CRO Advanced

Lineage: upgraded from IDS_CRO_Advanced.md. Practitioner base: synthesis of 50+ DTC CRO practitioners, UX researchers, and direct-response operators (original source framing); tool ecosystem referenced throughout: CXL Institute (Peep Laja), Intelligems, VWO, Convert.com, Hotjar, Microsoft Clarity, ReConvert, Okendo, Stamped.io. Current as of July 2026.

---

"Most brands treat CRO as a collection of tactics: change a button color, add a trust badge, try a new headline. Elite operators treat CRO as a system: understand the psychology of a specific customer segment at a specific moment of intent, remove every friction that stands between that intent and conversion, and then test everything at the hypothesis level." — Synthesis of DTC, SaaS, and direct-response CRO practitioners

What this module is not: a repeat of LUCE_08. If you haven't read LUCE_08 (Store CRO), read it first — it covers the 18-element product page, the shipping-speed promise, basic trust architecture, and the operating checklist a solo operator runs monthly. This module is the deep-dive twin: the statistics behind testing, the psychology behind every trust element, the checkout funnel broken into its stages, and the agentic-commerce and AI-copy science at a level of technical depth LUCE_08 deliberately left out to stay executable in a week.

---

#THE ONE-PAGE VERSION

1. CRO compounds — and the compounding math is bigger than most operators believe, even at 2026's lower CVR baselines. Doubling site-wide CVR on the same traffic doubles revenue with $0 additional ad spend. That's not a slogan; Part 1 works the actual numbers at both a $1k-operator scale and a mature-store scale.
2. Recalibrate every CVR benchmark you've internalized. The category tables in this module (Part 1.3) replace the aspirational 1.5–5% ranges common in older CRO literature. 2026 reality: Shopify-store average ~1.4% site-wide, cold paid-social 0.5–1.2%, email 4–5.3%, mobile 1.8–2.5% vs. desktop 3.5–4%. Category PDP-level CVR runs meaningfully higher than site-wide (food 4.5–6%, beauty 3–4%, electronics ~3.6%, luxury <1.2%) — know which number you're being quoted.
3. The conversion equation is unchanged, the inputs aren't: CVR ∝ (Clarity × Desire × Trust) / (Price Resistance × Risk Perception × Friction). In 2026, "Risk Perception" carries a new, specific component — the silent Amazon/Temu price comparison every cold visitor runs whether you address it or not.
4. The CRO Hierarchy of Needs still governs sequencing. Don't test button colors (Level 5) while your value proposition is unclear (Level 2) or your checkout is broken on mobile (Level 1). This module adds a Level 0: machine-readable — if your schema doesn't parse, an AI shopping agent can't even evaluate your page.
5. The checkout funnel has four measurable stages, and each has its own benchmark. Cart→checkout initiation runs 40–60%; initiation→completion runs 45–65%; overall cart→purchase runs ~25–40%. Below 25% overall, checkout itself — not the product or the ad — is where you're bleeding.
6. Express checkout (Shop Pay, Apple Pay, Google Pay) is conversion infrastructure, not a feature toggle. Shop Pay converts returning customers at 1.7× guest checkout; express buttons cut checkout friction by roughly 20%. Treat their absence as equivalent to a broken checkout button.
7. Post-purchase upsells deserve full LTV-aware modeling, not just an attach-rate estimate. A ReConvert-style Thank You Page offer converting 8–15% of buyers, at near-pure margin because CAC is sunk, materially changes contribution-margin math under 2026's compressed 3–7% generic-dropship baseline. Part 3.4 builds the full model.
8. A/B testing has hard statistical floors, and 2026's lower CVRs push those floors higher, not lower. At a 1.4% baseline, detecting a 10% relative lift needs roughly 14,000–15,000 visitors per variant. Below that, you are not testing — you are watching noise and calling it a decision.
9. Qualitative research generates the hypotheses that make quantitative testing worth running. Session recordings, heatmaps, exit surveys, post-purchase surveys, and the 5-second test are how elite operators know what to test before they spend the traffic to test it.
10. The 7 psychological principles are the substrate under every tactic in this course. Social proof, scarcity, authority, reciprocity, liking, commitment/consistency, and loss aversion aren't a checklist — they're why the checklist works. AIDA+ (Attention, Interest, Desire, Proof, Action, Reassurance) is the copy structure that deploys them in sequence.
11. Agentic commerce readiness has a technical floor: valid, complete Product schema. An AI shopping agent transacts on your JSON-LD, not your visual design. Broken or stale schema is worse than none — it signals unreliability to the exact channel growing 340%+ YoY.
12. AI-drafted copy at scale needs a QA rubric, not just a review gate. Part 8.2 gives the specific rubric a solo operator can run against every AI-drafted page before publish — factual accuracy, claim substantiation, brand-voice match, and regulatory exposure, each scored, not just eyeballed.
13. Price is the highest-leverage single test available, and almost nobody runs it. A $5 increase on a $45 product with no measurable CVR impact is an 11% revenue increase on every future unit — true price testing (not manual price changes) requires a tool built for it (Section 5.4).
14. The monthly CRO ritual is a four-week cycle, not a monthly glance. Data review → hypothesis generation → test running → winner implementation. Skipping a week breaks the cycle and produces untested opinions instead of validated wins.
15. This module hands off to LUCE_09 (Finance & Scaling) once your CRO discipline is producing a stable, testable funnel — the next problem is unit economics at scale, not conversion rate.

---

#PART 1: THE CONVERSION FRAMEWORK — FIRST PRINCIPLES, RECALIBRATED FOR 2026

#1.1 Why CRO Is Still the Highest-ROI Activity in DTC

Every dollar of traffic becomes more valuable when your store converts better. That's true at any scale, and it's worth re-deriving with numbers that match where most of this course's readers actually operate — not the mature-brand numbers the original module used.

The compounding math, lean-operator scale:

Store A: 10,000 monthly visitors x 1.4% CVR (Shopify average) x $65 AOV
  = 140 orders x $65 = $9,100 monthly revenue

Store B (same traffic): 10,000 visitors x 2.0% CVR x $65 AOV
  = 200 orders x $65 = $13,000 monthly revenue

Revenue increase from CRO alone: +$3,900/month (+42.9%), $0 additional ad spend

The compounding math, mature-store scale (the original module's frame, recalibrated):

Store A: 100,000 monthly visitors x 1.4% CVR x $65 AOV = $91,000/month
Store B (same traffic): 100,000 visitors x 2.5% CVR x $65 AOV = $162,500/month

Revenue increase: +$71,500/month (+78.6%), same traffic budget

The original module's illustrative jump (1.5%→3.0%, a full doubling) was aspirational even in 2024; at 2026's lower baseline it overstates what's realistically achievable through CRO alone. A 1.4%→2.0–2.5% improvement is the honest, achievable target range for a store that fixes its fundamentals (LUCE_08's 18-element page, checkout defaults, mobile-first execution) without needing a paid CRO agency. That range still produces a 40–80% revenue lift with zero incremental ad spend — which is why "improve conversion before scaling spend" remains the correct sequencing rule even after recalibration.

---

#1.2 The Conversion Equation

Conversion is not a random event. It follows a predictable logic that hasn't changed structurally since the original module — what's changed is one specific input.

Conversion happens when:
Perceived Value of Offer > (Price + Risk Perception + Effort)

CVR ∝ (Clarity x Desire x Trust) / (Price Resistance x Risk Perception x Friction)

The 6 conversion killers, in rough order of impact — with the 2026 addition:

1. Unclear value proposition — visitor can't immediately understand what the product does and for whom (LUCE_08 Section 2.3, the 3-second test)
2. Missing trust signals — no social proof, no reviews, no credibility markers
3. Price-value mismatch perception — price feels too high relative to perceived value, and in 2026 this perception is anchored against a specific, real comparison: the Temu/Amazon listing the visitor may have already seen (LUCE_08 Section 3.1) — this is now a specific, addressable input, not a vague "price resistance" term
4. Friction in the purchase path — slow load times, too many steps, missing express-checkout options (Part 3)
5. Wrong traffic-page match — ad creative sets an expectation the landing page doesn't deliver (Part 2.2)
6. No urgency — no reason to buy now vs. later; later, for most visitors, means never

---

#1.3 CRO Benchmarks by Vertical — 2026 Recalibration

The original module's benchmark table used ranges (1.5–5% for beauty, for instance) that read as achievable "good" targets in 2024–2025 CRO literature. Two things changed: the underlying baseline dropped (mobile-heavy, cold-paid-heavy traffic mixes convert lower than the literature's implicit assumptions), and the fact sheet gives us actual mid-2026 category data at a different level of the funnel than the original table measured. Reconciling those honestly matters more than picking a flattering number — a store benchmarking against the wrong denominator will chase a problem that doesn't exist, or miss one that does.

Table A — Site-wide CVR by vertical (blended traffic, all pages in the denominator), 2026 estimate anchored to the Shopify ~1.4% blended average:

These bands are directional estimates derived by scaling the original module's relative category ordering down to the 2026 blended-average anchor. Verify against your own GA4 data before treating any single number as gospel — this table exists to correct for "I read online that 3% is normal," not to replace your own measurement.

Table B — Session/PDP-level CVR, directly from mid-2026 market data (verified, not estimated):

Why Table A and Table B look so different for the same categories: they're measuring different denominators. Table A includes every page view — home, collection, blog, non-buying sessions — in the denominator, which is what "site-wide CVR" always means. Table B measures the narrower session/product-page-level conversion, which the original module correctly noted runs 2–3× higher than site-wide CVR because it excludes non-product traffic from the denominator. Use Table A to benchmark your Shopify Analytics "conversion rate" metric. Use Table B only if you're specifically measuring product-page or session-level conversion, and confirm which one your tool is actually reporting before you compare — this single confusion is one of the most common CRO misdiagnoses a solo operator makes.

---

#1.4 The CRO Hierarchy of Needs

Address in this order. The original 5-level hierarchy holds; 2026 adds a foundational layer underneath it.

Level 0 — Machine-readable (new for 2026, must work before Level 1 matters for AI-referred traffic):

• Product schema (JSON-LD) validates cleanly (Part 8.1)
• Price, availability, and shipping data in the schema match what's actually true on the page
• Agentic Storefronts toggle enabled (Part 8.1) — a $0, one-time setting

Level 1 — Functional (must work):

• Page loads in <3 seconds (mobile especially — Part 6.3)
• Checkout process works on all devices
• Images load at correct resolution
• No broken links or 404s in the purchase path

Level 2 — Usable (must be understandable):

• Clear product description: what it is, what it does, who it's for
• Obvious call to action, not buried or confusing
• Price clearly displayed
• Variants selectable without confusion

Level 3 — Trustworthy (must be believed):

• Reviews/ratings — the original module's "20+ reviews before meaningful trust signal" threshold still holds
• Guarantee/return policy visible before checkout
• Secure checkout signals (SSL badge, payment logos)
• Real brand presence (About page, contact info)
• New for 2026: the Amazon/Temu comparison addressed explicitly (LUCE_08 Section 3.1) — trust in 2026 includes trust that you know your visitor has cheaper alternatives and you're not pretending otherwise

Level 4 — Persuasive (should compel):

• Social proof (UGC, review highlights, "X customers love this")
• Scarcity and urgency (genuine low stock, limited time — Part 4.1)
• Comparison to alternatives, done honestly (why us vs. them)
• Benefits-forward copy that activates desire

Level 5 — Optimized (systematic refinement):

• A/B testing at every stage of funnel (Part 5)
• Personalization by traffic source, device, geography
• Advanced triggers and behavioral targeting
• Micro-optimization of copy, layout, and design

Don't skip levels. A store running Level 5 button-color tests while its Level 0 schema is broken is optimizing a channel that can't transact on it at all.

---

#PART 2: THE ADVANCED PDP AND LANDING PAGE SCIENCE

#2.1 The 12-Section PDP Architecture, Reconciled with LUCE_08's 18-Element Page

LUCE_08 gives you the operational 18-element checklist to build against. This module's 12-section architecture is the same underlying page, organized by psychological function rather than build order — useful when you're diagnosing why a page underperforms, not just confirming every element exists.

The advanced diagnostic use of this table: when a page underperforms, don't ask "which element is missing" (LUCE_08's job) — ask "which psychological function is failing." A page can have all 18 elements present and still fail if, for instance, the Trust strip and payment icons are visually buried (Risk Reduction function fails) even though the elements technically exist. Session recordings (Part 7.2) are how you catch this; a checklist alone can't.

---

#2.2 Landing Pages vs. PDPs — The Advertorial Funnel

For paid traffic — especially cold traffic — dedicated landing pages consistently outperform PDPs. This was true in the original module and remains true; nothing about 2026's platform shifts changed the underlying mechanism.

Why landing pages beat PDPs for paid traffic:

• Navigation removed = no distraction path out of the purchase funnel
• Message match: landing page headline mirrors the ad's hook exactly (continuity principle — a visitor who clicked on "back pain relief for remote workers" should land on a headline that says exactly that, not a generic product title)
• Layout optimized for conversion, not general browsing
• Room for longer-form copy, video, and sales-letter structure not appropriate on a PDP a returning or organic visitor also sees

When to use a dedicated landing page:

• Any ad spend above roughly $300/day directed at a single product
• Testing a new product concept — a landing page iterates faster than a full PDP rebuild (feeds directly into LUCE_03's validation ladder, Rung 4)
• Bundle or kit offers the standard PDP template isn't built to present
• Advertorial traffic specifically (below)

The advertorial funnel:

Ad -> Advertorial (editorial-style content page) -> Landing page -> Checkout

The advertorial "warms up" cold traffic by reading like a news article or personal story rather than a sales pitch — no visible Shopify navigation, no obvious "this is an ad" signaling. Conversion-rate lift over sending the same traffic direct-to-PDP: 20–50% in many verticals, strongest for supplements, beauty, and problem-solution products where the buying decision benefits from narrative context before the pitch.

The discipline point: an advertorial that doesn't feel like content — that reads as an ad wearing a content costume — converts worse than no advertorial at all, because it triggers the exact skepticism it's designed to avoid. If you can't write one that would pass as genuine editorial content to a skeptical reader, skip the layer and send traffic direct to a well-built landing page instead.

---

#2.3 Trust Architecture Deep-Dive: Winning the Amazon/Temu Comparison at Scale

LUCE_08 Section 3.1 gives you the comparison table pattern and a copy template. At scale — once you have enough traffic to test (Part 5) — this becomes a hypothesis-driven optimization problem, not a one-time copy write.

What to test, in priority order, once you have the traffic (Part 5.1 for the floor):

1. Placement — does the comparison module convert better as element [14] mid-page, or pulled up to a smaller version directly under the price (closer to where the mental comparison actually happens)?
2. Framing — direct comparison table (side-by-side) vs. narrative FAQ answer (LUCE_08's copy pattern) vs. a single trust-badge line ("Not the cheapest. Worth it.")
3. Specificity — generic quality claims vs. a specific, verifiable differentiator (a named material spec, a named guarantee term, a named support-response-time commitment)

Where the data comes from: post-purchase survey (Part 7.2) is the single best source here. The question "what almost stopped you from buying?" surfaces the actual Amazon/Temu objection in the customer's own words far more reliably than guessing — and if the answer never mentions price or marketplace alternatives, that's a signal your comparison module may be solving a problem your actual customers don't have, and the page real estate might convert better used elsewhere.

A note on honesty at scale: as you test comparison framings, resist the pull toward exaggerating the marketplace alternative's weaknesses. A comparison that's discovered to be misleading — inflated Temu shipping times, invented quality claims — costs more in review-section backlash and refund disputes than the conversion lift is worth. Test wording and placement; don't test dishonesty.

---

#PART 3: CHECKOUT SCIENCE

#3.1 The Checkout Funnel — Stage-by-Stage Conversion Math

The checkout funnel has four measurable stages. Each stage loses people, and each stage has its own diagnostic signature — treating "checkout conversion" as one number hides where the actual leak is.

Benchmark cart-to-purchase conversion rates:

The diagnostic rule: if your overall cart-to-purchase rate is below 25%, checkout itself is a significant leak — not the product, not the ad. Split the two stages separately before you touch anything: a weak cart→initiation number points at cart-page friction or price-shock (Section 3.2); a weak initiation→completion number points at checkout-page friction specifically (Section 3.3).

---

#3.2 Cart Optimization Deep Dive

The cart page is frequently neglected; treated correctly, it's an active conversion tool, not a passive holding pen.

Cart page best practices, with the mechanism each one addresses:

1. Product image + name + variant visible in cart — confirms the right item was added; removes a specific, common source of last-second doubt
2. Price subtotal prominent — no surprises carried forward into checkout; surprise costs at checkout are one of the most commonly cited abandonment reasons in exit-survey data (Part 7.2)
3. Free-shipping threshold indicator — "Add $15 more for free shipping" simultaneously increases AOV and reduces abandonment; it reframes an upsell as a reward, not a pitch
4. Trust badges on the cart page — guarantee, secure checkout, returns policy; repetition of Level 3 trust signals costs nothing and compounds
5. Cart upsell — cap at roughly 30% of cart value; "Complete your order: add [product] for $X" with one-click add. Beyond 30%, the upsell starts competing with the base purchase decision instead of complementing it
6. Discount code field — visible but not prominent; hide behind a "Have a promo code?" collapsible. A prominent field prompts coupon-hunting behavior that sends otherwise-ready buyers off-site to search for a code, some fraction of whom never return

---

#3.3 Checkout Page Optimization — Shop Pay, Apple Pay, Google Pay as Conversion Infrastructure

Shopify's checkout is largely templated; customization is limited outside Shopify Plus. What you can control is more consequential than most operators treat it.

What you can optimize:

• Order summary — ensure product images render in the order summary panel
• Trust elements — Shopify allows trust-badge customization in the checkout footer; use it
• Express checkout buttons at the TOP of checkout, not buried below the fold — this single placement decision is worth restating from LUCE_08 because it's the most commonly under-executed item on this list: Shop Pay, Apple Pay, and Google Pay reduce checkout-completion friction by roughly 20% when placed prominently, and materially less when placed at the bottom
• Address field autofill — Shop Pay and Google Pay autofill dramatically reduce manual entry friction, which matters disproportionately on mobile (Part 6)

The Shop Pay number, worked: Shop Pay converts returning customers at 1.7× the rate of guest checkout. For a store with a growing repeat-customer base (post-purchase flows from LUCE_20 build this over time), that's not a marginal optimization — it's close to the single highest-leverage lever available for the returning-customer segment specifically, and it costs nothing beyond enabling the setting.

Why this matters more in 2026 than in the original module's framing: in 2026, a visitor who has just closed a Temu or Amazon tab has already experienced one-tap express checkout as the default expectation, not a premium feature. A store without prominent express checkout isn't merely missing an optimization — it's presenting friction the visitor has been trained to consider abnormal.

---

#3.4 Post-Purchase Upsell — The Full ReConvert Economics Model

LUCE_08 Section 4.3 gives you the basic attach-rate math. Here's the fuller model, including the LTV dimension the basic version leaves out.

The core mechanism, restated: a post-purchase upsell app (ReConvert, Zipify OCU, or equivalent — verify current pricing at the vendor site; this category has historically run roughly free-to-$60/mo scaling with order volume) inserts a one-click offer between order placement and the confirmation page. No new checkout, no new payment entry — the highest-trust, lowest-friction moment in the entire customer relationship.

The single-order model (from LUCE_08):

Base order: $59.99 retail, $14.10 landed/fulfilled cost
  -> Contribution margin before ads: ~57.7% ($34.62)

Upsell offer: $19.99, $4.50 landed cost
  -> Contribution margin on upsell: ~77.5% ($15.49)

At a 10% attach rate, per 100 base orders:
  10 x $15.49 = $154.90 incremental profit, $0 incremental CAC

The LTV-aware extension: the upsell's value isn't just the immediate $15.49 per converted order — a customer who takes a post-purchase upsell has demonstrated a second buying decision within minutes of the first, which correlates with higher repeat-purchase propensity generally. If that customer's expected 12-month order count rises even modestly (say, from a baseline 1.3 orders/customer to 1.6 for upsell-takers — a figure you should measure in your own cohort data, not assume), the true value of the upsell moment includes:

True upsell value per converted customer ≈
    Immediate upsell margin ($15.49)
  + (Incremental expected future orders x average order margin)

Why this matters for a $1k operator specifically: at compressed generic-dropship margins (3–7% net, per the fact sheet), the post-purchase moment is disproportionately valuable because it's one of the only revenue sources in the entire funnel that arrives at 70%+ contribution margin instead of single-digit net margin. Prioritize building this flow correctly before optimizing almost anything else downstream of the first sale.

Testing the offer itself (once traffic supports it, Part 5.1): test price point (a lower-priced offer often lifts attach rate more than it costs in per-unit margin — model both), test single-offer vs. two-tier offer (a "good/better" choice architecture), and test the offer's framing (discount-off-list vs. flat bundle price) — these are Tier 2 tests per the priority table (Part 5.1), not the first thing to test on a low-traffic store.

---

#3.5 Checkout Abandonment Data Collection

Exit survey on checkout abandonment: install Hotjar or Microsoft Clarity; trigger a one-question survey on exit from the checkout page: "What stopped you from completing your order today?" with options like Price too high / Wasn't ready to buy / Found a better deal elsewhere / Technical issue / Just browsing.

Why this outranks A/B test results for diagnostic value: an A/B test tells you what worked, not why. The exit survey tells you the actual stated reason — correlation vs. plain evidence. If "found a better deal elsewhere" dominates your responses, that's a direct signal to strengthen the comparison module (Section 2.3) before testing anything else in checkout.

---

#PART 4: PSYCHOLOGICAL CONVERSION ARCHITECTURE

#4.1 The 7 Core Psychological Principles in Ecommerce

LUCE_08 Section 3.2 gives you the practical implementation table with directional lift figures. Here's the deeper mechanism behind each, plus the scale-of-persuasion ranking for social proof specifically — the principle with the most internal variation in effectiveness.

1. Social Proof — scale of persuasion, highest to lowest:

• Video testimonials from customers similar to the prospect
• Before/after photo reviews with a story attached
• Written reviews with a verified-purchase badge
• "X people are viewing this right now" (real-time social proof)
• "X sold in the last 24 hours"
• "Best seller" or "#1 rated" badges
• Star rating + review count in the header

Implementation discipline: don't just install a generic review widget. Feature specific reviews that answer specific objections. If your top objection (from exit-survey and post-purchase-survey data, Part 7.2) is "does it really work?", lead with transformation-focused reviews. If it's "is this legit?", lead with verified-purchase reviews that include photos.

2. Scarcity — must be genuine, or it destroys trust permanently.

Fake scarcity is discovered eventually — a customer who reorders and sees "only 2 left" for the third consecutive month learns the badge is theater, and the trust cost extends beyond that one interaction to everything else on the page.

3. Urgency — distinct from scarcity: time-based, not quantity-based. Shipping-cutoff countdowns ("order in the next 4 hours for Thursday delivery") are consistently the most effective and most authentic urgency mechanism available, because they're simply true — your actual fulfillment cutoff exists whether you display it or not.

4. Authority — relevance-gated. Press logos, "formulated by Dr. [Name]," certifications (cGMP, NSF, USDA Organic, dermatologist-tested), founder credibility. Authority only works when the credential is relevant to the claim — a nutrition credential matters for a supplement, is irrelevant (and can read as padding) for a clothing brand.

5. Reciprocity — give before you ask. Free valuable content, free samples, free-gift-with-purchase framing. Tests consistently show free-gift-with-purchase outperforms an equivalent-dollar discount on both AOV and CVR — customers register "getting more" even when the monetary value is identical to a straight discount.

6. Commitment and Consistency — the micro-commitment ladder:

1. Visitor reads a blog post (micro-commitment to content)
2. Visitor signs up for email (micro-commitment to the brand — feeds LUCE_20)
3. Visitor takes a quiz (micro-commitment to customization)
4. Visitor adds to cart (micro-commitment to purchase intent)
5. Visitor completes checkout (purchase)

A visitor who completes a product-recommendation quiz converts at roughly 3–4× the rate of a standard visitor — the mechanism is that they've invested time and received a personalized recommendation, both of which create psychological pressure toward consistency with that investment.

7. Loss Aversion — people are roughly twice as motivated by avoiding a loss as by gaining an equivalent gain.

---

#4.2 The AIDA+ Framework — Deploying the Principles in Sequence

Attention — interrupt the scroll; pattern interrupt in the first image or headline. Interest — connect to a pain or desire the customer actually has (mined via LUCE_03 Section 2.2's review-mining method). Desire — make the transformation visceral and real, not abstract. Proof — social proof, testimonials, data; make the claim believable, not just stated. Action — clear CTA, with genuine urgency/scarcity if applicable. Reassurance — the "+" most brands skip. Guarantee, risk reversal, easy returns.

Why Reassurance is the differentiator most brands miss: a strong money-back guarantee removes the final barrier for fence-sitters specifically — the segment that's already convinced on desire and proof but hasn't crossed the risk threshold. "If you're not 100% satisfied in 30 days, we'll refund every penny" converts skeptics that a merely well-argued page does not, because it addresses Risk Perception (Section 1.2) directly rather than trying to out-argue it.

---

#PART 5: STATISTICAL A/B TESTING MASTERY

#5.1 Testing Hierarchy — What to Test First

Not all tests are equal, and 2026's lower baseline CVRs make the traffic cost of testing the wrong tier first even more expensive than it was in the original module.

The rule, unchanged and still routinely violated: only test Tier 3 after Tiers 1 and 2 are fully optimized. Running a button-color test while the value proposition is unclear is wasted traffic — worse in 2026 than in the original module's era, because traffic itself costs more (blended CAC $68–84 per the fact sheet) and lower baseline CVRs mean you need more of it to reach significance in the first place.

---

#5.2 Statistical Validity Requirements

Minimum requirements for a valid A/B test:

• Sample size: minimum 300 conversions per variant — conversions, not visits
• Duration: minimum 1 business week to capture weekly seasonality; 2 weeks ideal
• Statistical confidence: 95% minimum before calling a winner; 90% acceptable only for genuinely low-stakes tests
• Single variable: change exactly one element per test — change headline and button copy together and neither result is attributable

The traffic math, recalibrated to 2026 baselines (this is the single most important update this module makes to the original's testing section):

The original module's illustrative figures assumed a 1% or 3% baseline; the 2026 reality is that most new stores sit below 1.4% blended on cold traffic specifically, which means the traffic requirement for a valid test is higher, not lower, than the original module implied. Implication, restated from LUCE_08 because it can't be overstated: below roughly 10,000 monthly visits, formal A/B testing of individual page elements is impractical. Focus on qualitative research (Part 7) and structural changes (the full 18-element page, checkout defaults) instead.

Statistical significance calculator: use ABTestGuide.com or your testing tool's built-in calculator. Never call a winner by eyeballing a percentage difference — a 20% observed lift on 40 conversions per variant is statistical noise, not a result.

---

#5.3 Testing Tools

A note on tool selection: the tool doesn't fix a traffic-floor problem. Buying VWO at 3,000 monthly visits doesn't make your tests statistically valid — it just makes the invalid tests easier to run. Match the tool tier to the traffic tier, not the other way around.

---

#5.4 Price Testing — the Highest-Leverage Test Available

Price testing is under-utilized relative to its leverage, because most operators (correctly) hesitate to show different prices to different customers without a compliant framework.

A $5 price increase on a $45 product, with no measurable CVR impact,
is an 11% revenue increase on every future unit sold at the new price.

Why this outranks almost every other test on the priority table: most CRO tests optimize conversion rate at a fixed price. A successful price test raises revenue per converted customer — and because margin, not just revenue, is what a compressed-margin 2026 dropship business actually needs (LUCE_03, LUCE_09), a price test that holds CVR steady while raising price is closer to pure margin improvement than almost any copy or layout test can achieve.

How to run it compliantly: true price testing (different visitors see genuinely different prices, split randomly and consistently) requires a tool built for it — Intelligems is the commonly cited Shopify-native option. Do not manually change your product's list price back and forth to "test" reaction; that's not a controlled test, it's guessing with extra steps, and it risks confusing returning customers who see inconsistent pricing.

---

#PART 6: MOBILE OPTIMIZATION AT DEPTH

#6.1 The Mobile-First Reality, Recalibrated

65–75% of ecommerce traffic is mobile in 2026 — the original module's "60–80%" range narrows and confirms at the high end. Most brands still build desktop-first and adapt down, which is backwards given the traffic mix.

The gap, in 2026 numbers: mobile converts 1.8–2.5%, desktop converts 3.5–4%. That's mobile running at roughly 51–71% of the desktop rate — a narrower framing than the original module's "30–50% lag" language, but applied to a larger share of total traffic, which means the absolute revenue opportunity in closing the gap is larger, not smaller, than the original module implied.

Mobile-specific CRO checklist, engineering detail:

Speed:

• Mobile PageSpeed score >70 (Google PageSpeed Insights)
• LCP <2.5 seconds on mobile (Part 6.3)
• Images served in WebP format — smaller file size, equivalent visual quality
• Videos lazy-loaded, never auto-playing on initial page load
• Theme not loading unnecessary JavaScript (Part 6.3's app-tax discipline)

Usability:

• CTA buttons tap-friendly (minimum 44×44px tap target)
• Text readable without zooming (minimum 16px body font)
• No horizontal scroll
• Sticky "Add to Cart" bar always visible as the user scrolls (Section 6.2)
• Product images in a swipeable carousel
• Checkout auto-fills correct keyboard type per field (numeric for phone, email keyboard for email)

Checkout:

• Apple Pay and Google Pay enabled and prominently displayed (Section 3.3)
• Shop Pay enabled — fastest checkout, one-tap for returning Shopify customers
• Checkout tested on both iOS Safari and Chrome for Android — the two environments render differently often enough to matter

---

#6.2 The Sticky ATC Bar

One of the highest-impact mobile CRO elements is a sticky "Add to Cart" bar that appears once the user scrolls past the main CTA button — carrying product name, price, variant selector, and the ATC button itself in a persistent footer strip.

Implementation: most modern Shopify OS 2.0 themes have this built in; enable it in theme settings before reaching for a third-party app. If your theme doesn't include it natively, a dedicated app (Sticky Add to Cart by Candy Rack, qikify Sticky ATC, or equivalent) fills the gap.

Average CVR lift when implemented correctly on long-content PDPs: 10–20% on mobile. The mechanism is straightforward — on a long product page, the primary CTA scrolls out of reach quickly on a small screen, and every scroll a visitor makes without a visible path to purchase is an opportunity to lose them to distraction or exit.

---

#6.3 The Page-Speed Budget — the Engineering Detail

LUCE_08 Section 5.2 gives you the operator-level speed budget and audit tool. Here's the engineering detail behind why each guardrail exists.

The underlying research, unchanged by any 2026 platform shift: every 100ms of added mobile load time measurably reduces conversion; sites loading in ~1 second convert meaningfully better than sites loading in ~5 seconds. This is page-speed physics, not a seasonal benchmark — it doesn't get revised by tariff policy or ad-platform algorithm updates.

Mobile LCP (Largest Contentful Paint) targets:

The most common Shopify speed killers, in order of frequency:

1. App JavaScript accumulation — each installed app typically loads its own JS bundle; 15–20 apps compounds into a meaningfully slower page even if each individual app seems lightweight
2. Unoptimized images — a multi-megabyte hero image served uncompressed to a mobile connection is still the single most common individual speed killer found in audits
3. Custom font loading — Google Fonts loaded without preload hints block render; limit to 2 font families and add preload hints
4. Third-party tracking pixels firing at page load, blocking initial paint instead of firing after
5. Autoplay video above the fold on mobile

The app-audit rule, restated with the underlying mechanism: remove any app unused in the last 30 days. An installed-but-unused app still loads its JavaScript on every page view — you're paying a speed tax for a feature nobody is using.

Theme choice matters, but less than app discipline: Shopify's free flagship theme (Horizon as of the 2025 Editions cycle, succeeding Dawn — verify the current default at themes.shopify.com) ships fast out of the box. A premium theme purchase ($350-class, Prestige/Impulse-tier) buys more design control, not automatically more speed — speed is overwhelmingly determined by app count and image discipline, not theme price point.

Speed audit tool: Google PageSpeed Insights (pagespeed.web.dev), run against the actual product page URL your traffic lands on — not the homepage, which is rarely the entry point for paid or AI-referred traffic.

---

#PART 7: QUALITATIVE CRO RESEARCH

#7.1 Why Qualitative Research Outperforms Guessing

Every A/B test needs a hypothesis. Hypotheses grounded in data and direct customer insight win far more often than hypotheses grounded in generic "best practice" advice. Qualitative research is how elite operators generate winning hypotheses before spending traffic to test them.

#7.2 The 5 Qualitative Research Methods

1. Session recording analysis (Hotjar / Microsoft Clarity / Lucky Orange)

Watch real users navigate your site. Look specifically for:

• Rage clicks — users clicking on non-clickable elements, meaning your design is confusing them about what's interactive
• Scroll depth — how far down the page do visitors actually get before dropping off
• Checkout drop-off points, stage by stage (Part 3.1)
• Form confusion — users starting a field, then abandoning it

Time investment: roughly 2 hours/month reviewing recordings. Insight quality: very high — you will see specific problems you'd never think to hypothesize, let alone test.

2. Heatmap analysis

• Click heatmaps: are users clicking on non-CTA elements, revealing a design that misleads about interactivity?
• Scroll heatmaps: where does the majority stop scrolling? If 60% drop off before your reviews section, move reviews higher (a structural change, testable per Part 5.1's Tier 1 traffic requirement)
• Move heatmaps (desktop): where do cursors hover, indicating reading focus?

3. Exit-intent survey — the checkout-specific version is Section 3.5; the same method applies site-wide on any high-value exit point.

4. Post-purchase survey — survey customers 3–7 days after purchase. Key questions:

• "What almost stopped you from buying?"
• "What convinced you to buy?"
• "How did you find us?"
• "What other options did you consider?" — this question specifically surfaces the Amazon/Temu comparison data feeding Section 2.3

This is the highest-value qualitative source available: it reveals your actual value proposition from the customer's perspective, which is frequently different from what you assumed it was, and the real objections you need to address in copy and page structure.

5. The 5-second test — show your homepage or PDP to someone who hasn't seen it, for exactly 5 seconds, then ask: "What does this brand sell? Who is it for? Why would someone buy it?" If they can't answer clearly, your value proposition isn't clear enough — this test is humbling and consistently reveals major CRO opportunities that internal review misses because you already know the answers.

---

#PART 8: AI-ERA CRO AT DEPTH

#8.1 Agentic Commerce Readiness — the Technical Checklist

LUCE_08 Section 6.1 gives you the three-step operator hedge: enable Agentic Storefronts, keep schema clean, validate it. Here's what "clean schema" means technically.

The Level 0 requirement (Part 1.4): an AI shopping agent evaluates your product on structured data (JSON-LD Product schema), not on your visual page design. A page that looks complete to a human visitor can be functionally invisible or misleading to an agent if the schema is missing, incomplete, or stale.

Minimum required Product schema fields:

• name, description, image
• offers — price, priceCurrency, availability (must match real-time inventory, not a cached default)
• aggregateRating and review — if you're claiming a star rating on-page, the schema should reflect the same number, or the mismatch is a credibility signal against you
• shippingDetails — this is the schema-level equivalent of LUCE_08's above-the-fold shipping-speed promise; an agent parsing your page for "how fast does this arrive" reads this field, not your marketing copy

Validation, not assumption: run Google's Rich Results Test and Schema.org's validator against your live product pages, not just your theme's template. A theme can ship with schema markup that looks correct in code but breaks on your specific product data (missing a required field for a particular product type, for instance) — validate the actual rendered page, per SKU category, not the theme documentation.

Why stale data is worse than no data in agentic commerce: a human visitor who sees "in stock" on an actually out-of-stock item bounces or contacts support — an annoyance. An AI agent that transacts on stale "in stock" schema may complete a purchase your inventory can't fulfill, generating a cancellation, a refund, and a damaged trust signal with the platform surfacing your catalog. Treat schema accuracy as an operational discipline (tie it to your inventory-sync cadence), not a one-time setup task.

The scale of the opportunity, honestly stated: AI referrals are still roughly 1% of web traffic. This is not where most of your CRO effort should go in 2026. It is, however, growing 340%+ YoY and converting +42% better than non-AI traffic when it does arrive — which is exactly why it's a "free hedge, do it once, maintain it passively" item rather than a dedicated workstream. Don't let a vendor sell you an ongoing "AEO service" against a channel this small; the fact sheet is explicit that top-10 Google results appear in AI answers only about 8% of the time, meaning there's no reliable paid lever to pull here yet.

---

#8.2 AI-Drafted Copy Workflows at Scale — the QA Rubric

LUCE_08 Section 6.2 gives you the operator-level four-step workflow and review gate. At scale — drafting copy for a growing catalog, or iterating faster on Tier 1 test variants (Part 5.1) — the review gate needs a rubric, not just a checklist, so review quality doesn't degrade as volume increases.

The QA rubric — score each AI-drafted section 1–3 on four axes before publish:

Publish threshold: every axis must score 3 before a section goes live. A single 1 on any axis is an automatic hold, regardless of how strong the other three scores are — this mirrors the hard-gate logic from LUCE_03's regulatory filter: one failure kills the section, a strong average doesn't save it.

Batch workflow for scale:

1. Draft in batches by section type (all FAQ answers together, all
   benefit-bullet sets together) — reviewing similar content types
   consecutively catches pattern errors (e.g., a repeated inflated
   claim across multiple SKUs) that spot-checking individually misses.

2. Score every batch against the rubric before any section publishes.

3. Log the failure axis for anything that doesn't pass — over time,
   this log tells you which prompt inputs (Section from LUCE_08 6.2:
   spec sheet, mined reviews, shipping facts) are weakest, so you
   fix the input, not just the output, going forward.

4. Re-run the human review gate on any SKU whose underlying facts
   change (new landed cost, new guarantee terms, new supplier) —
   AI-drafted copy that was accurate at write-time can go stale
   silently if nobody re-checks it against updated facts.

---

#PART 9: THE CRO OPERATING CADENCE

#9.1 The Monthly CRO Ritual

Week 1 — Data review:

• Pull CVR by device, traffic source, and landing page in GA4 and Shopify Analytics
• Review session recordings (30 minutes)
• Check heatmaps for anomalies
• Review exit-survey and post-purchase-survey data (Part 7.2)
• Identify the top 3 funnel leaks, ranked by estimated revenue impact

Week 2 — Hypothesis generation:

• For each funnel leak, generate 1–3 testable hypotheses
• Prioritize by (expected impact x confidence level) / effort required
• Document in a test backlog: hypothesis, test type, expected impact, traffic required (Part 5.1/5.2 tables)

Weeks 3–4 — Test running:

• Implement the highest-priority test that clears the traffic floor (Part 5.2)
• Run for a minimum of 2 weeks to capture a full business cycle
• Do not peek at results daily and call a winner before statistical significance is reached

Monthly — Winner implementation and documentation:

• Document all test results, including losers — losing tests contain hypotheses worth invalidating and not re-testing
• Implement winners into the permanent theme
• Update the test backlog with new hypotheses generated from this cycle's learnings

---

#9.2 The CRO Stack — Tool Budget by Stage

Minimum viable CRO stack (under $200/month):

• Microsoft Clarity — free — session recordings, heatmaps
• Google Analytics 4 — free — funnel analytics, traffic segmentation
• Hotjar Basic — verify current pricing (historically ~$32/mo) — surveys, additional recordings
• A/B testing via Shopify theme duplication — free

Growth stack (~$200–500/month):

• Hotjar Business — unlimited recordings + surveys, verify current tier pricing
• Intelligems — true A/B testing with price testing, verify current tier pricing
• Okendo or Stamped.io — review platform with UGC, ~$99–199/mo historically
• Triple Whale — from $129/mo (fact-sheet verified) — attribution layer once you're past basic GA4 needs

Scale stack (~$500–1,500/month):

• VWO or Convert.com — enterprise A/B and multivariate testing
• Hotjar Scale — included at this tier
• Typeform for structured post-purchase surveys
• ReConvert or equivalent at its higher tier — post-purchase upsell optimization at volume

The sequencing discipline: don't buy the growth-stack tools while you're still below the traffic floor (Part 5.2) for meaningful A/B testing. A $500/month scale stack on a 5,000-visit store is paying for statistical infrastructure you can't yet feed enough data to use.

---

#DECISION TREES

#Tree 1 — Checkout funnel breakpoint diagnosis

START: Overall cart-to-purchase rate is below the 25-40% benchmark (Section 3.1).

STEP 1 - Split the funnel into its two stages.

  IF cart -> checkout initiation is below 40%
    -> Cart-page problem. Check:
       - Is the free-shipping threshold indicator present and accurate?
       - Is the cart upsell under the 30%-of-cart-value cap (over-cap
         upsells compete with, rather than complement, the purchase)?
       - Is the discount-code field hidden/collapsible, not prominent?
       -> Fix per Section 3.2, re-measure after 1-2 weeks.

  IF checkout initiation -> completion is below 45%
    -> Checkout-page problem. Check:
       - Are Shop Pay / Apple Pay / Google Pay enabled AND at the
         top of checkout, not buried?
       - Is a progress indicator present?
       - Run the exit survey (Section 3.5) for the actual stated reason.
       -> Fix per Section 3.3, re-measure after 1-2 weeks.

  IF both stages are individually within benchmark but the compound
  rate is still weak
    -> Recalculate the compound math (0.5 x 0.55 = 0.275, for instance,
       is a "passing" compound from two borderline-healthy stages) -
       treat borderline-healthy stages as a warning even if neither
       triggers its own threshold individually.

#Tree 2 — Which CRO tool tier do you actually need

START: You're deciding whether to buy a CRO/testing tool.

IF monthly visits < 10,000
  -> Minimum viable stack only (Section 9.2): Microsoft Clarity + GA4,
     both free. Do not buy Intelligems, VWO, or Convert.com yet -
     you cannot generate statistically valid tests to justify the cost
     (Part 5.2 traffic floor).

IF monthly visits 10,000-50,000
  -> Growth stack becomes justifiable IF you have a documented test
     backlog (Part 9.1, Week 2) with at least 2-3 Tier 1/Tier 2
     hypotheses ready to run. Buying the tool before the backlog
     exists is buying capacity you won't use.

IF monthly visits > 50,000 AND running 2+ concurrent tests
  -> Scale stack (VWO/Convert.com-tier) is justified. Below this
     traffic and concurrency level, the enterprise tools' main
     advantage - running multiple simultaneous multivariate tests -
     goes unused.

---

#KPI TABLE — TARGETS, WARNINGS, KILL SWITCHES

---

#THE 2026 REALITY LAYER

The statistical floor moved up, not down. The original module's traffic-requirement examples assumed baseline CVRs (1% and 3%) that don't match 2026's actual distribution. Because most new stores sit below 1.4% blended on cold traffic, the real traffic requirement for a valid Tier 1 test is higher than the original module's numbers suggested — Part 5.2's recalibrated table is the correction, and it changes the practical answer to "can I A/B test yet" for a meaningful share of readers from "yes, cautiously" to "not yet."

The vertical CVR benchmark tables needed splitting into two, not just updating. The original single table conflated site-wide and session-level CVR under one set of ranges. The fact sheet's category data (food, beauty, electronics, luxury) reports at the session/PDP level, which runs 2–3× the site-wide number — presenting one table as if it answered both questions would have quietly reintroduced a fantasy-baseline problem in a different form. Part 1.3's two tables exist specifically to prevent that.

Checkout's express-payment expectation hardened from "nice to have" to "assumed default." The original module's checklist framing ("what you can optimize") still applies mechanically, but the psychological stakes changed: a visitor arriving from a Temu or Amazon session has already experienced one-tap checkout as normal, which means its absence on your store now reads as a specific, noticeable deficiency rather than a missing enhancement.

Post-purchase upsell economics are proportionally more important under compressed 2026 margins. At generic-dropship net margins of 3–7%, the near-pure-margin post-purchase moment (Section 3.4) is one of the few places in the funnel where the margin math is genuinely favorable — worth disproportionate build effort relative to its complexity.

Agentic commerce added a Level 0 to the CRO hierarchy that didn't exist in the original module at all. Schema validity wasn't a CRO concern when the original was written because AI shopping agents weren't a meaningful traffic source. At ~1% of traffic and +340% YoY growth, it's still small — but it's now a real, distinct layer underneath even the most basic "does the page load" concern, because a broken schema can make an otherwise-perfect page invisible to a growing referral channel.

AI-drafted copy shifted the review-gate problem from "does this take too long" to "does this scale correctly." The original module didn't address AI drafting at all. At 2026's drafting speed, the constraint isn't time — it's maintaining review rigor at volume, which is why Part 8.2 replaces a simple checklist with a scored rubric: a checklist degrades under volume pressure, a rubric with a hard publish threshold does not.

---

#FAILURE MODES

---

#SOPs & CADENCES

Weekly:

1. Execute the current week of the Monthly CRO Ritual (Part 9.1) — data review, hypothesis generation, or test running, depending on which week of the cycle you're in.
2. Spot-check the checkout funnel split (cart→initiation, initiation→completion) against the KPI table.
3. Scan session recordings for any new friction pattern introduced by a recent change.

Monthly:

1. Complete the full four-week CRO ritual cycle (Part 9.1) and document every test result, including losers.
2. Re-validate product schema on any SKU with a pricing, inventory, or supplier change (Part 8.1).
3. Re-run the QA rubric (Part 8.2) on any AI-drafted section whose underlying facts changed.
4. Review the CRO tool stack against Decision Tree 2 — confirm you're not paying for a tier your traffic doesn't support yet, or under-investing once you've cleared a traffic threshold.
5. Review post-purchase and exit-survey data for any shift in the dominant stated objection (Part 7.2).

Quarterly:

1. Refresh the site-wide CVR benchmark comparison (Part 1.3) against your own trailing-90-day data — the tables in this module are a starting anchor, not a permanent target.
2. Reassess whether traffic has crossed a threshold (10,000, 50,000 monthly visits) that justifies a CRO tool-stack upgrade (Section 9.2, Decision Tree 2).
3. Review the full checkout funnel stage-by-stage math (Section 3.1) for drift, especially after any theme, app, or checkout-settings change.

---

#WEEK-1 ACTION PLAN

1. Day 1: Pull your last 90 days of GA4/Shopify Analytics data, segmented by device and traffic source. Map your numbers against Table A (Section 1.3) to establish your honest current-state baseline.
2. Day 2: Split your checkout funnel into its two stages (Section 3.1) using Shopify's checkout analytics. Identify which stage, if either, is below benchmark.
3. Day 3: Install or audit your qualitative research stack — Microsoft Clarity (free) at minimum. Review your last 20–30 session recordings for the specific friction patterns in Part 7.2.
4. Day 4: If you have an active customer base, launch a post-purchase survey (Part 7.2) with the four core questions. If pre-launch, review your LUCE_03 review-mining notes for the same objections.
5. Day 5: Validate your product schema against Google's Rich Results Test (Part 8.1) on your top 3 SKUs. Fix any errors before moving on.
6. Day 6: Build your test backlog document (Part 9.1, Week 2 format) with at least 3 hypotheses, even if you're below the traffic floor to test them yet — this becomes the foundation once traffic grows.
7. Day 7: Check your current traffic level against the Part 5.2 traffic-floor table. If you're below 10,000 monthly visits, commit explicitly to shipping documented best practice (LUCE_08) over formal testing for now, and set a calendar reminder to re-check this threshold monthly.

---

#SELF-TEST

1. A store's Shopify Analytics reports a 3.6% "conversion rate." Per Section 1.3, what question must you answer before comparing that number to the electronics category benchmark of "~3.6%" from the fact sheet?
2. A store's cart→checkout-initiation rate is 55% and checkout-initiation→completion rate is 50%. What's the compound overall cart-to-purchase rate, and does it clear the 25% benchmark from Section 3.1?
3. At a 1.4% baseline site-wide CVR, roughly how many total visitors are needed to validly detect a 10% relative lift in an A/B test, per the Section 5.2 table?
4. A store's AI-referred traffic is converting at roughly the same rate as its non-AI traffic, not the expected +42% premium. Per Part 8.1, what's the first thing to check?
5. Per Decision Tree 2, a store at 8,000 monthly visits is considering buying VWO. What should it do instead, and why?

<details> <summary>Answers</summary>

1. Whether the reported 3.6% is a site-wide figure (Table A) or a session/PDP-level figure (Table B) — the fact sheet's "~3.6%" for electronics is a session-level number, roughly 2–3× the site-wide figure the Table A electronics band (0.5–1.8%) would predict. Comparing a site-wide number against a session-level benchmark without reconciling the denominator produces a false read in either direction.
2. Compound rate = 0.55 × 0.50 = 27.5% — this clears the 25% floor from Section 3.1, but sits close enough to the threshold to flag as a warning-tier result worth monitoring, not a clean pass to ignore.
3. Roughly 28,000–30,000 total visitors (~14,000–15,000 per variant), per the Section 5.2 recalibrated table.
4. Check whether the product schema (Part 8.1) is stale — mismatched price, wrong availability status, or missing shippingDetails relative to the live page. Stale schema is the most common reason AI-referred traffic underperforms its expected conversion premium.
5. Stick with the minimum viable stack (Microsoft Clarity + GA4, both free) — per Decision Tree 2, 8,000 monthly visits is below the 10,000-visit floor needed to generate statistically valid tests, so a paid testing tool like VWO would be capacity without the traffic to use it.

</details>

---

#CROSS-REFERENCES

• → LUCE_08 (Store CRO): the core module this one extends. Read it first — it holds the 18-element product page, the shipping-speed promise, the operator-level trust architecture, and the $0 monthly audit this module's deeper science supports.
• → LUCE_03 (Product Selection): supplies the review-mining method (Section 2.2 there) that feeds hypothesis generation (Part 9.1) and AI-copy drafting inputs (Part 8.2) here.
• → LUCE_04 (Advertising): the advertorial funnel (Part 2.2) and landing-page-vs-PDP logic here directly inform paid-traffic landing page builds there.
• → LUCE_07 (Brand Building): supplies the defined brand voice the QA rubric (Part 8.2) checks AI-drafted copy against, and the deeper narrative architecture behind the Amazon/Temu comparison (Section 2.3).
• → LUCE_09 (Finance & Scaling): this module's next stop. Once CRO discipline is producing a stable, statistically testable funnel and a working post-purchase upsell model (Section 3.4), the next problem is unit economics and scaling infrastructure — covered there.
• → LUCE_20 (Email/SMS Advanced): consumes the email-traffic CVR benchmark (4–5.3%) and the micro-commitment ladder (Section 4.1) for list-growth and flow design.

---

LUCE — Launch. Unit Economics. Compound. Exit.

Next: → LUCE_09_Finance_Scaling.md — the financial architecture that keeps you solvent as you scale, and the unit economics every operator must master once conversion is no longer the bottleneck.