EllieMD x Sadra Khazaei Targhi

UX Audit

EllieMD Design Audit

Five prioritized recommendations to reduce friction and increase conversions on the weight-loss page.

A note before we begin: The current site is clearly working. EllieMD is growing, customers are converting, and the foundation is solid. The suggestions here are opportunities to build on that success, not criticisms of what exists. Every recommendation comes from a place of respect for what the team has already built.

Reviewed on a 13" MacBook at 100% browser zoom

Opportunities

Prioritized by potential impact on conversion and user confidence. Each includes observations, potential impact, and a suggested approach.

Priority Key
P0Fix first
P1Very high
P2High
P0Fix first

1. Critical conversion blockers at key funnel moments

Why this is first priority: If users can't click a button or get stopped by an unexpected modal, they never make it to the next step. Addressing this first clears the path for everything else.

What I noticed

Two issues directly prevent conversion at the most critical moments. First, "Personalize Your Formula" CTAs appear across multiple sections, but only the hero CTA works reliably. Buttons in the Experience and Weight-Loss Kit sections fail intermittently. Second, after users complete personalization and see their custom formula with pricing, clicking "Begin Medical Intake" triggers an account creation modal. This hard stop happens at the exact moment users are ready to commit.

Evidence

Hero: CTA works reliably

Hero: CTA works reliably

The EllieMD Experience: CTA sometimes fails

The EllieMD Experience: CTA sometimes fails

Ultimate Weight-Loss Kit: CTA sometimes fails

Ultimate Weight-Loss Kit: CTA sometimes fails

Formula complete: user at peak intent, ready to commit

Formula complete: user at peak intent, ready to commit

Auth modal blocks progress at the commit moment

Auth modal blocks progress at the commit moment

Potential impact

For users

  • Immediate trust damage: if a button fails, users question whether checkout and medical intake will work
  • Momentum interrupted at peak intent: users were ready to proceed, now they face unresponsive buttons or account steps
  • Perceived time cost increases: signals the process may be longer than expected

For business

  • Direct conversion leak at the most valuable funnel stages
  • Highest-intent funnel point: any friction here disproportionately increases drop-off
  • Corrupted analytics: we can't learn which sections drive conversion if triggers don't fire consistently

Suggested approach

Unify all CTAs to a single reliable entry point and defer account creation until after eligibility

  • Every CTA scrolls to the selector and focuses it, or opens it as a modal/drawer
  • Progress before auth: let users start intake with a quick eligibility check (3-6 questions) before requiring account creation
  • Gate account creation after showing 'You're likely eligible' so it feels necessary, not arbitrary
  • Preserve user state: selections carry over, progress is shown throughout

Proposed Flow

Step 1
3-5 eligibility questions
No account needed
Step 2
"You're likely eligible"
Creates momentum
Step 3
"Create account to save progress"
Account feels necessary
vs current: auth modal at peak intent

This flow reduces friction at the auth step. But there's an even bigger win hiding here.

The bigger opportunity

This isn't just about reducing drop-off. It's about capturing users we'd otherwise lose forever.

Current flow: early exit
1
User builds formula
2
Hits auth modal
Leaves without converting
We know nothing about them.
No way to recover this user.
Proposed flow: recovery path
1
User builds formula
2
Completes quick intake
Eligible: continue to GLP-1
Not eligible: route to other products
We have baseline data.
Every user becomes a conversion opportunity.

By collecting eligibility data before requiring an account, we transform an early exit into a warm lead. A user who isn't eligible for GLP-1 might be perfect for wellness supplements, nutrition coaching, or other EllieMD offerings. With the current flow, we'd never know.

Success metrics

  • CTA click-to-selector engagement rate
  • Repeat-click rate on CTAs (indicates unresponsive buttons)
  • Auth modal abandonment rate
  • Time from personalization complete to intake submission

Effort estimate

Medium: front-end event wiring, intake flow restructuring, and session state handling

P1Very high

2. Content doesn't fit on standard laptop screens

What I noticed

At 100% zoom on a 13" laptop (one of the most common screen sizes), critical content doesn't fit in the viewport. The navigation header consumes significant vertical space, and the personalization flow requires constant scrolling. Users can't see their selection, the product, and the CTA together. Navigation buttons are cut off, making the flow feel stuck. This affects both the landing page hero and every step of the personalization flow.

Evidence

Hero at 100% zoom: header consumes space needed for value props

Hero at 100% zoom: header consumes space needed for value props

Potential impact

For users

  • Fragmented decision context: can't see selection, product, and CTA together when deciding
  • Navigation buttons cut off, making the flow feel less intuitive
  • Users on 13-14" laptops see a page that feels cramped and unfinished

For business

  • Conversion friction: every scroll between decision and action is a chance to lose the user
  • Perception of poor quality: a page that doesn't fit feels untested
  • Header + content overflow = compounding friction across every screen

Suggested approach

Optimize layout for common viewport heights (700-800px) and reduce header footprint

  • Audit header padding and logo size: tighten vertical spacing to reclaim 20-40px
  • On the personalization flow, use a minimal header (logo + close/exit) to maximize decision UI space
  • Use two-column layout on desktop: product preview fixed left, options and CTA scrollable right
  • Add sticky footer bar on shorter viewports with primary CTA and selection summary
  • Consider collapsing header that shrinks on scroll

Optimized Layout

Current (100% zoom, 13" laptop)
CTA cut off below fold
Proposed
CTA always visible
Reduce header height (20-40px savings)
Two-column layout: product fixed left, options scrollable right
Sticky CTA bar on shorter viewports

Success metrics

  • CTA visibility on page load (% of sessions where CTA is above fold)
  • Scroll-to-CTA distance per step
  • Step completion rate segmented by viewport height

Effort estimate

Medium: CSS/layout work across components. Quick wins on header, more work on flow layout.

P1Very high

3. Page doesn't address what users actually worry about

What I noticed

The page doesn't yet address the primary concerns of GLP-1 shoppers upfront. Side effects (the #1 concern) are tucked away in an FAQ accordion at the bottom. There's no competitor comparison, encouraging users to leave and research elsewhere. After the personalization selector, the next section shows feature tiles instead of addressing why users hesitated. The content talks about what EllieMD offers, not what users are wondering about.

Evidence

Post-CTA content: feature tiles don't address user concerns

Post-CTA content: feature tiles don't address user concerns

Potential impact

For users

  • Users arrive anxious about nausea, injection pain, and other known side effects
  • Without proactive reassurance, they may leave to research elsewhere
  • Real questions go unanswered: Will I qualify? What about side effects? How does this compare to Hims/Ro?

For business

  • Side effect anxiety is a primary reason prospects don't convert
  • Users who leave to research may find competitors who address concerns more transparently
  • Prime real estate underutilized: post-CTA space could be a high-ROI conversion driver

Suggested approach

Create dedicated content that proactively addresses side effects, eligibility, and competitive positioning

  • Add 'What to Expect' section above fold or after primary CTA: list common side effects with percentages and typical duration
  • Reframe as 'manageable and temporary' with specific support (nausea meds included, clinician messaging)
  • Replace post-CTA feature tiles with objection-recovery block: eligibility, billing, side effects, timeline
  • Add 'How We Compare' section: EllieMD vs typical telehealth GLP-1 providers
  • Include real patient quotes about side effect experience and how they were managed

Objection Recovery Block

Accordion: Address real concerns
What are the side effects?
+
Will I qualify?
+
When am I charged?
+
How does EllieMD compare?
+
Persistent: Selection summary + CTA
Your Selection
Type
Formula
Additive
Total$299/mo
Begin Medical Intake
Always visible as users explore concerns
"What to Expect" Section (expanded)
Nausea (30-40%, usually weeks 1-2)
Fatigue (20%, temporary)
Nausea meds included
Unlimited clinician messaging

Built-in customer research

These accordion clicks aren't just UX, they're data. Track which objections get opened most and we'll know exactly what's blocking visitors from converting.

42%
clicked "side effects"
31%
clicked "eligibility"
18%
clicked "billing"
9%
clicked "compare"

If 42% click "side effects," that's a clear signal to address it in ads, email sequences, and above-the-fold content. The page becomes a learning engine, not just a conversion tool.

Success metrics

  • Scroll depth to side effects content (should be much higher on page)
  • Time on site (expect increase when comparison content keeps users engaged)
  • Exit rate from landing page (expect decrease)
  • Objection card open rates (reveals what's blocking users)

Effort estimate

Medium: content creation, competitive research, section design, and page restructuring

P1Very high

4. Personalization flow creates confusion instead of confidence

What I noticed

The personalization experience has multiple UX issues that compound into decision paralysis. Selection cards don't look obviously clickable (soft styling, chevrons suggest expansion not selection). Dense paragraph-heavy content requires too much parsing. The entire flow frames treatment as a shopping menu rather than clinician-guided care, repeatedly signaling users are responsible for choosing the "right medication" for a high-stakes medical decision.

Evidence

Step 1: cards look passive, chevrons suggest expansion not selection

Step 1: cards look passive, chevrons suggest expansion not selection

Potential impact

For users

  • Unclear what to do: cards look like info displays, not interactive selectors
  • First-time GLP-1 shoppers may feel: 'Am I choosing the right one for me?'
  • Too many permutations trigger choice anxiety and 'I'll research later' behavior

For business

  • Slows funnel: users who don't immediately understand what to click will pause or leave
  • Increases support burden with 'Which one should I pick?' questions
  • Risk of refunds when expectations don't align with clinical matching

Suggested approach

Add clear interaction affordances and reframe around clinician guidance instead of user shopping

  • Add hover states with clear visual feedback (border change, subtle lift/shadow) on selection cards
  • Replace chevrons with radio-style indicators that signal 'select one'
  • Change language from 'Select your medication' to 'Tell us your preference. Your clinician will recommend what's appropriate.'
  • Reformat education sections into scan-first decision cards: Best for, What to expect, Common concerns
  • Thread consistent narrative: Tell us about you, Clinician matches you, Medication + support

Improved Selection Cards

Current: Ambiguous affordances
›
›
Chevrons suggest "expand" not "select"
Proposed: Clear selection indicators
Radio indicators signal "select one"
Reframe: Clinician-guided, not shopping
Current copy
"Select your medication"
"Choose your additive"
Proposed copy
"Tell us your preference"
"Your clinician will recommend"

Brand positioning through UX

This isn't just about reducing anxiety. It's about positioning EllieMD differently in a crowded market.

Competitors (Hims, Ro, etc.)
E-commerce shopping UX
"Add to cart" framing
User as shopper
EllieMD opportunity
Clinical consultation UX
"We'll guide you" framing
User as patient

Everyone else uses shopping UX because it's familiar. EllieMD can own "medical-first" positioning through interaction design alone. When users feel like patients instead of shoppers, price becomes less of an objection and trust becomes the differentiator.

Success metrics

  • Time to first selection per step
  • Selector to Begin Medical Intake click-through rate
  • Intake completion rate (drop-off at option-choice steps should decrease)
  • Fewer support questions about which option to pick

Effort estimate

Medium to High: interaction design, copy updates, content restructuring, and selector logic

P2High

5. Inconsistent button styles erode trust and brand perception

What I noticed

Buttons across the site use different components, casing, and terminology for the same actions. The header uses "SIGN IN" (outlined, all caps with letter spacing) while the auth modal uses "Login" (filled, title case). Primary CTAs use all caps ("BEGIN MEDICAL INTAKE") but the modal uses title case. The Register button is coral while other primary actions are teal. This inconsistency signals lack of attention to detail.

Evidence

Header: 'SIGN IN' outlined, 'GET STARTED' filled, both all caps

Header: 'SIGN IN' outlined, 'GET STARTED' filled, both all caps

Potential impact

For users

  • Visual inconsistency erodes trust: the site feels assembled from different templates
  • Terminology confusion: 'Sign In' vs 'Login' for the same action creates micro-friction
  • The auth modal feels disconnected from the rest of the experience

For business

  • Brand perception suffers: inconsistency signals lack of polish
  • Increased cognitive load at conversion points
  • Technical debt: maintaining multiple button variants slows future development

Suggested approach

Establish and enforce a unified button system

  • Define 3-4 button variants: Primary (teal filled), Secondary (outlined), Tertiary (text link)
  • Standardize typography: choose all caps with letter spacing OR title case, apply consistently
  • Unify terminology: pick 'Sign In' or 'Log In' and use it everywhere
  • Rebuild auth modal with the same button components used elsewhere
  • Document the button system to prevent future drift

Unified Button System

Define 3 consistent variants
Primary
BUTTON TEXT
Main actions, CTAs
Secondary
BUTTON TEXT
Secondary actions
Tertiary
BUTTON TEXT
Text links, navigation
Current: Inconsistent
Header: "SIGN IN" (all caps, outlined)
Modal: "Login" (title case, filled)
CTA: "BEGIN MEDICAL INTAKE" (all caps)
Modal: "Register" (title case, coral)
Proposed: Unified
All caps with letter spacing
Consistent terminology: "Sign In" everywhere
Same component across all contexts
Single accent color for primary actions

Success metrics

  • Design system audit: count of non-conforming button instances
  • Auth modal completion rate (before/after unification)
  • User feedback mentioning 'confusing' or 'inconsistent' UI

Effort estimate

Low to Medium: component standardization and find-replace across templates

Suggested Roadmap

One way to phase this work based on effort and impact. Quick wins first to build momentum, then the higher-lift items.

Week 1

Quick Wins

Low effort, immediate impact

  • #5Standardize button components and terminology
  • #2Reduce header height, tighten padding
  • #1Unify CTA event handlers across sections

Weeks 2-3

Phase 1

Highest conversion impact

  • #1Defer account creation until after eligibility check
  • #4Add hover states and selection indicators to cards
  • #2Implement sticky CTA bar for short viewports

Weeks 4-6

Phase 2

Content and structural improvements

  • #3Create side effects and comparison content
  • #3Redesign post-CTA section for objection handling
  • #4Reframe flow language around clinician guidance

Task 2: Redesign

Personalization Flow Prototype

I chose to redesign Opportunity #4 (Decision Flow Friction) because it touches the core product experience and has compounding effects on conversion. The prototype includes two variants to test different approaches to decision fatigue.

What's different in this redesign:

1
2
3

Progress bar and clickable stepper

Visual progress at the top. Completed steps are tappable to jump back.

What's the difference?

Education at the point of decision

Helper buttons expand inline comparisons exactly where users need them.

Clear selection affordances

Radio indicators, hover effects, and smooth animations.

Best for results
Budget friendly

Scannable decision cards

Key differentiators surfaced immediately, not buried.

Continue

Sticky CTA button

Continue button stays visible even when helper content is expanded.

Viewport-optimized layout

Product preview, selections, and CTA visible together.

Two variants to test:

Variant AQuick Start

Smart defaults are pre-selected on load. Users start with the recommended configuration and only change what they want.

NAD+ (wellness boost) auto-selected in Step 3
No skip buttons, so users engage with each step
Helper buttons available for education

Hypothesis: Pre-selecting the best option reduces decision fatigue. Users who are unsure will go with the default rather than abandoning.

Variant BDoctor Decide

No defaults pre-selected. Users start fresh and must actively select an option, with an escape hatch if they're unsure.

Continue button disabled until selection made
"Let my doctor decide" skip button on all steps
Skip auto-selects smart default and advances

Hypothesis: An explicit "let my doctor decide" escape hatch prevents abandonment from decision paralysis while still letting engaged users choose.

What's the same in both: The 4-step flow (Method → Plan → Boost → Review), all card options, pricing, educational content, and the product showcase with animations. Toggle between variants using the floating A/B button or the ?variant=a or ?variant=b URL parameter.

Live Prototype

Use the floating A/B button in the prototype to switch between variants

That covers both the audit and the redesign. Happy to walk through the prototype live, discuss tradeoffs, or talk about how we'd approach implementation with engineering.