeometry practice with hints revealed one at a time
Result details
It went live on Oct 8, 2026, and grew through six versions with energy check-ins, surprise reward cards, a pet with 11 unlockable outfits, a tour planner, and Euclid's Elements, Book I.
Why it matters
It's the same approach as my CRO work: make the first step smaller, give quick feedback, and celebrate progress.
AI 02 · AI PROJECT · Agent Model Tracker
A daily dashboard comparing AI coding agents
The problem
New AI coding tools ship updates almost daily. It's hard to tell which one is strongest at what, and where the claims come from.
The fix
I built a dashboard that compares OpenAI Codex, Claude Code and Grok across 13 coding categories.
The result
13categories scored for three AI coding tools, updated daily
The full story
I built a dashboard that compares OpenAI Codex, Claude Code and Grok across 13 coding categories. It has score gauges, a radar chart, a score-over-time chart, a pricing chart, a news feed with filters, and an Experts tab that sums up what 25 verified commentators agree on. Every score keeps its reason and source.
Agent Model Tracker dashboard: a countdown to the next daily update and the latest news
Result details
An automated update runs every day at 7:16 AM PT, logging every score change with a reason and source. Charts are hand-built inline graphics with no chart library, so the page also works offline.
Why it matters
Picking the right AI tool is a decision like any other: show the evidence and how it changes over time.
AI 03 · AI PROJECT · Blueberry Brook Yorkies
Raising Yorkies for therapy and service work, with an app to track every puppy
The problem
At Blueberry Brook Yorkies, we breed Yorkies to become therapy and service dogs. That starts with careful early care: feedings, weights, deworming, vaccines and vet visits for every puppy, tracked across two phones.
The fix
I built an installable app for two people with puppy profiles, one-tap logs (feeding, weight, meds, deworming, vaccines, vet, potty, notes), a care schedule built from each puppy's birth date, and reminders on both phones.
The result
8one-tap puppy care logs, from feeding to vet visits
The full story
I built an installable app for two people with puppy profiles, one-tap logs (feeding, weight, meds, deworming, vaccines, vet, potty, notes), a care schedule built from each puppy's birth date, and reminders on both phones. Every change can be undone, and nothing is ever hard-deleted. A breeding and heat tracker and a pedigree check for pairings came later, plus a growth tracker that charts weekly weigh-ins for each puppy. With AI, I researched how to predict adult size; the research recommends showing a range, not a single number.
Puppy profile with quick-log buttons (sample puppy)Weight chart and care schedule (sample puppy)
Result details
It works offline with the last saved data, syncs between both phones, and never fills in a medicine dose: those items say “Dose: per your vet.”
Why it matters
Healthy, well-tracked puppies are the start of a good therapy or service dog. Good UX helps: fewer taps, clear reminders, and an undo button instead of “Are you sure?” pop-ups.
Proof, Close to Home
This Portfolio Was Built With AI Coding
I designed and built this site with an AI coding assistant. Every update runs automated checks before it ships: WCAG 2.1 AA accessibility at phone and desktop sizes, mobile layout, and crawler blocking. It is the same approach I bring to client work: fast iteration, with guardrails.
Before focusing on CRO, I designed brands and websites at Web Lamb, LLC, my San Francisco studio (2012 – 2017). These projects are where I learned to build brands and sites that have to convert.
Startup launch · 2016SnapRoute
WEB 01 · BRAND & WEB · SnapRoute
A brand that snaps together
The problem
A networking startup needed a brand and a website ready for its product announcement. The site had to explain a technical product quickly and capture leads from day one.
The fix
A logo built from modular pieces that snap together, echoing how SnapRoute’s networking components can be customized and swapped.
The result
2016brand and launch site, ready for the product announcement
The full story
A logo built from modular pieces that snap together, echoing how SnapRoute’s networking components can be customized and swapped. Custom lowercase letterforms, a set of friendly illustrations and icons, a custom video player to showcase the codebase, and a launch landing page designed for clear messaging and lead capture.
Result details
A modular brand that mirrors the product, and a launch page built for clear messaging and lead capture.
Why it matters
Friendly illustrations make an open-source networking stack feel simple, and a clear page turns announcement traffic into leads.
AssurX makes quality management system (QMS) software for regulated industries and needed a brand that set it apart from close competitors.
The fix
A discovery phase to find a position AssurX could own, then a new logo and color palette with a color for each of five industries, a responsive website organized by industry, landing pages connected to Pardot for campaigns and demo requests, SEO, copywriting and redesigned infographics.
The result
5industries, each with its own brand color
The full story
A discovery phase to find a position AssurX could own, then a new logo and color palette with a color for each of five industries, a responsive website organized by industry, landing pages connected to Pardot for campaigns and demo requests, SEO, copywriting and redesigned infographics.
Result details
Every buyer lands on content made for their industry, and the marketing team got a brand asset kit to stay on-brand.
Why it matters
In a crowded category, a clear position and content for each buyer is what sets a brand apart.
Mediasmith, a global media advertising agency, needed a refreshed brand and a website its own team could update with news, jobs and speaking engagements.
The fix
A rebrand with a logo revision, a new tagline, messaging and copywriting, and a 2017 WordPress site the team updates themselves.
The result
5years of partnership: two website redesigns and a rebrand
The full story
A rebrand with a logo revision, a new tagline, messaging and copywriting, and a 2017 WordPress site the team updates themselves. SEO, a PowerPoint template and print collateral keep the brand consistent everywhere else.
Result details
Web Lamb, LLC has been Mediasmith’s agency since 2012, delivering two website redesigns and one rebrand.
Why it matters
A site the team can run themselves keeps the brand current between redesigns.
Example: on Cuadra, only 45% of visitors reached the first products.
Example: on EndeavorRx, “Get a Prescription” became “Get Started.”
Example: the EndeavorRx test ran July 21 to August 11, 2022.
Example: on American Financing, mortgage applications rose 33.8%.
In-depth case studies
The full story behind each result, with test reports and screenshots. Each one opens with “Read the in-depth case study.”
Case 01 · EndeavorRx · One button
+85.7%
More parents starting the telehealth flow."Get a Prescription" → "Get Started" · 99% confidence · 25,891 sessions
Case 01 · EndeavorRx: The Main CTA Test
A/B Test · Jul 21 – Aug 11, 2022
ClientAkili · EndeavorRx
RoleGrowth Strategist & Sr. UX Designer, Ostro Health
KPITelehealth flow starts
Sample25,891 sessions
Result+85.7%
ProblemThe main button asked parents to “Get a Prescription” before they were ready, so few started the telehealth flow.
SolutionI A/B tested a softer “Get Started” button, then redesigned the next steps so parents could talk to a doctor or a nurse.
The First Ask Was Too Big
EndeavorRx is the first doctor-prescribed video game treatment for kids with ADHD. Its main button asked parents to "Get a Prescription" and sent them to a telehealth partner, Ostro (formerly Rx Define), to book a doctor. For a parent still learning about a new kind of treatment for their child, that is a big first step.
Hypothesis
If we change the CTA from "Get a Prescription" to "Get Started", more parents will start the flow, because the first step will match how ready they are.
What We Tested
One change only: the button copy on the homepage hero. Same page, same position, same color. The test ran July 21 to August 11, 2022, with traffic split evenly.
A · Original
The only doctor-prescribed video game treatment for kids with ADHD
Get a Prescription
8.04%1,045 of 13,004 sessionsB · Variant · Winner
The only doctor-prescribed video game treatment for kids with ADHD
Get Started
14.92%1,923 of 12,887 sessions
See the Test Report Screenshots
A · Original"Get a Prescription": 1,045 of 13,004 sessions (8.04%)B · Variant (Winner)"Get Started": 1,923 of 12,887 sessions (14.92%)
A · Get a Prescription8.04%1,045 / 13,004
B · Get Started14.92%1,923 / 12,887
Conversion = sessions that started the Ostro flow. Predicted lift range +53% to +113%.
With more parents starting, the next screens had to match the softer promise. In Figma I designed the mobile flow to follow "Get Started": instead of choosing a doctor appointment type, parents can talk to a doctor or to a nurse to learn more.
Before
Doctor-first: choose an appointment typeAfter
"Get Started" → Doctor or nurse
Nurse Chat That Doesn't Lose People
For parents who choose the nurse, the chat collects an email or phone number while it connects them, and after hours it turns into a callback form. Interested parents become leads instead of leaving.
Online: capture contact while connectingOffline: after-hours visits become leads
What I Learned
In healthcare, the words on a button set expectations about what happens next. A smaller first commitment nearly doubled conversion. The test only measured clicks, though, so we flagged that bounce rates at the telehealth partner should be watched after rollout. Knowing what a test does not tell you is part of the result.
Revenue from pointing shoppers to what was in stock.Homepage recommendations · plus +15% mobile purchases from a simpler checkout
Case 02 · Wahoo Fitness: CRO During a Sell-Out
At Publicis Sapient
ClientWahoo Fitness
RoleSr. Ecommerce CRO Strategist, Publicis Sapient
MarketsUS, EU, UK
Tests4 shipped
Best Result+19% revenue
ProblemPandemic traffic tripled while products sold out minutes after restocking, and mobile shoppers were abandoning carts.
SolutionFast tests on navigation, in-stock recommendations, a simpler mobile checkout and a new homepage, then a fly-out cart with upsells.
The Problem: More Traffic Than Stock
During the pandemic, traffic to Wahoo's indoor training gear jumped about 3x. Products sold out minutes after restocking, bounce rates on the homepage were high, and mobile shoppers were abandoning carts. The goal was to turn that traffic into revenue anyway.
Four Tests, Each Building on the Last
Navigation"Shop" moved from last to first+3% revenue · +6% purchasesRecommendationsHomepage recs pointing to in-stock products+19% revenue · +6% purchasesCheckoutDistractions removed from mobile checkout+15% mobile purchasesHomepageNew design tested 50/50 in 3 markets+6% revenue · +4% purchases
"By implementing CRO, personalization, and design enhancements, we achieved astounding results in just one month."
Mackenzie Housman, Director of Ecommerce, Wahoo Fitness
Fly-Out Cart: Upsells at the Moment of Add to Cart
If we open a recommendations panel beside the mini cart when someone adds to cart, more shoppers will buy, because the add-on they need shows up while they are already deciding.
I tested two recommendation strategies, Popular and Similar, against no panel. Desktop, all audiences, July 20 to 28, 2023.
Control · No Panel$9.177,285 users
Similar Products$9.777,063 users
Popular Products$25.576,847 users
Revenue per user · Popular: +178.9% revenue, +145.8% purchases, >99% probability to be best · 21,195 users
+146%Purchases
+179%Revenue per user
>99%Probability to be best
See the Test Report Screenshots
The fly-out panel opens to the left of the mini cartDynamic Yield report, USA-192
The Next Phase: Goals and Segments
In a later review I reset the program around three goals: grow Wahoo X subscriptions from 54K to at least 65K, raise product-page add-to-cart from 2% to 2.4%, and cut cart abandonment from 46% to 37%. The roadmap moved to audience segments. An early read on a cart-recommendations test showed "Popular" recommendations leading for the Runner segment, so I recommended waiting until the two-week minimum before calling it.
What I Learned
When demand outruns supply, CRO is about routing people to what you can sell. Small, fast tests on navigation and recommendations paid off within a month. The fly-out cart showed something else: which algorithm you pick matters as much as where the panel goes. Similar products barely moved revenue, while Popular nearly tripled it.
More completed mortgage applications.Goldie 2.0 · 100% probability · 26,465 visitors
Case 03 · American Financing: Testing a Mortgage Funnel
May 2021 – Aug 2022
ClientAmerican Financing
RoleSr. Conversion Optimization Analyst
KPIApplication ("Goldie") completions
ToolsA/B testing platform, Optimizely
Top Result+33.8% · 26K visitors
ProblemVisitors from TV, radio and search weren’t starting or finishing “Goldie,” the online mortgage application.
SolutionNine tests that made the first steps simpler: two clear buttons instead of a form, and one question per screen in Goldie 2.0.
The Funnel
American Financing is a mortgage lender. Visitors arrive from TV, radio and paid search, then start "Goldie," a guided online application. Every test aimed at one thing: more people starting and finishing Goldie.
Homepage Test: A Form vs. Two Clear Choices
If we replace the hero lead form with two buttons, Refinance and Purchase, more visitors will start the application, because choosing a goal is an easier first step than handing over contact details.
Before ControlLead form in the heroAfter WinnerTwo buttons: Refinance or Purchase
A · Lead Form2.99%6 / 222
B · Two Buttons7.24%15 / 215
Application completions, desktop, Aug 2021. 97.99% probability to beat control (the chance the new version really is better than the original), but only 437 visitors, so it was a strong signal rather than a final answer.
The Biggest Win: Goldie 2.0
The redesigned application flow, "Goldie" 2.0, asks one simple question per screen with a progress bar and a slider instead of typed numbers. Tested on all devices over three weeks, it beat the old flow decisively.
Goldie 2.0: one question per step, slider input, progress bar
Control5.96%225 / 3,787
Goldie 2.07.97%1,807 / 22,678
Application completions · Feb 3–24, 2022 · +33.8% · 100% probability to beat control (the chance the new version really is better than the original) · 26,465 visitors
Landing page: one choice, Refinance or PurchaseMobile version of the same pattern
What I Learned
Simpler first steps kept winning: one question per screen beat a long form, two buttons beat a hero form, and a video lost to a plain page. The losing and flat tests mattered too. They stopped the team from shipping a video everyone liked, and saved weeks on a layout change that made no difference.
More engagements, and a MedTech Breakthrough Award.2× high-value actions · 2023
Case 04 · Paragard: A Growth Strategy From Launch to Experiments
Oct 2022 – Feb 2023
ClientCooperSurgical · Paragard
RoleGrowth Strategist & Sr. UX Designer, Ostro Health
KPIsNurse chats, CRM sign-ups, doctor visits
Timeline5 months
Result4x engagements
ProblemThe brand needed more visitors to take high-value actions, like talking to a nurse or booking a doctor visit.
SolutionAn engagement flow with nurse navigators and a “right for me?” quiz, then a ranked seven-test roadmap built from heatmap and funnel data.
The Goal: More People Taking the Next Step
Paragard is a hormone-free birth control. The brand wanted more visitors to take high-value actions (HVAs): talk to a nurse navigator, sign up for emails, look up insurance costs and schedule a doctor visit. I agreed these KPIs with the client first, so every design and test had a clear target.
The Plan
Oct 2022Launch the FlowEngagement flow and optimized pages go live.
Nov 2022AnalyzeQuantitative and qualitative data on user behavior.
Dec 2022RoadmapTurn the findings into a ranked list of test ideas.
Feb 2023ExperimentStart A/B testing for ongoing growth.
The Engagement Flow
Visitors choose how they want help: talk to a nurse navigator, take a short "Is Paragard right for me?" quiz, or browse other options. Nurses can be reached by chat, phone or a requested callback.
Choose a path → Choose how to talk to a nurse → Confirmation
What the Data Showed
Heatmaps from Oct 3, 2022 to Jan 3, 2023 showed "See if it could be right for you" was the most-clicked element on the homepage, with 1,217 clicks. 76% of people who started the quiz finished it, and the biggest drop-off was at step 2, where they were asked for their name.
Homepage heatmap: the quiz CTA got the most clicksQuiz funnel: 76% complete, biggest drop at step 2
The Testing Roadmap
Seven ideas, each with a hypothesis. A few of them:
Idea
Hypothesis
Flow entry on the top-clicked homepage CTA
Putting entry points where people already click will bring more of them into the flow and to HVAs.
Entry point on "What to Expect"
An entry point on a high-traffic page will raise engagement and conversion to HVAs.
Above-the-fold CTA copy
A CTA that meets people where they are in the buying cycle will raise click-through.
QR codes and SMS short codes
More ways to engage from TV and media ads will raise engagement from patients and doctors.
4xMore engagements
2xMore high-value actions
2,917Tracked engagements
The program was named Best Patient Education Solution at the 2023 MedTech Breakthrough Awards.
Revenue from returning shoppers after shortening one message.Free shipping copy · conclusive in Dynamic Yield · 260,693 users
Case 05 · CAL Ranch: Personalization by Shopper Type
Dynamic Yield · 2021 – 2023
ClientCAL Ranch Stores
RoleSr. Ecommerce CRO Strategist, Publicis Sapient
ToolDynamic Yield
AudiencesNew, returning, pickup states
Best Result+177% revenue
ProblemOne experience served very different shoppers: new and returning, pickup and ship-to-home.
SolutionPersonalized tests in Dynamic Yield by shopper type, from shorter shipping copy to live social proof and exit-intent offers.
The Problem: One Site, Very Different Shoppers
CAL Ranch sells farm, ranch and outdoor gear, from feed to chainsaws, with stores across the West. Some shoppers buy online and pick up locally, others ship home. New visitors are researching, while returning ones know what they want. One experience for everyone was leaving money on the table.
Biggest Win: Say Less About Shipping
If we shorten the free shipping message for returning shoppers, more of them will buy, because they already know the offer and a long message gets in the way of the product.
A · Original
Full free shipping message with conditions and details
Long Message
BaselineReturning shoppersB · Variant · Winner
The same offer in a few words
Short Message
+177%Revenue · 260,693 users
The Other Results
Every test ran against a control group, split by audience. Results are Dynamic Yield's conclusive calls.
Test
Audience
Result
"Going fast" social proof badge, sporting goods
Products with 40+ views
+108% purchases · +97% revenue 19,773 users
Exit-intent pop-up: "Don't miss out, popular now"
New users, 1+ pageview
+41% revenue 60,403 users
Move out-of-stock products down the page
Desktop
+32% purchases · +35% revenue 42,902 users
Sticky cart summary
All users
+16% revenue 3,454 users
Hide product tags on product pages
All users
+5% purchases 2,634,582 users
Back to top button
Desktop
−3.4% revenue · lost 959,530 users
The sticky cart summary had a small sample, so treat it as a signal rather than a final answer.
Recommendations by Shopper Type
Alongside the copy tests, I ran redesigned product recommendations on the homepage, product pages and cart, split by new and returning shoppers and by whether a store offered pickup. A dashboard of incremental revenue per experience showed which ones to keep and which to switch off.
See the Test Report Screenshots
CRO-168: mini cart opens for returning visitorsIncremental revenue by experience, August 2023
+177%Revenue, shipping copy
+108%Purchases, social proof
+41%Revenue, exit intent
What I Learned
The biggest win removed words. Returning shoppers already knew the offer, so less copy let them get to the product. Social proof worked best where it was true and specific: only on items with real traffic. And the back-to-top button everyone liked cost revenue across nearly a million users. A losing test is still a result.
Of homepage visitors ever reached the first products.VWO scrollmap · 100 views · plus a 0.48% store conversion rate on 287,985 sessions
Case 06 · Cuadra: Finding the Leaks Before Testing
VWO · Jun – Oct 2023
ClientCuadra (luxury leather boots)
RoleSr. Ecommerce CRO Strategist, Publicis Sapient
ToolsVWO, session recordings, Shopify
Baseline0.48% conversion
OutputResearch and test designs
ProblemA 0.48% conversion rate on high-priced boots, with no clear picture of where shoppers were dropping off.
SolutionScrollmap, session recording and funnel research, turned into four targeted test designs.
The Baseline
Cuadra sells handcrafted boots, many of them $300 to $500. Over 30 days, 287,985 sessions produced 1,375 orders. People were interested, since 15,362 added to cart, but most never finished.
Sessions287,985100%
Added to Cart15,3625.3%
Reached Checkout6,5142.3%
Ordered1,3750.48%
Shopify, Sep 20 – Oct 19, 2023. Fewer than half of the people who added to cart reached checkout.
What the Research Showed
Scrollmap · Homepage
Only 45% of visitors scrolled past the hero to the first row of boots.
The hero was a large lifestyle image with no product and no clear next step.
Session Recordings
Recordings logged JavaScript errors during shopping sessions.
Red shipping warnings ("can not be sent to Nevada") sat right under the price.
Cart
A "We Do Not Ship to PO Boxes" pop-up covered the page as soon as the cart opened.
Free shipping was mentioned in the header but not where people decide, in the cart.
Mobile
Search was hidden behind an icon, and the header scrolled away.
Shoppers deep in a collection had no quick way back to search or the cart.
The Tests I Designed
Each finding became a test in VWO:
Test
Hypothesis
Mobile sticky nav with open search
Keeping the logo, cart and a visible search bar on screen will help mobile shoppers find boots and return to the cart.
Back in Stock badge
Telling shoppers a restocked size is back, with "Grab it before it's gone," will add urgency on high-priced items.
Free shipping message in the mini cart
Restating free shipping right above Check Out will lower doubt at the moment of decision.
Review star color (orange, gold, brand gold)
Stars people recognize from big marketplaces will make ratings easier to see on collection pages.
A · ControlHeader scrolls awayB · VariantSticky header with open search
See the Research and Test Screenshots
Scrollmap: 45% reach the first productsA blocking pop-up when the cart opensSession recording with errors loggedBack in Stock badge variantFree shipping message in the mini cartShopify baseline, 0.48%
What I Learned
With a 0.48% baseline and high prices, every test needs a lot of traffic to reach a result, so picking the right tests matters more than running many. Research made the list short: get people to the products, stop interrupting the cart, and make it easy to find a boot again on mobile.
For brands without a testing program yet, I start with a 10-part audit: first impression, navigation, content, trust, user journey, conversion points, checkout, interaction, forms and search. Each finding becomes a hypothesis on the roadmap.
Loren Stewart · Jewelry
Session recordings caught JavaScript errors right as shoppers added to bag.
Rage clicks on a mobile image carousel with no visible controls.
Hero CTA below the fold on mobile → first test on the roadmap.
Before focusing on CRO, I designed brands and websites at Web Lamb, LLC, my San Francisco studio. These projects are where I learned to build brands and sites that have to convert.
ProblemA networking startup needed a brand and a website ready for its product announcement. The site had to explain a technical product quickly and capture leads from day one.
SolutionA modular brand that mirrors the product, friendly illustrations that make networking feel simple, and a launch page built for clear messaging and lead capture.
What We Did
Logo: a mark built from modular pieces that snap together, echoing how SnapRoute’s networking components can be customized and swapped.
Logotype: custom lowercase letterforms for a simple, approachable, forward-looking feel.
Illustrations and icons: a set that makes an open-source networking stack feel simple and friendly.
Custom video player: built to showcase the codebase.
Launch landing page: designed for clear messaging and lead capture.
RebrandAssurX
AssurX: Standing Apart in Quality Management Software
ProblemAssurX makes quality management system (QMS) software for regulated industries and needed a brand that set it apart from close competitors.
SolutionA new identity with a color for each industry it serves, and a website organized so every buyer lands on content made for them.
What We Did
Discovery: a discovery phase to find a position AssurX could own.
Brand identity: a new logo and color palette, with a color for each of five industries: Energy & Utilities, Life Sciences & Medical Devices, Manufacturing, High Tech, and Food & Beverage.
Website: a responsive site organized by industry.
Lead generation: landing pages connected to Pardot for campaigns and demo requests.
Content: SEO, copywriting and redesigned infographics.
Handoff: a brand asset kit so the marketing team could stay on-brand.
ProblemMediasmith, a global media advertising agency, needed a refreshed brand and a website its own team could update with news, jobs and speaking engagements.
SolutionA rebrand with a new tagline and messaging, and a WordPress site the team runs themselves, backed by SEO and matching presentation and print materials.
What We Did
Long-term partnership: Web Lamb, LLC has been Mediasmith’s agency since 2012, delivering two website redesigns and one rebrand.
Rebrand: logo revision, a new tagline, messaging and copywriting.
Website (2017): a WordPress site the team updates themselves with news, job listings and speaking engagements.
Everywhere else: SEO, a PowerPoint template and print collateral to keep the brand consistent.
About Jennie
Experience & Skills
I'm a growth and product designer who lives where design meets data. I started in cognitive neuroscience research at Stanford School of Medicine, and I still run every test like an experiment. I've run CRO programs for ecommerce brands at Publicis Sapient, built and tested a mortgage application funnel at American Financing, and designed patient engagement flows for healthcare brands like EndeavorRx and Paragard.
My favorite part is the moment a test proves something nobody expected. I build AI agents, and I teach humans. Since college, I've been known for unusually sharp critical thinking and for explaining complex things in simple terms. I design and build AI agents, apps and workflows, with guardrails.
