{"id":"4c3a5b0a-43ff-4bdf-8305-1e070b9bc4d1","shortId":"TMMWz8","kind":"skill","title":"cro-optimization","tagline":"Run conversion rate optimization through hypothesis-driven testing including audit, hypothesis generation, test design, statistical analysis, and rollout decisions. Use this skill whenever the user wants to optimize conversion, run A/B tests, audit a funnel, generate test hypothe","description":"# CRO Optimization\n\nRun conversion rate optimization as a structured discipline: audit → hypothesize → test → decide. Stack-agnostic. Tool-agnostic.\n\nThis skill is for running tests against existing pages and flows. For writing landing page copy from scratch, use `landing-page-copy`. For setting up the analytics that make CRO possible, use `analytics-strategy`.\n\n---\n\n## When to use\n\n- Converting traffic at lower rate than expected\n- Specific funnel step has high drop-off\n- Pages with high traffic that could move the needle if optimized\n- A/B testing infrastructure exists (or can be set up)\n- Statistical significance and sample size questions\n\n## When NOT to use\n\n- Without sufficient traffic to test (under ~5,000 monthly conversions per variant)\n- Pre-launch (no users to test on yet)\n- Strategy or messaging-level questions that need qualitative research first\n- Brand-defining choices (CRO can't optimize a fundamentally wrong brand)\n\n---\n\n## Required inputs\n\n- The page or flow under optimization\n- Current conversion rate and traffic volume\n- Access to analytics (event tracking, funnel data)\n- An A/B testing tool (or willingness to set one up)\n- Time and budget for testing (typically 2 to 8 weeks per test)\n\n---\n\n## The framework: 4 phases\n\n### 1. Audit\n\nDiagnose before treating.\n\n**Quantitative audit:**\n\n- **Funnel data.** Where are users dropping off? The biggest drop is the biggest opportunity.\n- **Segmentation.** Does the funnel perform differently by source, device, geography, audience type?\n- **Performance data.** Are slow pages dragging conversions?\n- **Search Console / on-site search.** What are users looking for that they can't find?\n\n**Qualitative audit:**\n\n- **Session replay.** Watch 20+ sessions of users on the target flow. Note friction, confusion, hesitation.\n- **Heatmaps.** Where do users click? Where do they scroll? Where do they not?\n- **User interviews / surveys.** Why did users not convert? Survey people who started but abandoned.\n- **Form analytics.** Which fields cause abandonment? Which cause errors?\n- **Customer support tickets.** What conversion-related questions come in?\n\n**Heuristic audit:**\n\n- Apply CRO heuristics to the flow:\n  - Is the value proposition clear in 5 seconds?\n  - Is there a single primary CTA per page?\n  - Is the form length appropriate to the offer?\n  - Is the trust/social proof present?\n  - Are objections handled?\n  - Is the page accessible? (Accessibility issues hurt conversion silently.)\n\nThe audit produces a list of suspected friction points. Each becomes a hypothesis candidate.\n\n### 2. Hypothesis\n\nA testable statement.\n\n**Hypothesis structure:**\n\n> Because [observation from audit], we believe that [change] will produce [predicted outcome] for [user segment], because [reason].\n\n**Example:**\n\n> Because session replays show users abandoning at the shipping step (audit), we believe that adding visible shipping cost to the product page (change) will increase add-to-cart conversion by 5 percent (outcome) for desktop users (segment), because users are surprised by shipping cost and abandon (reason).\n\n**Hypothesis quality criteria:**\n\n- Specific change (not \"improve the design\")\n- Measurable outcome (with a target)\n- Grounded in evidence (audit, research, prior tests)\n- Tied to a known mechanism (why would this work?)\n\n**Hypothesis prioritization (ICE or PIE):**\n\n- **Impact:** How much could this move the metric?\n- **Confidence:** How likely is the hypothesis to be right?\n- **Ease:** How easy to test? (Time, complexity, risk)\n\nScore each 1 to 10. Highest combined scores test first.\n\n### 3. Test design\n\nA test that produces an unambiguous answer.\n\n**Sample size and duration:**\n\nUse a sample size calculator (most A/B tools have one) before launching. Inputs:\n\n- Baseline conversion rate\n- Minimum detectable effect (the smallest lift you'd care about)\n- Statistical power (typically 80%)\n- Significance level (typically 95%)\n\nThis produces required sample size per variant. Run the test until that sample is reached, OR for a minimum duration that captures full business cycle (typically 2 weeks minimum, to cover weekends and weekly patterns).\n\n**Common test setup mistakes:**\n\n- Stopping the test the moment significance is hit (peeking)\n- Running tests for too short to capture a full business cycle\n- Running multiple overlapping tests on the same flow\n- Testing during atypical periods (Black Friday, holidays, major campaigns)\n- Excluding mobile when 50%+ of traffic is mobile (or vice versa)\n- Testing on too small a slice of traffic (low statistical power)\n- Not segmenting analysis (overall lift can hide negative impact on a segment)\n\n**Test parameters to define before launch:**\n\n- Primary metric (one)\n- Guardrail metrics (do not go down)\n- Sample size\n- Duration (minimum and maximum)\n- Decision criteria (when to ship, when to kill, when to extend)\n- Segments to analyze in addition to overall\n\n### 4. Decide\n\nAfter the test concludes.\n\n**Decision framework:**\n\n| Outcome | Decision |\n|---|---|\n| Variant clearly wins (>95% significance, exceeds minimum effect) | Ship variant. Document. Continue testing. |\n| Variant clearly loses | Kill. Capture the lesson. Iterate hypothesis. |\n| Inconclusive (neither significant) | Larger test, different angle, or move on. Don't ship \"tied\" variants. |\n| Small lift, lots of variance | Probably not worth shipping. Even if \"winner,\" may not replicate. |\n| Wins overall, loses for important segment | Investigate segment. Consider segment-specific solution. |\n\n**Anti-patterns:**\n\n- \"It looks like it's winning, ship it\" before reaching significance\n- Shipping a variant because the team wants to (HiPPO - highest paid person's opinion)\n- Killing tests too early because they look bad\n- Re-running tests until they \"win\" (false positive risk)\n- Not capturing the learning when a test loses\n\n---\n\n## Statistical foundations\n\n### Significance and confidence\n\nA 95% significance level means: if there were truly no difference between variants, there's only a 5% chance you'd see results this extreme by chance.\n\nThat's not the same as \"95% chance the variant wins.\"\n\nMost CRO tools report Bayesian probabilities (\"95% chance of being best\"). Read the methodology your tool uses.\n\n### Sample size\n\nConversion testing needs more sample than people intuit. Quick reference:\n\n| Baseline rate | Minimum detectable effect | Sample per variant |\n|---|---|---|\n| 2% | 10% relative lift | ~75,000 |\n| 2% | 20% relative lift | ~19,000 |\n| 5% | 10% relative lift | ~30,000 |\n| 5% | 20% relative lift | ~7,500 |\n| 10% | 10% relative lift | ~14,000 |\n| 10% | 20% relative lift | ~3,500 |\n\n(Approximate. Use a calculator.)\n\nIf your monthly conversions per variant don't reach these numbers, A/B testing won't produce reliable results. Iterate via design and qualitative research instead.\n\n### Multiple testing\n\nThe more variants and metrics tested simultaneously, the more false positives. Adjust significance thresholds for multiple comparisons (Bonferroni or similar).\n\n---\n\n## Workflow\n\n1. **Audit.** Quantitative + qualitative + heuristic.\n2. **Generate hypotheses.** From audit findings. Apply hypothesis structure.\n3. **Prioritize.** ICE or PIE. Top 3 to 5 to test next.\n4. **Design the test.** Sample size, duration, primary and guardrail metrics, decision criteria.\n5. **Implement.** Build variants. QA carefully (broken variants invalidate tests).\n6. **Run.** Don't peek. Don't stop early.\n7. **Analyze.** Overall and by segment. Note interesting patterns regardless of significance.\n8. **Decide.** Ship, kill, or extend.\n9. **Document.** Hypothesis, design, results, decision, lesson.\n10. **Compound.** Apply lessons to next round of hypotheses.\n\n---\n\n## Failure patterns\n\n- **Testing without audit.** Random changes, random results.\n- **Vague hypotheses.** \"Make it better\" is not a hypothesis.\n- **Peeking and early stopping.** Bias toward false positives.\n- **Underpowered tests.** Not enough sample for a real conclusion.\n- **HiPPO override.** Highest paid person's opinion overrides the data.\n- **Testing during atypical periods.** Holidays distort results.\n- **Single metric obsession.** Conversion ups but average order value craters. Net loss.\n- **No guardrail metrics.** Testing for one outcome, missing damage to others.\n- **Documentation gap.** Wins captured, losses forgotten. Same hypothesis re-tested 3 times.\n- **Treating each test in isolation.** Compounding learning across tests is where CRO programs really win.\n\n---\n\n## Output format\n\nDefault output: a markdown test plan at `cro-test-[hypothesis-slug].md` per test. After the test runs, append the results section.\n\nStructure:\n\n```markdown\n# Test: [Hypothesis short name]\n\n## Hypothesis\nBecause [observation], we believe that [change] will produce [outcome] for [segment], because [reason].\n\n## Audit evidence\n[What evidence supports this hypothesis]\n\n## Test design\n- Primary metric:\n- Guardrail metrics:\n- Sample size required:\n- Duration: minimum X, maximum Y\n- Variant traffic split:\n- Segments to analyze:\n\n## Decision criteria\n- Ship if: [conditions]\n- Kill if: [conditions]\n- Extend if: [conditions]\n\n## Results (filled after test)\n- Sample reached:\n- Duration actual:\n- Primary metric: [variant vs control + significance]\n- Guardrail metrics: [results]\n- Segment analysis: [findings]\n\n## Decision\n[Ship / Kill / Extend / Iterate] - [Why]\n\n## Lesson\n[What this teaches us, regardless of outcome]\n```\n\n---\n\n## Reference files\n\n- [`references/hypothesis-library.md`](references/hypothesis-library.md) - Common high-impact hypothesis patterns by funnel stage.","tags":["cro","optimization","claude","skills","rampstackco","agent-skills","anthropic","awesome-claude-code","awesome-claude-prompts","awesome-claude-skills","claude-code","claude-skills"],"capabilities":["skill","source-rampstackco","skill-cro-optimization","topic-agent-skills","topic-anthropic","topic-awesome-claude-code","topic-awesome-claude-prompts","topic-awesome-claude-skills","topic-claude","topic-claude-code","topic-claude-skills","topic-good-first-issue","topic-mcp","topic-product-management","topic-seo"],"categories":["claude-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/rampstackco/claude-skills/cro-optimization","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add rampstackco/claude-skills","source_repo":"https://github.com/rampstackco/claude-skills","install_from":"skills.sh"}},"qualityScore":"0.540","qualityRationale":"deterministic score 0.54 from registry signals: · indexed on github topic:agent-skills · 181 github stars · SKILL.md body (9,583 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-18T18:55:14.792Z","embedding":null,"createdAt":"2026-04-30T01:01:27.851Z","updatedAt":"2026-05-18T18:55:14.792Z","lastSeenAt":"2026-05-18T18:55:14.792Z","tsv":"'000':154,974,980,986,998 '1':238,555,1057 '10':557,970,982,993,994,999,1140 '14':997 '19':979 '2':228,420,637,969,975,1062 '20':299,976,988,1000 '3':563,1003,1071,1077,1235 '30':985 '4':236,760,1083 '5':153,371,476,911,981,987,1079,1096 '50':690 '500':992,1004 '6':1106 '7':991,1115 '75':973 '8':230,1127 '80':606 '9':1133 '95':610,773,895,927,938 'a/b':35,128,213,583,1020 'abandon':337,343,450,491 'access':205,400,401 'across':1244 'actual':1343 'ad':459 'add':471 'add-to-cart':470 'addit':757 'adjust':1047 'agnost':59,62 'analysi':20,711,1354 'analyt':90,97,207,339 'analytics-strategi':96 'analyz':755,1116,1324 'angl':798 'answer':572 'anti':836 'anti-pattern':835 'append':1274 'appli':359,1068,1142 'appropri':385 'approxim':1005 'atyp':680,1196 'audienc':269 'audit':14,37,53,239,244,295,358,407,430,455,510,1058,1066,1153,1298 'averag':1207 'bad':870 'baselin':590,961 'bayesian':936 'becom':416 'believ':432,457,1288 'best':942 'better':1162 'bias':1171 'biggest':253,257 'black':682 'bonferroni':1053 'brand':180,190 'brand-defin':179 'broken':1102 'budget':224 'build':1098 'busi':634,668 'calcul':581,1008 'campaign':686 'candid':419 'captur':632,665,787,882,1227 'care':601,1101 'cart':473 'caus':342,345 'chanc':912,920,928,939 'chang':434,467,497,1155,1290 'choic':182 'clear':369,771,784 'click':315 'combin':559 'come':355 'common':646,1374 'comparison':1052 'complex':551 'compound':1141,1242 'conclud':765 'conclus':1183 'condit':1329,1332,1335 'confid':536,893 'confus':309 'consid':830 'consol':279 'continu':781 'control':1348 'convers':5,33,46,156,200,277,352,404,474,591,951,1012,1204 'conversion-rel':351 'convert':102,331 'copi':78,85 'cost':462,489 'could':122,531 'cover':641 'crater':1210 'criteria':495,743,1095,1326 'cro':2,43,93,183,360,933,1248,1262 'cro-optim':1 'cro-test':1261 'cta':378 'current':199 'custom':347 'cycl':635,669 'd':600,914 'damag':1221 'data':211,246,272,1193 'decid':56,761,1128 'decis':23,742,766,769,1094,1138,1325,1356 'default':1254 'defin':181,724 'design':18,501,565,1029,1084,1136,1306 'desktop':480 'detect':594,964 'devic':267 'diagnos':240 'differ':264,797,904 'disciplin':52 'distort':1199 'document':780,1134,1224 'drag':276 'driven':11 'drop':115,250,254 'drop-off':114 'durat':576,630,738,1089,1314,1342 'earli':866,1114,1169 'eas':545 'easi':547 'effect':595,777,965 'enough':1178 'error':346 'even':816 'event':208 'evid':509,1299,1301 'exampl':444 'exceed':775 'exclud':687 'exist':70,131 'expect':108 'extend':752,1132,1333,1359 'extrem':918 'failur':1149 'fals':878,1045,1173 'field':341 'file':1371 'fill':1337 'find':293,1067,1355 'first':178,562 'flow':73,196,306,364,677 'forgotten':1229 'form':338,383 'format':1253 'foundat':890 'framework':235,767 'friction':308,413 'friday':683 'full':633,667 'fundament':188 'funnel':39,110,210,245,262,1381 'gap':1225 'generat':16,40,1063 'geographi':268 'go':734 'ground':507 'guardrail':730,1092,1214,1309,1350 'handl':396 'heatmap':311 'hesit':310 'heurist':357,361,1061 'hide':715 'high':113,119,1376 'high-impact':1375 'highest':558,858,1186 'hippo':857,1184 'hit':657 'holiday':684,1198 'hurt':403 'hypoth':42 'hypothes':54,1064,1148,1159 'hypothesi':10,15,418,421,425,493,523,541,791,1069,1135,1166,1231,1265,1281,1284,1304,1378 'hypothesis-driven':9 'hypothesis-slug':1264 'ice':525,1073 'impact':528,717,1377 'implement':1097 'import':826 'improv':499 'includ':13 'inconclus':792 'increas':469 'infrastructur':130 'input':192,589 'instead':1033 'interest':1122 'interview':325 'intuit':958 'invalid':1104 'investig':828 'isol':1241 'issu':402 'iter':790,1027,1360 'kill':749,786,863,1130,1330,1358 'known':517 'land':76,83 'landing-page-copi':82 'larger':795 'launch':161,588,726 'learn':884,1243 'length':384 'lesson':789,1139,1143,1362 'level':172,608,897 'lift':598,713,808,972,978,984,990,996,1002 'like':538,840 'list':410 'look':287,839,869 'lose':785,824,888 'loss':1212,1228 'lot':809 'low':706 'lower':105 'major':685 'make':92,1160 'markdown':1257,1279 'maximum':741,1317 'may':819 'md':1267 'mean':898 'measur':502 'mechan':518 'messag':171 'messaging-level':170 'methodolog':945 'metric':535,728,731,1040,1093,1202,1215,1308,1310,1345,1351 'minimum':593,629,639,739,776,963,1315 'miss':1220 'mistak':649 'mobil':688,694 'moment':654 'month':155,1011 'move':123,533,800 'much':530 'multipl':671,1034,1051 'name':1283 'need':175,953 'needl':125 'negat':716 'neither':793 'net':1211 'next':1082,1145 'note':307,1121 'number':1019 'object':395 'observ':428,1286 'obsess':1203 'offer':388 'on-sit':280 'one':220,586,729,1218 'opinion':862,1190 'opportun':258 'optim':3,7,32,44,48,127,186,198 'order':1208 'other':1223 'outcom':438,478,503,768,1219,1293,1369 'output':1252,1255 'overal':712,759,823,1117 'overlap':672 'overrid':1185,1191 'page':71,77,84,117,194,275,380,399,466 'paid':859,1187 'paramet':722 'pattern':645,837,1123,1150,1379 'peek':658,1110,1167 'peopl':333,957 'per':157,232,379,616,967,1013,1268 'percent':477 'perform':263,271 'period':681,1197 'person':860,1188 'phase':237 'pie':527,1075 'plan':1259 'point':414 'posit':879,1046,1174 'possibl':94 'power':604,708 'pre':160 'pre-launch':159 'predict':437 'present':393 'primari':377,727,1090,1307,1344 'prior':512 'priorit':524,1072 'probabl':812,937 'produc':408,436,569,612,1024,1292 'product':465 'program':1249 'proof':392 'proposit':368 'qa':1100 'qualit':176,294,1031,1060 'qualiti':494 'quantit':243,1059 'question':142,173,354 'quick':959 'random':1154,1156 'rate':6,47,106,201,592,962 're':872,1233 're-run':871 're-test':1232 'reach':625,847,1017,1341 'read':943 'real':1182 'realli':1250 'reason':443,492,1297 'refer':960,1370 'references/hypothesis-library.md':1372,1373 'regardless':1124,1367 'relat':353,971,977,983,989,995,1001 'reliabl':1025 'replay':297,447 'replic':821 'report':935 'requir':191,613,1313 'research':177,511,1032 'result':916,1026,1137,1157,1200,1276,1336,1352 'right':544 'risk':552,880 'rollout':22 'round':1146 'run':4,34,45,67,618,659,670,873,1107,1273 'sampl':140,573,579,614,623,736,949,955,966,1087,1179,1311,1340 'score':553,560 'scratch':80 'scroll':319 'search':278,283 'second':372 'section':1277 'see':915 'segment':259,441,482,710,720,753,827,829,832,1120,1295,1322,1353 'segment-specif':831 'session':296,300,446 'set':87,135,219 'setup':648 'ship':453,461,488,746,778,804,815,844,849,1129,1327,1357 'short':663,1282 'show':448 'signific':138,607,655,774,794,848,891,896,1048,1126,1349 'silent':405 'similar':1055 'simultan':1042 'singl':376,1201 'site':282 'size':141,574,580,615,737,950,1088,1312 'skill':26,64 'skill-cro-optimization' 'slice':703 'slow':274 'slug':1266 'small':701,807 'smallest':597 'solut':834 'sourc':266 'source-rampstackco' 'specif':109,496,833 'split':1321 'stack':58 'stack-agnost':57 'stage':1382 'start':335 'statement':424 'statist':19,137,603,707,889 'step':111,454 'stop':650,1113,1170 'strategi':98,168 'structur':51,426,1070,1278 'suffici':148 'support':348,1302 'surpris':486 'survey':326,332 'suspect':412 'target':305,506 'teach':1365 'team':854 'test':12,17,36,41,55,68,129,151,165,214,226,233,513,549,561,564,567,620,647,652,660,673,678,698,721,764,782,796,864,874,887,952,1021,1035,1041,1081,1086,1105,1151,1176,1194,1216,1234,1239,1245,1258,1263,1269,1272,1280,1305,1339 'testabl':423 'threshold':1049 'ticket':349 'tie':514,805 'time':222,550,1236 'tool':61,215,584,934,947 'tool-agnost':60 'top':1076 'topic-agent-skills' 'topic-anthropic' 'topic-awesome-claude-code' 'topic-awesome-claude-prompts' 'topic-awesome-claude-skills' 'topic-claude' 'topic-claude-code' 'topic-claude-skills' 'topic-good-first-issue' 'topic-mcp' 'topic-product-management' 'topic-seo' 'toward':1172 'track':209 'traffic':103,120,149,203,692,705,1320 'treat':242,1237 'truli':902 'trust/social':391 'type':270 'typic':227,605,609,636 'unambigu':571 'underpow':1175 'up':1205 'us':1366 'use':24,81,95,101,146,577,948,1006 'user':29,163,249,286,302,314,324,329,440,449,481,484 'vagu':1158 'valu':367,1209 'varianc':811 'variant':158,617,770,779,783,806,851,906,930,968,1014,1038,1099,1103,1319,1346 'versa':697 'via':1028 'vice':696 'visibl':460 'volum':204 'vs':1347 'want':30,855 'watch':298 'week':231,638,644 'weekend':642 'whenev':27 'willing':217 'win':772,822,843,877,931,1226,1251 'winner':818 'without':147,1152 'won':1022 'work':522 'workflow':1056 'worth':814 'would':520 'write':75 'wrong':189 'x':1316 'y':1318 'yet':167","prices":[{"id":"f0a6681e-2aa8-4254-8007-0d8b5fa996ab","listingId":"4c3a5b0a-43ff-4bdf-8305-1e070b9bc4d1","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"rampstackco","category":"claude-skills","install_from":"skills.sh"},"createdAt":"2026-04-30T01:01:27.851Z"}],"sources":[{"listingId":"4c3a5b0a-43ff-4bdf-8305-1e070b9bc4d1","source":"github","sourceId":"rampstackco/claude-skills/cro-optimization","sourceUrl":"https://github.com/rampstackco/claude-skills/tree/main/skills/cro-optimization","isPrimary":false,"firstSeenAt":"2026-04-30T01:01:27.851Z","lastSeenAt":"2026-05-18T18:55:14.792Z"}],"details":{"listingId":"4c3a5b0a-43ff-4bdf-8305-1e070b9bc4d1","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"rampstackco","slug":"cro-optimization","github":{"repo":"rampstackco/claude-skills","stars":181,"topics":["agent-skills","anthropic","awesome-claude-code","awesome-claude-prompts","awesome-claude-skills","claude","claude-code","claude-skills","good-first-issue","mcp","product-management","seo","show-hn","showcase","showdev","web-design","web-development"],"license":"mit","html_url":"https://github.com/rampstackco/claude-skills","pushed_at":"2026-05-10T22:40:22Z","description":"Stack-agnostic Claude Skills covering the full website lifecycle: brand, design, content, SEO, dev, ops, growth, and research. Build, ship, audit, optimize.","skill_md_sha":"d1ea82f4c1574cbd671d519ffe3b950cc6e0d309","skill_md_path":"skills/cro-optimization/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/rampstackco/claude-skills/tree/main/skills/cro-optimization"},"layout":"multi","source":"github","category":"claude-skills","frontmatter":{"name":"cro-optimization","description":"Run conversion rate optimization through hypothesis-driven testing including audit, hypothesis generation, test design, statistical analysis, and rollout decisions. Use this skill whenever the user wants to optimize conversion, run A/B tests, audit a funnel, generate test hypotheses, design experiments, or analyze test results. Triggers on conversion optimization, CRO, A/B test, split test, multivariate test, hypothesis, conversion funnel, funnel audit, experiment design, statistical significance, lift, optimization. Also triggers when the user has a conversion problem and isn't sure where to start, or when test results are ambiguous and need interpretation."},"skills_sh_url":"https://skills.sh/rampstackco/claude-skills/cro-optimization"},"updatedAt":"2026-05-18T18:55:14.792Z"}}