Conversion Rate Optimization (CRO)

Turn existing traffic into more signups, leads, and purchases by fixing the page and the flow, not by buying more clicks. This skill covers the whole on-site journey: the hero that decides the bounce, the page structure and copy that build desire, the trust and friction levers, the checkout surgery, and the testing discipline that proves what actually worked.

The governing formula for every page here: Purchase Rate = Desire minus (Labor plus Confusion). Raise desire, cut steps, cut confusion.

When to use this

• A page gets traffic but converts poorly, or visitors bounce before scrolling.
• You are writing or rewriting a hero, landing page, product page, or checkout.
• Paid traffic lands on a generic homepage instead of a purpose-built page.
• Checkout or signup has a measurable drop-off between add-to-cart and purchase.
• A multi-step funnel leaks and you can only see where people click, not why they left.
• You have a backlog of test ideas and need to prioritize what to run next.
• Reviews, popups, or upsells exist but feel like passive widgets that do nothing.

When NOT to use this (reach for instead)

• The words inside the ad or the page (headlines, feature copy, CTA phrasing as a

craft): use copywriting-messaging. This skill tells you WHERE copy goes and whether it passes the conversion litmus test; that skill is the deeper writing craft.

• Why a tactic works on the human (loss aversion, goal-gradient, social proof,

scarcity psychology): marketing-psychology covers the underlying principles.

• Measuring whether a change actually moved the number (significance, event tracking,

attribution beyond a single A/B tool): analytics-data.

• Deciding IF paid traffic is your channel at all: strategy-fundamentals.
• Ecommerce-specific merchandising (bundles, promo framing, shipping, marketplace,

cart-recovery email): ecommerce. This skill owns the field-level checkout surgery and the belief-level funnel audit; ecommerce owns store construction and AOV.

How this works (decision path)

1. Diagnose before you build. If a whole funnel is leaking, run a belief-and-energy

audit first (funnel-audit-frameworks) to find the real cause, not the symptom. Fixing the wrong step wastes the traffic you already paid for.

1. Fix the hero first. The above-the-fold does roughly 90% of a landing page's

conversion work. Get the value prop, hook, persona fit, and CTA right (above-the-fold-formula) before touching anything below.

1. Lay the skeleton. Every high-converting page uses the same 7-section structure

with social proof placed next to CTAs (landing-page-structure). Match the page TYPE to the campaign feeding it (campaign-landing-page-types).

1. Make the copy descriptive, not clever. Run every headline and feature block

through the "do they know what you sell?" litmus test (landing-page-copy). For SaaS, show the real product UI (saas-visual-design).

1. Add trust and the right friction. Layer cheap trust cues and honest objection

handling, and classify each step as good or bad friction (trust-and-attention-cues, review-display-optimization, objection-smasher-pages, strategic-friction).

1. Fix the mid-funnel surfaces. Make PDP content discoverable, pin the two mobile

elements that convert, time popups, and personalize by segment (product-page-information-architecture, mobile-sticky-elements, popup-timing-segmentation, personalization-conversion-tactics).

1. Do checkout surgery. Cut fields (not steps), engineer "free" correctly, and stack

contextual upsells (checkout-field-optimization, free-offer-checkout-mechanics, upsell-cross-sell-tactics).

1. Prove it. A/B test only funnel-moving changes, run to a real sample, and log the

result (ab-testing-process). The winner becomes the new baseline.

The plays

Above-the-fold and hero

• Write a hero that sells in seconds with a value-prop then hook then persona formula

(10 to 30% lift when nailed)

Page structure, type, and copy

• Build from the fixed 7-element skeleton and audit with a 4-step review
• Match a purpose-built page to each campaign (7 types), and CTR-test the headline in

ads before you build the page

• Pass the "do they know what you sell?" litmus test with descriptive headers, not

slogans

• Always show the real product UI and enforce four visual-hierarchy rules for SaaS

Trust and strategic friction

• Add cheap trust and attention cues: DIY video testimonials, a visible form, and

averted-gaze imagery

• Engineer your review display: 5-star first, a 4.0 to 4.5 aggregate, kept honest

negatives

• Publish an objection-smasher page that names why customers churn or pick rivals
• Add the right friction on purpose: classify good vs bad friction and match it to

the business model

Product page and mobile

• Kill horizontal tabs and "Shop" dropdowns: make PDP content discoverable, not

compact

• Pin the two mobile elements that convert: sticky search and sticky Add-to-Cart

Popups and personalization

• Time and segment popups so they convert 10% instead of 3% (the 60% rule, audience

offers, intent gating)

• Personalize site, email, and ads with first-party data (10 to 15% revenue lift)

Checkout and offers

• Cut checkout fields, not steps (5 to 20%+ completion lift) with merging, jargon

fixes, and a goal-gradient progress bar

• Convert with "free" done right: $0 products through full checkout, refundable

deposits, and proving "it's free" at the button

• Stack contextual cross-sells and loss-aversion upsells at checkout

Diagnosis and testing

• Audit a funnel by belief and energy, not just clicks (Psych Energy, micro-beliefs,

Moments of Truth, the 7-question check)

• Run A/B tests that reach real significance: macro vs micro variants, a 4-factor

priority score, hard sample thresholds

Key numbers and benchmarks

• Above-the-fold does ~90% of a landing page's conversion work; nailing it lifts

conversion 10 to 30%.

• Checkout leaks 30 to 70% of revenue; 60 to 70% of abandonment happens in the first

15 seconds. Removing one field lifts completion 5 to 10%; removing three lifts it 20%+.

• A/B significance thresholds: 1,000+ visits to validate a 6.3%+ lift; 10,000+ visits

for a 2%+ lift. Target 70%+ "probability to be best" before shipping.

• Macro test variants (significant rethinks) carry 50 to 300% lift potential; micro

changes cap around 2%. The highest-value micro is rewriting the hero header.

• Reviews: leading with a 5-star review = 84% higher sales; target a 4.0 to 4.5

aggregate (5.0 reads fake); comparative positive reviews drive +26% sales; replying to reviews adds +12% submission volume; wait 10+ days to ask (+40 to 60% response).

• Popups average ~3% conversion but reach ~10% when timed to 60% of average

time-on-page, copy-matched, and segmented. 92% of first-timers don't buy but ~75% plan to return.

• Mobile: a sticky search bar drove +267% search clicks; a sticky Add-to-Cart drove

+78% ATC clicks. Keep combined sticky elements under half the screen height.

• Trust cues: video testimonials drive 25 to 50% lifts for 89% of enterprises; a

visible form drove 53.2% more calls than a no-form control.

• Personalization drives a 10 to 15% revenue lift (up to 25% for DTC). 71% of

consumers expect it; 76% are frustrated without it.

• PDP tab-blindness: 27% of Sephora and 43% of REI users never found content hidden

in unopened tabs.

• "It's free" microcopy drove an 8x CTA lift (Grammarly); a $50 refundable deposit

drove 100,000+ signups in a month (Slate). The biggest wins usually resolve one specific objection, not a full redesign.

Reference library

Every play above, in full.

Run A/B tests that reach real significance, macro vs. micro variants, a 4-factor priority score, and hard sample thresholds

The strategy

A disciplined 5-step A/B cycle: pick a change, split traffic, run to a statistically significant sample, read the probability-to-win, and log the result to inform future tests. Only test changes that move funnel metrics (email captures, signups, purchases, retention), never vanity metrics like button clicks.

When to use it

Continuously, year-round, once a page has enough traffic to reach significance. Prioritize when you have real conversion data and a backlog of ideas.

How to execute (steps)

1. Decide the change to test, sourcing ideas from: support/sales (customer objections), user surveys (loved features + concerns), best-performing ads (value props + imagery), competitor sites, on-site behavior via Hotjar/FullStory, and past test results.
2. Split traffic with a testing tool (Google Optimize, VWO, Optimizely): show half the change.
3. Run to a significant sample. Thresholds: 1,000+ visits to validate a 6.3%+ conversion lift; 10,000+ visits to validate a 2%+ lift. Target 70%+ "probability to be best" with sufficient sessions before implementing.
4. Read the report, likelihood of a significant effect on conversion.
5. Log design + result to inform the next experiment.

Notes / caveats / examples

• Macro vs. micro variants:
• Macro (significant rethinks) → 50-300% lift potential. Five approaches: mimic a competitor's page sections, rewrite for a different persona, cut the page in half for concision, take a competitive stance, or combine several micro-wins toward one goal.
• Micro (small changes) → ~2% at best. Levers: text, imagery, CTA, social proof, forms, section order, design, offers. The most impactful micro is rewriting the above-the-fold header/subheader.
• Prioritization (4 factors): Confidence, Impact, Implementation complexity, Uniqueness (not too similar to a past failed test). Brand consistency may outweigh a conversion gain.
• Test-design rules: run one A/B at a time (avoid device/session crossover); run variants in parallel, not sequentially (controls for traffic-source variation); consider testing new users only for a consistent familiarity baseline.
• Don't implement negligible wins, unknown downsides often exceed the small expected gain.
• Paid/low-intent traffic shows larger A/B returns than high-intent organic, retry an underperforming variant on paid before dismissing it.
• Schedule a recurring event (~every 2 months) to brainstorm macro variants so tests never stop.

→ Skill conversion note

Becomes an "A/B test planner" SKILL.md: takes a page + traffic volume, tells you the minimum detectable lift you can validate at that sample, scores candidate tests on the 4 factors, and outputs a prioritized macro/micro test queue.

Write an above-the-fold section that sells in seconds using a value-prop → hook → persona formula

The strategy

The above-the-fold (ATF) area, everything visible before scrolling, does roughly 90% of a landing page's conversion work. The ATF formula is a 3-step build: (1) state how users gain value, (2) add a hook, (3) speak to specific personas. Nailing ATF can lift conversion 10-30%.

When to use it

Any homepage, product page, or persona landing page. Highest leverage on pages fed by paid or cold traffic, where the visitor decides to scroll or bounce in the first few seconds.

How to execute (steps)

1. Find your value prop. Answer: what inferior alternative do people use without your product, and how are you better? Convert the answer into a concrete action statement. Be specific, no vague phrasing. (Duolingo: "short-form, interactive lessons" vs. traditional textbooks.)
2. Add a hook, one of two ways:

• Bold, specific claim that triggers "wow, I didn't know that was possible", but don't make it sound too good to be true or you trip the visitor's skepticism radar.
• Address the biggest buying objection directly in the header. Find it by surveying customers: "What almost stopped you from buying?" (Webflow expanded "Build your own website" to signal "without coding skills / without design training." HelloSign shows a product GIF to remove uncertainty.) Don't address every objection here, leave the rest for the body.

1. Speak to personas. List target personas, prioritize by ROI, and write in their language (not corporate speak). If you serve multiple major personas, add a "choose your own adventure" selector routing each to the right page section (Xeal Energy splits apartment vs. workplace owners).
2. Write the subheader, one or two sentences only, never an essay. Sentence 1: what the product is. Sentence 2: why the header's claim is believable / how it works. (Jupiter.co: first sentence names the category, second justifies "in just 5 minutes.")
3. Make the CTA a continuation of the header narrative. "Find food," "Start learning", not vague "Request a meeting." For complex/dense products, make the first CTA "Keep scrolling" and place the real conversion CTA further down after value lands.

Notes / caveats / examples

• Design rules: keep plenty of negative space so copy and visuals pop; use 2-4 nav links plus one CTA (fewer links = less analysis paralysis); show the product in action (GIF/demo), blur irrelevant text to keep focus. 3 Leaf Tea shows matcha in aesthetic packaging to trigger unboxing excitement.
• Don't be avant-garde, the page is a familiar medium for communicating what's unique about the product; save novelty for the product itself.
• Reported impact of doing this well: 10-30% conversion increase.

→ Skill conversion note

Becomes a "score & rewrite my above-the-fold" SKILL.md: takes a URL or hero copy, checks value-prop specificity / hook presence / persona fit / subheader length / CTA-narrative continuity, and outputs rewritten header + subheader + CTA variants.

Match a purpose-built landing page to each ad campaign (7 types), and test the copy in ads before you build the page

The strategy

Sending paid traffic to your homepage wastes it. Each campaign's intent and traffic temperature deserves its own landing-page type. There are seven, with a claimed conversion range of "5% to 500%" versus a generic homepage. Bonus: you can validate the winning headline in cheap ad tests before building the page.

When to use it

Any paid campaign (Meta, Google, TikTok) currently pointing at the homepage or one generic LP. Especially when scaling spend across multiple audiences/intents.

How to execute (steps)

1. Pick the LP type that matches the campaign:
2. Interactive calculator, quantify the user's problem/savings.
3. Competitor-comparison (e.g. Webflow vs Squarespace), honestly show where the rival wins, then where you win. Best for consideration-stage traffic.
4. Pre-sales / story page, warm cold traffic with narrative before the ask.
5. Free-trial / welcome-offer (e.g. AG1), for ready-to-try traffic.
6. Quiz funnel (e.g. ThirdLove bra fitting), for products needing personalization.
7. Advertorial blog, a long-form "growth guide" advertorial can drive leads for years.
8. Listicle with your tool ranked #1, best for retargeting.
9. Ad-test the copy first. Run headline/body variants in Meta Ads Manager optimized for CTR, ads reach statistical significance far faster than LP traffic, then port the winning message to the landing page.
10. Build one LP per active campaign, not one LP for all.

Notes / caveats / examples

• 7 types + the "5% to 500%" claim, plus the ad-tested-copy method. Comparison example = Webflow vs Squarespace; quiz example = ThirdLove.
• Complements (doesn't replace) the landing-page-structure and above-the-fold cards, those govern how each page is built; this governs which page to build.

→ Skill conversion note

Skill candidate: "Pick + spec my campaign landing page." Input = campaign audience + intent + traffic temperature. Output = the recommended LP type, a section skeleton for it, and 3 ad-headline variants to CTR-test before building.

Cut checkout fields (not steps) to lift completion 5-20%+, field merging, jargon fixes, and goal-gradient progress

The strategy

Most checkout revenue leaks come from field friction, not from having "too many steps." The industry loses 30-70% of revenue at checkout, and 60-70% of abandonment happens in the first 15 seconds. Removing a single field lifts completion 5-10%; removing three lifts it 20%+. Reduce fields, clarify labels, and show progress.

When to use it

Any ecommerce or signup checkout with a measurable add-to-cart → purchase drop-off. Pair with the ecommerce cart-abandonment card (which covers shipping cost, forced accounts, and email recovery); this card is the field-level surgery.

How to execute (steps)

1. Merge and hide fields. Combine First/Last into a single "Full Name"; hide Address Line 2, Company, and Coupon behind expandable links; drop Company/Phone/Title unless essential.
2. Auto-detect what you can. Populate city/state from ZIP; pre-check "billing = shipping."
3. Move account creation to the confirmation page. ~1 in 5 abandon on complexity, and forced accounts are a top-5 abandonment cause, let them buy first, create the account after.
4. Label required AND optional fields, and strip jargon. 85% of sites label only one of the two; ~50% of checkouts have poor field UX. Heatmap-test confusing labels like CID/CSC (the CVV code).
5. Add a goal-gradient progress bar. Show % complete / steps left ("Almost there! One more question") to pull users through, the closer people feel to done, the harder they push. (Shopify: iCart app.)
6. Nail the confirmation page. Restate the exact total, say "what happens next," and include a tracking link.

Notes / caveats / examples

• Benchmark stats: 5-10% completion lift per field removed, 20%+ for three, 30-70% checkout revenue loss, 60-70% of abandonment in the first 15s. Field tactics follow a "processing-fluency" checklist: nav <7 links, 8th-grade reading level, "clarity > cleverness". The goal-gradient progress bar is a separate lever.
• Principle: reduce interaction cost, not visible step count, a one-page checkout crammed with fields can convert worse than a clean multi-step one.

→ Skill conversion note

Skill candidate: "Audit my checkout for field bloat." Input = current field list + labels. Output = a merge/hide/auto-detect map, a required/optional labeling pass, a jargon rewrite, and a confirmation-page checklist, with the "move account creation to post-purchase" default.

Convert with "free" done right, $0 products through full checkout, refundable-deposit tripwires, and proving "it's free" at the button

The strategy

"Free" and low-commitment offers convert best when they're engineered as real commitment mechanics, not instant giveaways. Three moves: route free items through the full buying motion so users rehearse it, use a refundable deposit to capture commitment without financial risk, and prove "it's free" at the exact decision point.

When to use it

Any funnel with a free tier, lead magnet, freebie, waitlist, or free trial, where you want the free step to prime the eventual paid purchase rather than just harvest an email.

How to execute (steps)

1. List free resources as $0 products that go through the full checkout (create account → add to cart → check out), not one-click downloads. Users rehearse the exact buying motion, which smooths later paid conversion (Sara Renae Clark case).
2. Use a refundable deposit as a tripwire. A small, fully-refundable reservation fee captures real commitment with zero financial risk, a $50 reservation fee drove 100,000+ signups in one month (Slate). Best when the deposit mirrors the product's own philosophy.
3. Prove "free" at the button. Adding the words "it's free" to a CTA drove an 8x lift for Grammarly (a top-10 result across hundreds of tests, worth millions). The win came from removing a real user doubt at the decision point, not from changing many variables.

Notes / caveats / examples

• $0-product-through-checkout / Sara Renae Clark (issue #104); $50 refundable deposit → 100,000+ signups / Slate (issue #273); Grammarly "it's free" 8x CTA lift (issue #311).
• Lesson from Grammarly: the biggest lifts often come from resolving one specific objection, not from redesigning everything.

→ Skill conversion note

Skill candidate: "Engineer my free-offer funnel." Input = the free asset + eventual paid offer. Output = a full-checkout flow for the freebie, a refundable-deposit option where commitment matters, and CTA microcopy that proves "free" at the button.

Audit a funnel by belief and energy, not just clicks, Psych Energy, micro-beliefs, Moments of Truth, and the 7-question LP check

The strategy

CRO stalls when you only measure where people click. These four complementary frameworks audit a funnel by what the visitor feels and believes at each step, emotional energy, the specific belief each asset must earn, and the moments that decide the purchase, so you fix causes, not symptoms.

When to use it

Diagnosing a leaky multi-step funnel, planning a full-page or full-journey rewrite, or auditing a single landing page before ship. Use one framework or layer several.

How to execute (steps)

1. Psych Energy audit (Darius Contractor). Map emotional "energy" 0-100 through the funnel. Identify what adds energy (clean design, humor, bonuses, clear copy) vs what depletes it (form fields, slow pages). Sequence high-drain asks (email, payment) only after energy is high.
2. Micro-beliefs, one per asset. Map one product through a 4-stage funnel where each asset earns exactly ONE belief: Hook (name the pain) → Story (solution/differentiation) → Risk Removal (free trial, no card) → Payoff (deliver the promise). Worked example: Flowline.
3. Filter → Commit → Return. In the 3-second Filter scan, reuse structural patterns (nav/footer/buttons) but break content patterns (headline/visual/proof); lead with a familiar category then differentiate ("Like known product], but for audience]"). Pass three questions: what category is this, is it relevant, what would I learn by continuing?
4. Three Moments of Truth. Optimize ZMOT (research → win "best-of"/comparison listicles, Reddit/Quora answers), FMOT (purchase → distinctive packaging, social proof, value-based pricing), and SMOT (first experience → unboxing, onboarding, immediate visible effectiveness).
5. 7-question landing-page check. Grandma-clear product, named audience, legitimacy design, real-tweet social proof, upfront pricing, obvious support/FAQ, and CTA intensity matched to lead temperature.
6. Over-complexity pass. Remove email gatekeeping before the core offer, and consolidate "funnel stages" that perform the same job, timing and messaging usually beat more segmentation.

Notes / caveats / examples

• Psych Energy (issue #155, tools: Smart Recognition IDs ~40% of anon traffic); micro-beliefs/Flowline (#301); Filter→Commit→Return (#305); 3 Moments of Truth (#171); 7-question audit (#140); over-complexity checks (#303).
• These are diagnostic lenses, not layout rules, pair with the landing-page-structure and above-the-fold cards for the build.

→ Skill conversion note

Strong skill candidate: "Run a funnel-belief audit." Input = the funnel's asset list. Output = a Psych-Energy score per step, the one belief each asset should earn, the Moment-of-Truth it serves, and a prioritized fix list, a genuinely differentiated CRO skill.

Write landing-page copy that passes the "do they know what you sell?" litmus test, descriptive headers, not slogans

The strategy

Landing-page copy converts when it is descriptive and specific, not clever or corporate. The governing litmus test: if a visitor reads only this text, will they know exactly what you sell? Clear beats clever. Lead with the customer outcome, and answer objections where they naturally arise.

When to use it

Writing or rewriting any headline, subheader, feature block, or CTA on a landing page. Use it to kill slogan-style copy.

How to execute (steps)

1. Write descriptive headers, not slogans.

• Bad: "Improve your workflow!" / "Supercharge collaboration!"
• Good: "Visually design sites from scratch. No coding." / "Groceries delivered in 1 hour." / "Rent people's homes to experience cities like locals."

1. Header writing process: identify the value prop (what bad alternative do people use, how are you better → action statement), then add a hook, either a bold specific claim ("Faster" → "Saves 5 hours per week") or the biggest objection surfaced by surveying "What almost stopped you from buying?"
2. Subheader: max 1-2 sentences, first names what the product is, second explains why the header claim is possible. Keep reading breezy.
3. Write each of the 3-6 feature blocks with three parts:

• Feature header, blunt, specific value prop, not vague words like "Empower." Portable grill example: "Cooks and sears" / "No clean up" / "Cooks more than just meat."
• Feature paragraph, concise; address an objection only if it blocks purchase. For complex features, link to a detail page or use an expandable button (preferred, keeps users on page).
• Feature image, show the product in action (GIF for software; use cases + build-quality close-ups for physical goods); blur unimportant text. Each feature should call back to the hero's dominant value prop.

1. Make CTA copy a natural continuation of the hero narrative: "Find food," "Start learning", never a generic "Request a meeting."

Notes / caveats / examples

• Maximize specificity over corporate vagueness at every level of the page.
• Address objections where they naturally arise in the flow, not all at once in the header.
• For software feature imagery, avoid abstract art, show the actual product being used.

→ Skill conversion note

Becomes a "landing-page copy critic" SKILL.md: runs the litmus test on every headline/subhead/feature header, flags slogans and vague verbs, and rewrites them into descriptive, outcome-led copy with CTA-continuity checks.

Build a high-converting landing page from a fixed 7-element skeleton and audit it with a 4-step review

The strategy

Every high-converting landing page, homepage, persona page, or product page, uses the same skeleton and optimizes for one formula: Purchase Rate = Desire − (Labor + Confusion). Raise desire, cut steps, cut confusion. Layout should be proven and practical so the product's differentiation stands out, not the design.

When to use it

Building or rebuilding any landing page. Use the skeleton as a checklist and the review framework before shipping.

How to execute (steps)

1. Lay down the 7 required sections in order:
2. Navbar, logo + minimal links.
3. Hero, header, subheader, captivating imagery (see the above-the-fold-formula card).
4. Social proof, press or customer logos; creates FOMO ("elite club"). Startup tactic: give the product free to employees at recognized companies, then show their logos.
5. CTA, signup button with a concise incentive.
6. Features & objections, 3-6 feature blocks (see landing-page-copy card).
7. Repeat CTA, a second signup opportunity after value lands.
8. Footer, miscellaneous links.
9. Place social proof where decisions happen, next to CTAs, so it converts second-guessing into clicks ("buyers trust other buyers more than they trust you").
10. Design mobile-first. Pages built on desktop fall apart on a phone; cut page weight so slow loads don't bounce visitors before they convert.
11. Run the 4-step optimization loop (a proven agency process):

• Audit, line-by-line skeptical read + pressure-test the data for drop-off points.
• Diagnose the root cause: weak headline, buried offer, unanswered objections, or too many required clicks.
• Redesign around proven patterns and ship dev-ready, not slides.
• Test & scale, A/B the new page vs. old with real traffic; the winner becomes the new baseline for the next iteration.

Notes / caveats / examples

• Pre-ship review framework: get two reviewer types, people outside your market (test clarity/context) and people inside it (test uniqueness). Ask each for six things: Conversion (would you sign up, what's missing?), Interest (1-10, what to rewrite?), Clarity (what's unclear?), Expansion (what needs more detail?), Brevity (if cutting half, what goes?), Disbelief (what felt scammy?).
• Same skeleton reused for homepage (catch-all), persona pages (tailored), and product pages (functionality walkthrough).
• Client outcomes reported from this process: Snitcher ~4x revenue from paid in 12 months; WorkOS 9x qualified leads from Google; Videogen scaled to $493k/month in 90 days at 8x ad spend; FirmPilot 8x+ ROAS.

→ Skill conversion note

Becomes a "landing-page skeleton audit" SKILL.md: checks a page for the 7 sections in order, flags missing/misplaced social proof, and runs the 6-question review as generated critique prompts.

Pin the two mobile elements that move conversion, a sticky search bar and a sticky Add-to-Cart CTA

The strategy

On mobile, the two highest-leverage UI changes are keeping search and the primary action persistently visible while the user scrolls. Making users scroll back to find the search bar or the buy button is silent conversion loss on the single biggest commerce surface (US mobile commerce ≈ $387B/yr).

When to use it

Any mobile ecommerce homepage (search) or product page (Add-to-Cart). Highest impact on catalogs where users browse a lot before deciding.

How to execute (steps)

1. Add a fixed/sticky search bar to the mobile homepage. Calvin Klein HK saw +267% search-bar clicks and +19% of users reaching the results page.
2. Fix a sticky Add-to-Cart CTA on mobile product pages so it stays visible as the user scrolls specs, reviews, and images. Baume & Mercier lifted Add-to-Cart clicks +78%.
3. Keep a persistent basket/cart indicator (Sephora's persistent basket vs Missha's scroll-back friction).
4. Cap sticky footprint. Keep all sticky elements combined under half the screen height so they don't crowd the content.

Notes / caveats / examples

• +267% / +19% (Calvin Klein HK, issue #095); +78% Add-to-Cart (Baume & Mercier), $387B US mobile commerce, Sephora vs Missha (issue #121).
• Caveat: sticky elements eat viewport, test that the pinned bar doesn't hide price or the next section on small screens.

→ Skill conversion note

Skill candidate: "Mobile conversion quick-wins audit." Input = mobile homepage + PDP screenshots. Output = sticky-search and sticky-ATC recommendations with the <50%-screen guardrail and a persistent-basket check.

Publish an "objection smasher" page that names why customers churn or pick rivals, and win trust by being first to say it

The strategy

Prospects already have objections and are already comparing you to rivals. A dedicated page that names those objections and credible alternatives out loud disarms skepticism, captures comparison-stage search intent, and signals confidence, the brand willing to say where it loses reads as the trustworthy one.

When to use it

B2B SaaS and considered purchases where buyers research alternatives before committing, and where sales/support hear the same objections repeatedly. Pairs naturally with comparison/alternatives pages and retargeting.

How to execute (steps)

1. Harvest real objections from sales calls, support tickets, churn-survey answers, and win/loss notes. Rank by frequency.
2. Name why customers quit. Publish a candid "reasons customers leave" page (Drift's "4 Reasons Customers Quit Drift in 2017"). Address each reason with an honest response, a fix, a workaround, or "we're not for you if…".
3. Name credible alternatives. Reference the rival buyers are already weighing (Monday.com naming Salesforce as a credible alternative). Concede where they genuinely win, then show where you win.
4. Answer objections at the point of decision too, mirror the same objections near CTAs and pricing.
5. Optimize for search. These pages rank for "you] vs rival]", "you] alternatives", and "why leave you]" queries, high-intent comparison traffic.

Notes / caveats / examples

• Real examples: Drift's "4 Reasons Customers Quit Drift in 2017" and Monday.com naming Salesforce as a credible alternative.
• Caveat: only works if the concessions are honest, a fake objection ("our only flaw is we care too much") destroys the trust the tactic is meant to build.

→ Skill conversion note

Strong skill candidate: "Draft my objection-smasher page." Input = top objections + main rival + honest concessions. Output = a structured page (objection → honest response) plus target comparison keywords, directly reusable alongside God of Skills' competitor/alternatives page tooling.

Personalize site, email, and ads to lift conversion with first-party data

The strategy

Use collected first-party data to tailor on-site search, merchandising, email offers, and retargeting to each segment. Personalization drives a 10-15% revenue lift across businesses (up to 25% for DTC/online); 71% of consumers expect personalization while 76% get frustrated without it.

When to use it

Once you have profile/behavioral data and a way to act on it (ideally a CDP). Reserve heavier tactics (upsells) for proven high-value segments.

How to execute (steps)

• Customize on-site search by attributes (Nike returns women's running shoes using Run Club app + profile data).
• Show in-stock merch by preference: score purchase history by category (Tennis:5, Cycling:3, Golf:0), rank by interest, hide out-of-stock matches.
• Upsell/cross-sell only to high-AOV customers, and only when the current order is below their average (HelloFresh targets $150+ AOV customers with premium add-ons).
• Product-fit recommendations: collect body type/measurements at signup, combine with size charts, show "buy one size up for a relaxed fit" to cut return rates.
• Email: send hyperspecific offers by buying pattern (segment by frequency/timing, gender-targeted "Guys/Ladies Night Out").
• Ads: match retargeting creative to the exact content visited (golf-pain blog reader sees golf-specific ads, not generic); auto-build negative audiences to exclude existing users (exclude registered app users from download ads).

Notes / caveats / examples

• Spotify Wrapped (personalization at scale): 3B playlist streams and 1.2M tweets in month one (2019); +21% app downloads in the first week of December (2020).

→ Skill conversion note

A "personalization playbook builder" skill: given available data fields, output a prioritized list of site/email/ad personalization rules with segment definitions.

Time and segment popups so they convert 10% instead of 3%, the 60% rule, audience-based offers, and intent gating

The strategy

Popups average ~3% conversion but hit ~10% when they're timed to engagement, copy-matched to the page, and served to the right audience segment. The fix is never "add a popup", it's when it fires and who sees which offer.

When to use it

Any site running email-capture or offer popups, especially content pages, ecommerce PDPs, and returning-visitor flows. Deploy when a blunt sitewide popup is annoying users or under-converting.

How to execute (steps)

1. The 60% rule for timing. Fire the popup at 60% of a page's average time-on-page (e.g. 30s if the page average is 50s), not on immediate load. Match the popup copy to the page content.
2. Smart engagement triggers for content. Only fire after real engagement, e.g. 5+ minutes on page AND 40%+ scroll, and pair it with a lead magnet tied to that specific content. Gate the second half of long-form guides behind email.
3. Segment by audience (three-popup system):

• First-time visitors → discount / giveaway.
• Returning subscribers → subscriber-exclusive offer + countdown.
• Existing customers → new-product / welcome-back.

1. Intent-gate offers. Only fire an offer popup for returning visitors who meet a criterion (e.g. a $50+ basket before showing a free-shipping unlock). 92% of first-timers don't buy but ~75% plan to return, don't burn the offer on the first session.
2. Exit-intent as a recovery layer. On cart abandonment, trigger an exit-intent chatbot with an exclusive offer or low-stock nudge; escalate to a human if it can't close.

Notes / caveats / examples

• ~3% avg → ~10% timed-and-matched, plus the 92%/75% first-timer stats, come from aggregated popup CRO tests. Exit-intent chatbot example: Tidio (Casimir Rajnerowicz).
• Guardrail: engagement-gating beats time-gating on content; audience-gating beats one-size offers on commerce.

→ Skill conversion note

Skill candidate: "Design my popup logic." Input = page types + visitor segments + offer inventory. Output = a trigger table (60%-time / 5min+40%-scroll / exit-intent) mapped to segment-specific offers, ready to paste into an OptinMonster/Klaviyo-style rule set.

Kill horizontal tabs and "Shop" dropdowns, make product-page content discoverable instead of compact

The strategy

Compact page layouts hide the content that actually converts. Horizontal tabs, generic "Shop" dropdowns, and standalone FAQ pages all bury decision-driving information behind a click most users never make. The fix is a bias toward discoverability over compactness: expose content inline, expand the important stuff by default, and label everything clearly.

When to use it

Any product detail page (PDP) or site nav where key info (specs, sizing, FAQs, categories) is tucked into tabs, accordions, or dropdowns. Especially content-heavy considered purchases.

How to execute (steps)

1. Replace horizontal PDP tabs with vertically stacked collapsible sections, or a single long page + a sticky table of contents. 27% of Sephora and 43% of REI users never found content hidden in unopened tabs; 18-21% searched for it but still missed it.
2. Auto-expand critical sections (returns, key specs, sizing) instead of leaving everything collapsed.
3. Give sections clear multi-word titles ("Shipping & returns", not "More") so users can scan and find them.
4. Embed FAQs directly on the product page, ordered by journey stage, phrased as real customer Q&As, linking to supporting content, instead of a separate FAQ page.
5. Replace a generic "Shop" dropdown with explicit category links in the header; group many categories, and if you have one clear best-seller, link it directly.
6. Let the nav follow the business model. Low-AOV/simple products → push blog/info links to the footer; high-AOV/story-driven products → keep info in the navbar. Test question for any nav item: does it help convert or distract?

Notes / caveats / examples

• Sephora 27% / REI 43% tab-blindness plus the 18-21% missed-search stats come from usability research on hidden PDP content.
• Principle: every extra click to reach decision-critical content is a place users drop.

→ Skill conversion note

Skill candidate: "PDP information-architecture audit." Input = current PDP + nav structure. Output = a tabs→stacked-sections rewrite, an auto-expand list, inline-FAQ placement, and a nav keep/move recommendation based on AOV.

Engineer your review display, star ordering, comparative wording, incentive rules, and negative reviews you keep on purpose

The strategy

Reviews are not a passive widget, how you order, word, aggregate, and incentivize them changes sales measurably. The counterintuitive core: a perfectly clean 5.0 wall of praise converts worse than a mostly-positive page with a few honest negatives, because uniform praise reads as fake and any information beats none.

When to use it

Any product/PDP, marketplace listing, or SaaS pricing page that shows social proof. Highest leverage once you have 50+ reviews (a known conversion threshold) or when you're deciding whether to hide/show critical feedback.

How to execute (steps)

1. Show a 5-star review first. Leading with a 5★ vs a 1★ review = 84% higher sales. Order matters more than average.
2. Target a 4.0-4.5 aggregate, not 5.0. 4.9+ reads as manufactured. Keep minor, irrelevant 1-star reviews (e.g. "shipping box was dented"), irrelevant negatives lift credibility ~15%.
3. Feature slightly negative (4-star) reviews rather than hiding them; 93% of consumers say reviews affect purchase.
4. Prompt comparative wording. Pre-populate/seed review fields to elicit comparisons, comparative positive reviews drive +26% sales, and comparative negative reviews are 47% less harmful than absolute negatives.
5. Reply to every review (positive and negative): +12% review-submission volume and +0.12★ average.
6. Time the ask. Wait 10+ days after purchase/delivery to request a review (+40-60% likelihood of getting one).
7. Incentivize carefully, never for positivity. A sweepstakes entry lifted positive reviews +83.4%; even a flat $0.25 immediate incentive lifted positivity +20.6%. Never ask specifically for a positive review (violates platform policy + trust).
8. Publish ANY social proof now. A listing with negative reviews + poor photos still beats one with zero information, because any information reduces buyer uncertainty.

Notes / caveats / examples

• Science-backed review stats: +84% (5★-first), +26% (comparative), 47% less harmful (comparative negatives), +83.4% / +20.6% (incentives), +12% / +0.12★ (replies), +40-60% (10-day wait).
• Off-site option: push customers to review on Amazon/Trustpilot and feature that UGC on your site, since external reviews are trusted more.

→ Skill conversion note

Strong skill candidate: "Audit + rewrite my review section." Inputs = current reviews, average, ordering. Output = a re-ordered display (5★ first), an incentive/ask cadence (10-day trigger, $0.25 or sweepstakes, no positivity asks), and a reply template, with the target-4.3-average guardrail baked in.

Always show your product UI, enforce four hierarchy rules, and compete on "design care" now that AI raised the baseline

The strategy

For SaaS and digital products, the page's visual choices carry as much conversion weight as the copy. Three compounding rules: show the real product (hiding the UI implies it's bad), enforce a clear visual hierarchy so one message lands per section, and, now that AI-generated sites are the norm, invest in deliberate "design care" to stand out.

When to use it

SaaS homepages, product pages, and pricing pages, and any digital-product landing page where the current design leans on abstract illustration or looks template-default.

How to execute (steps)

1. Show real product UI, not abstract illustration. Use actual screenshots; add short 6-second looping videos and scrubbing/hover previews. Engagement ranking: images > GIFs > videos, so lead with crisp static UI. Use "Try a demo," not "Book a demo."
2. Apply four hierarchy rules per section: size = importance; contrast creates focus; one message per section; and always show the product UI. Start from Relume/Figma/Webflow/Framer templates and change as little as possible.
3. Add deliberate "design care" to differentiate. Once AI made a competent homepage trivial (35.3% of 2025's new sites are AI-generated), the edge is one deliberately distinctive element, a custom serif, a physical transition, a mascot, plus extending that care past the homepage into copy, email, and ads, and building reusable brand assets.

Notes / caveats / examples

• Design-care cases: Supermemory got a 40% conversion lift from a Claude-Opus homepage rebuild; TypingMind (Tony Dinh) reached $1M lifetime solo; Arc/The Browser Company's ~$610M Atlassian acquisition with ~50 people; Linear at $100M ARR.
• Caveat: "design care" ≠ decoration, the distinctive element must not fight the hierarchy or hide the product.

→ Skill conversion note

Skill candidate: "SaaS page visual audit." Input = current page. Output = a show-the-UI checklist, a per-section hierarchy pass (size/contrast/one-message), a "Try a demo" CTA fix, and one deliberate design-care suggestion.

Add the right friction on purpose, good vs bad friction, and why a slower flow can raise conversion and retention

The strategy

"Remove all friction" is wrong. Friction that increases perceived control, commitment, or comprehension raises conversion and retention; friction that adds cost or confusion kills it. The job is to classify each step as good or bad friction and match it to your business model.

When to use it

When auditing a signup/paywall/trial flow, or when a "frictionless" one-click flow is producing low-quality or low-commitment conversions (high refunds, fast churn). Complements the PLG friction-as-filter card (which is about gating/qualifying users).

How to execute (steps)

1. Classify every step as good or bad friction:

• Good: onboarding personalization, a well-placed checkout upsell, tooltips, an explanatory page that sets expectations, letting the user choose a tier.
• Bad: forcing a credit card up front, early push-notification permission prompts, mandatory account creation before purchase.

1. Match friction to your model. High-consideration/high-AOV products can afford (and benefit from) more upfront explanation and choice; low-consideration impulse products should strip it.
2. Use friction to raise commitment before the paywall. Add a short explanatory page that makes the value and the ask explicit rather than dropping users straight onto a payment screen.
3. A/B test the counterintuitive version. Test "add a step" against "remove a step", don't assume shorter wins.

Notes / caveats / examples

• Headspace added an explanatory page before the paywall (raised conversion). Duolingo removed the pre-selected 14-day trial. Peloton swapped a one-click trial for a dense info page + tier choice: friction raised perceived control and commitment.
• Caveat: friction must focus, not frustrate, every added step needs a hypothesis for why it raises commitment or clarity.

→ Skill conversion note

Skill candidate: "Classify friction in my funnel." Input = the current step list. Output = each step tagged good/bad friction with a keep/remove/add recommendation and a paired A/B test to run.

Cheap trust and attention cues that lift conversion, DIY video testimonials, a visible form, and averted-gaze imagery

The strategy

Three low-cost visual cues punch above their weight on conversion: real (even scrappy) video testimonials, a visible form as a "next step" signal, and imagery where models look toward your product/CTA to steer the viewer's eye. None require a big budget, just deliberate setup.

When to use it

Landing pages, product pages, and lead-gen pages that need more trust or a clearer focal point. The video-testimonial play is especially strong for higher-consideration B2B/enterprise offers.

How to execute (steps)

1. Produce DIY video testimonials. Pick your top-NPS / repeat customers, offer a small incentive, book a 20-30 min Zoom via a free Calendly, ask narrative questions (before/after, what almost stopped them), edit in iMovie with AudioJungle background music, and repurpose the clips across site, ads, and email. Video testimonials drive 25-50% conversion lifts for 89% of enterprises.
2. Add a form as a visual conversion cue. The mere presence of a form signals a next step, a law-firm test generated 53.2% more calls with a form present than a no-form control, even with the same primary CTA.
3. Use averted-gaze imagery. Models looking toward the product or CTA pull the viewer's attention there and make the creative more memorable (Katelyn Bourgoin). Direct eye contact holds attention on the face; a gaze toward the button moves it to the action.

Notes / caveats / examples

• Benchmarks: video testimonials drive 25-50% lifts for 89% of enterprises; form-as-cue drove 53.2% more calls; averted-gaze principle credited to Katelyn Bourgoin.
• Caveat: authenticity beats polish on testimonials, a real customer on a webcam outperforms a slick actor.

→ Skill conversion note

Skill candidate: "Trust-cue quick wins." Input = current page + customer list. Output = a video-testimonial shortlist + interview script, a form-placement recommendation, and an image-direction note (averted gaze toward the CTA).

Raise order value with contextual cross-sells and benefit-stacked checkout upsells that use loss aversion

The strategy

Generic "you may also like" widgets barely move order value. The upsells and cross-sells that work are specific and contextual, tied to what the customer is already buying, and the checkout stack is framed around benefits and loss aversion, not discounts.

When to use it

At the cart/checkout of any ecommerce or digital-product flow, and anywhere you show related-product recommendations. Best once you know real co-purchase patterns.

How to execute (steps)

1. Make cross-sells contextual and specific. "Would you like a blueberry muffin with your coffee?" beats a generic upsell, Second Cup staff doubled order values with named, relevant add-ons. Recommend one clearly complementary item, not a competing alternative.
2. Stack benefit-framed upsells at checkout. Present each add-on by the benefit it delivers, not the feature.
3. Auto-apply with removal friction (loss aversion). Pre-include the upsell so removing it feels like losing something, the default carries more than an opt-in ask.
4. Layer social proof on the upsell ("most buyers add this") at the decision point.
5. Pre-empt the specific objection the upsell raises (Remote OK pre-empts "will I actually get applicants?" right on the upsell).

Notes / caveats / examples

• Contextual cross-sell / Second Cup doubling order value; Remote OK checkout upsell stacking: auto-apply + removal friction + social proof + objection pre-empt.
• Complements the personalization card (reserve heavier upsells for proven high-AOV segments) and the ecommerce bundle/AOV card. Caveat: auto-applied add-ons must be genuinely easy to remove or they read as a dark pattern and drive refunds.

→ Skill conversion note

Skill candidate: "Design my checkout upsell stack." Input = product + top co-purchased items. Output = one contextual cross-sell line, a benefit-framed auto-applied upsell with a clear remove control, and a matched social-proof + objection line.

From God of Skills: a curated, hand-tested directory of AI skills, prompts, templates and image style guides. Source: https://godofskills.com/skills/conversion-cro?ref=claude-skill