conversion-optimizer
Increase conversion rates — CRO audits, A/B testing, landing page optimization, and experimentation programs.
- License 许可
- MIT
- 最近更新
- 2026-11-07
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- conversion-optimizer
- category
- business-marketing
- description
- Increase conversion rates — CRO audits, A/B testing, landing page optimization, and experimentation programs.
03 SKILL.md 正文
Overview
Conversion rate optimization squeezes more value from existing traffic: systematic research, hypothesis-driven tests, and iterative improvements to pages and funnels. This skill covers CRO audits, building test hypotheses, running valid A/B tests, landing page best practices, and running an experimentation program.
CRO is research-led, not opinion-led. "I think the button should be green" is not a hypothesis.
When to use
-
Improving landing page conversion rates
-
Reducing checkout or signup abandonment
-
Building an A/B testing program
-
Diagnosing why traffic doesn't convert
-
Prioritizing website improvements
-
Fixing leaky funnels
-
Auditing a checkout flow for abandonment points
-
Prioritizing a CRO test backlog
-
Deciding between A/B testing and personalization
-
Improving trial-to-paid conversion
-
Optimizing mobile conversion specifically
-
Testing pricing page variations
Core concepts
The CRO process. Research → hypothesize → prioritize → test → analyze → implement/learn. Skipping research produces random tests that rarely win.
Research methods. Analytics (funnel drop-offs, heatmaps, session recordings), user feedback (surveys, polls, support tickets), usability testing (watch real users struggle), heuristic evaluation (expert review against best practices). Combine quantitative (what) with qualitative (why).
Hypothesis format. "Because [research insight], we believe changing [element] from [A] to [B] will cause [metric] to [change], measured by [primary metric]." Testable, grounded, specific.
A/B test validity. One variable at a time (mostly), adequate sample size (use a calculator — don't guess), full business cycles (run at least 1–2 weeks to cover day-of-week effects), statistical significance (95%+), and no peeking/stopping early.
Prioritization (PIE/ICE). Score ideas on Potential impact, Importance (page traffic/value), Ease of implementation. Test high PIE first.
Landing page anatomy. Headline (value prop in 5 seconds), subhead, hero visual, social proof (logos, testimonials, stats), benefits (not features), single clear CTA repeated, objection handling (FAQ, guarantees), minimal navigation/distraction.
Fogg Behavior Model. Behavior = Motivation × Ability × Prompt, all at the same moment. CRO maps to this: increase motivation (value clarity, urgency), increase ability (reduce friction, simplify), and place prompts well (CTAs at decision moments). If any factor is zero, no conversion — diagnose which one is missing.
The test-result repository. Every test documented: hypothesis, variant, results, learning, decision. This compounds into institutional knowledge — new team members learn years of insights in days, and failed tests prevent repeated mistakes.
Statistical fundamentals. Significance (95%+ confidence), power (enough sample size — calculate before, not after), and minimum detectable effect (what lift is worth detecting?). Peeking at results early and stopping at first significance inflates false positives — pre-commit to sample sizes. For low traffic, use sequential testing methods or go qualitative; underpowered tests are worse than no tests. Segmentation of results. Overall winners can hide segment losers: new vs. returning, mobile vs. desktop, traffic source. Always segment key tests — a change that lifts desktop 10% but tanks mobile 15% is not a win. Document segment findings; they become targeting rules for personalization later. Prioritization (PIE/ICE). Potential (how much improvement possible?) → Importance (how valuable is this page?) → Ease (how hard to implement?). Score ideas 1–10 on each; work top-down. Re-score quarterly — data from tests should update your priors.
Practical workflow
- Research. Pull funnel analytics: where do users drop? Watch session recordings of drop-off points. Survey exiting users ("what stopped you?"). Review support tickets for confusion themes.
- Build the hypothesis backlog. Convert insights into hypotheses in the standard format. Score with PIE. Keep 10–15 queued.
- Design tests. Wireframe the variant, write the copy, define primary metric + guardrails (e.g., "conversion up, average order value not down"). QA the test implementation thoroughly — broken tests waste weeks.
- Run properly. Calculate required sample size first. Run full weeks. Don't peek and stop early. Document everything.
- Analyze honestly. Significant win → implement and bank the learning. Inconclusive → was the hypothesis wrong or the test underpowered? Loss → what did we learn about users? Every test teaches something.
- Compound wins. Roll out winners, update the playbook, feed learnings into the next hypotheses. CRO compounds: small wins stack into large lifts.
Quick-win audit checklist: headline clarity (5-second test), CTA visibility and copy, form length (every field costs conversions), page speed on mobile, trust signals near the CTA, mobile layout, message match from ad/email.
Heuristic audit framework (per page): value proposition (clear in 5 seconds?) → relevance (matches traffic source?) → clarity (what to do next?) → friction (what makes action hard?) → anxiety (what creates doubt?) → distraction (what pulls attention away?) → urgency (why act now?). Score each; fix the worst two first.
Test documentation template: hypothesis (because [insight], changing [element] will [effect]) → primary metric → secondary/guardrail metrics → audience and traffic split → duration and sample size → results → decision (ship/iterate/kill) → learning. The learning field is the real output — even killed tests teach. Research-first testing: analytics review → heuristic audit → user testing (5 users) → surveys/polls → then test ideas. Tests born from research win 2–3x more often than tests born from opinions. Research is the highest-ROI CRO activity.
Common pitfalls
- Testing opinions. Random button-color tests without research. Research first, test second.
- Stopping tests early. Peeking at day 3 and declaring victory. Run to significance and full cycles.
- Underpowered tests. Low-traffic pages can't support A/B tests — use before/after or qualitative methods instead.
- Ignoring mobile. Testing desktop while most traffic is mobile. Optimize where the users are.
- No guardrail metrics. Increasing signups while destroying lead quality. Watch secondary metrics.
- One-and-done. Running three tests, declaring CRO "doesn't work." It's a program, not a project.
- Copying competitors' winners. Their audience isn't yours. Steal test ideas, not conclusions — validate yourself.
- Testing without traffic. Running A/B tests on pages with 100 visitors a month. Low-traffic pages need qualitative research or sequential testing, not split tests.
- Novelty effects. New variants winning briefly because they are new, then regressing. Run tests long enough for novelty to fade (2+ weeks minimum).
- HIPPO testing. Testing the highest-paid person's opinion instead of research-backed hypotheses. Data beats hierarchy.
- Ignoring qualitative. Numbers show what; user tests show why. Winners without understood whys do not generalize.
04 同仓库更多 Skills
web-scrape
Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open/fetch fails on JS-heavy pages, or when you need clean markdown/structured extraction from a site. Requires FIRECRAWL_API_KEY.
design-taste-frontend
tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML/CSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, "AI-made", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).