seo-reporting-measurement
Skill14 starsChanged 7 months agoMITView on GitHub
Build reports that stakeholders trust, rooted in data sources that are actually reliable. Modern SEO measurement has to reckon with a very specific constraint: **the primary data source (GSC) is ~75% incomplete**, and the AI visibility layer is **probabilistic**, not deterministic. Building reports that ignore these realities produces false confidence.
In this repository28 other files it gives its agents.
- AGENTS.mdAGENTS.md
- ad-spend-allocatorSkill
- ai-search-visibility-aeo-geo-llmoSkill
- bofu-seo-aeo-strategySkill
- campaign-analyzerSkill
- content-effort-auditSkill
- google-ads-account-auditSkill
- google-ads-campaign-builderSkill
- google-ads-keyword-engineSkill
- google-ads-optimizerSkill
- google-ads-scriptsSkill
- google-merchant-centerSkill
- link-building-digital-prSkill
- local-seo-google-business-profileSkill
- meta-ads-account-auditSkill
- meta-ads-campaign-builderSkill
- meta-ads-creative-engineSkill
- meta-ads-optimizerSkill
- meta-ads-playbookSkill
- meta-ads-tracking-setupSkill
- on-page-seo-aeo-optimizationSkill
- paid-media-reporterSkill
- paid-search-ads-playbookSkill
- prompt-pickerSkill
- seo-aeo-content-strategySkill
- seo-keyword-research-intent-mappingSkill
- seo-max-ktSkill
- technical-seo-ai-crawler-auditSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 10 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈10
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
Build attribution: Default to multi-touch when possible.
- Distinctive: line 21
Handle algorithm updates: When Google updates or an AI model shifts, reports need to flag and contextualize — not just show numbers going …
- Distinctive: line 94
Document your model: Every report should state which attribution model is used and its known limitations.
- Very specific: line 98
When reporting AI visibility to stakeholders, every metric must include sample size and the measurement protocol.
- Distinctive: line 110
When Google rolls out a core update or an AI platform shifts behavior:
- Distinctive: line 113
Segment the impact: Don't report "traffic is down 12%." Break it down: which pages?
- Distinctive: line 115
Compare to industry: If traffic dropped 12% and the industry dropped 15%, you're relatively stable.
- Distinctive: line 203
Update weekly; review monthly in the structured report.
- Distinctive: line 293
Algorithm update impact takes 2-4 weeks to stabilize: Don't recommend reactive changes within the rollout window — wait for the dust to …
- Distinctive: line 296
Build for decisions, not dashboards: Every metric in every report should answer the question "what would we do differently if this number …
- Distinctive: line 303
Read these when the task warrants:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Why this matters
- GSC data is ~75% incomplete. Google filters approximately 75% of search impressions for "privacy" (industry analysis, Feb 2026). Single-source GSC decisions are unreliable.
- Multi-touch attribution reveals 30-60% more organic value than last-click models — which is why most SEO teams under-report their actual business contribution.
- GSC branded query filter (released November 2025) finally enables native branded vs non-branded segmentation — one of the biggest reporting improvements of the last five years.
- Median publisher: 10% YoY decline in organic traffic H1 2025 despite maintained visibility — this is the zero-click effect showing up in the numbers. Report on impressions and rankings, not just clicks.
- AI visibility is probabilistic. <1% chance of same brand list from same prompt; <0.1% chance of same order (multi-run replication studies). All AI metrics must be reported with sample sizes and confidence notes.
- Zero-click reality: 58.5% of US searches and 77.2% of mobile searches end without a click (third-party search studies, 2025). Click-based KPIs alone understate visibility.
Process
- Data foundation audit. Is GSC connected? Is GA4 set up correctly? Are conversions tracked? Without this, nothing else matters.
- Define the KPI hierarchy. Leading → core → business → AI visibility. Each layer has its own cadence and audience.
- Build attribution. Default to multi-touch when possible. Document the model and its known limitations.
- Structure the reports. Weekly tactical, monthly strategic, quarterly deep-dive. Each has a specific audience and purpose.
- Handle algorithm updates. When Google updates or an AI model shifts, reports need to flag and contextualize — not just show numbers going up or down.
- Report on AI visibility with appropriate caveats. Sample sizes, platform breakdowns, quarterly trends.
- Continuously refine. Retire metrics that don't drive decisions; add metrics that do.
Frameworks
Critical data caveat to lead every report with
⚠️ GSC impression data is approximately 75% incomplete. Google filters most search impressions for privacy. Single-source GSC decisions are unreliable. All figures below should be treated as directional trends, not absolute numbers. Cross-reference with third-party tools and focus on deltas over time, not point values.
This is not boilerplate — stakeholders consistently misinterpret SEO reports when they don't understand the data limits. Lead every report with this caveat.
KPI hierarchy
Organized from leading (early signals) to lagging (business outcomes). Each tier has different review cadence and different audiences.
Leading indicators (weekly review):
- Impressions (GSC)
- CTR by position band
- Scroll depth by landing page type
- Time on page by intent segment
- Indexed pages count (GSC Coverage)
- Core Web Vitals field scores (GSC)
- Crawl stats (GSC Crawl Stats report)
- New issues detected in GSC
Core SEO metrics (monthly review):
- Organic sessions segmented by intent (not aggregate) — informational/commercial/transactional
- Branded vs non-branded traffic — use GSC branded query filter (November 2025+)
- Ranking distribution — top 3, top 10, top 20 keyword counts
- New vs lost keywords — velocity of keyword portfolio
- SERP feature presence — featured snippets, PAA, AI Overviews
- Internal link coverage on priority pages
- Content freshness — how many priority pages updated in last quarter
Business metrics (monthly review, executive audience):
- Organic-attributed revenue (multi-touch model preferred)
- Lead generation from organic
- Conversion rate by landing page type
- Customer acquisition cost from organic
- Lifetime value from organic customers (if tracked)
AI visibility metrics (quarterly review minimum):
- Brand mention frequency across platforms (60-100+ runs per prompt)
- AI referral traffic (source/medium reports in GA4)
- Citation rate — how often your domain is cited across AI queries
- Competitor share-of-voice in AI
- Per-platform visibility — never collapse into a single score
Metrics to retire
Per industry reporting, these metrics should NOT be used as primary KPIs:
| Metric | Why retire it | What to use instead |
| Organic traffic as standalone KPI | Lacks intent context; treats all traffic as equal | Organic sessions segmented by intent |
| Average keyword position | Meaningless in aggregate (averaging ranks across different queries is statistically incoherent) | Ranking distribution buckets (top 3, top 10, top 20) |
| Domain authority as a business metric | A third-party proxy score, not a Google metric | Referring domain count and quality, actual ranking performance |
| Bounce rate in isolation | GA4 redefined it; often misinterpreted | Engagement rate + task completion signals |
| Single-run AI visibility "score" | Non-deterministic; meaningless without sample size | Visibility % across 60+ runs per prompt per platform |
| Individual meta description CTR | Google rewrites ~60% of meta descriptions; you're measuring Google's rewrite, not yours | Page-level CTR trends with content changes logged |
Revenue attribution formula
Monthly Organic Value =
(Organic Sessions × Conversion Rate × Average Order Value)
+
(Lead Gen Conversions × Lead Value × Close Rate)Important: This is a last-click formula. Multi-touch attribution typically reveals 30-60% more organic value than last-click. If you can, use GA4's data-driven attribution model or build a custom model that credits organic for its role in multi-touch journeys.
Document your model. Every report should state which attribution model is used and its known limitations. Stakeholders will ask why "SEO revenue" differs from "paid social revenue" — the model is the answer.
AI visibility measurement protocol (reporting layer)
When reporting AI visibility to stakeholders, every metric must include sample size and the measurement protocol. See references/ai-metrics-reporting.md for the detailed reporting patterns.
Core rules:
- Sample size minimum: 60 runs per prompt per platform. Anything less is noise.
- Never report "rank position" in AI output. <0.1% consistency — it's not a stable metric.
- Report per-platform, not aggregated. 86% of top-cited sources are platform-unique; aggregating hides the pattern.
- Quarterly cadence is the default; monthly if you're actively optimizing.
- Flag the confidence of every AI metric: Data-backed, Directional, or Experimental.
Algorithm update assessment framework
When Google rolls out a core update or an AI platform shifts behavior:
- Pause before reporting. Google core updates take 2-4 weeks to fully roll out. Initial ranking shifts are not final.
- Segment the impact. Don't report "traffic is down 12%." Break it down: which pages? which intent buckets? which query types?
- Cross-reference with known changes. Was there a recent migration? A new competitor? A CWV regression? Update impact gets blamed for unrelated issues constantly.
- Compare to industry. If traffic dropped 12% and the industry dropped 15%, you're relatively stable. If the industry is flat and you dropped 12%, it's specific to your site.
- Flag what you don't know. "Some of this may be algorithm-driven and some may be zero-click expansion from AI Overviews — it will take 30-60 days to distinguish."
- Recommend observation windows, not reactive changes. The worst thing post-update is to start rewriting content before the update finishes rolling out.
See references/algorithm-update-playbook.md for the full diagnostic playbook.
Report structure template
Adapt to the audience. Executives want the one-pager; SEO teams want the detail.
## SEO Performance Report — [Period]
### Executive Summary
[2-3 sentences: what happened this period, what it means for the business, what to do next. No jargon. No fluff.]
### Business Impact
| Metric | This Period | Prior Period | YoY | Target |
|--------|-------------|--------------|-----|--------|
| Organic Revenue | | | | |
| Organic Leads | | | | |
| Organic Sessions (non-branded) | | | | |
| Organic CAC | | | | |
⚠️ Attribution model: [last-click / multi-touch / data-driven]. GSC data ~75% filtered — trends more reliable than absolutes.
### Search Visibility
- Ranking distribution (top 3 / top 10 / top 20)
- Branded vs non-branded impressions and clicks
- SERP feature presence (featured snippets, PAA, AI Overviews)
- New vs lost keywords this period
### AI Search Visibility (if tracking)
| Platform | Visibility % | Δ vs last | Sample size |
|----------|-------------|-----------|-------------|
⚠️ AI metrics are probabilistic. Replication studies: <1% chance of same brand list from same prompt. All figures represent statistical patterns across [X] runs per prompt per platform. Focus on quarterly trends.
### Content Performance
- Top performing pages (by revenue/conversions, not just traffic)
- Decaying content (pages losing rankings or traffic)
- Content refresh priorities (decayed priority pages)
- New content performance (published this period)
### Technical Health
- Core Web Vitals (LCP, INP, CLS — field data)
- Indexation status (indexed vs discovered vs excluded)
- Critical issues (GSC errors, coverage issues, manual actions)
- AI crawler access status
### Competitive Landscape
- Key ranking movements in the competitive set
- Opportunities (pages where we gained position)
- Threats (pages where competitors gained)
### Recommended Actions (Prioritized)
| Priority | Action | Expected Impact | Effort | Timeline |
|----------|--------|----------------|--------|----------|
### Appendix: Methodology Notes
- Data sources: GSC, GA4, rank-tracking / backlink tool, [other]
- Attribution model in use
- AI measurement protocol and sample sizes
- Known limitations: GSC filtering, AI non-determinism, attribution caveats
- Period definitions: [date range, comparison period]Reporting cadence
| Cadence | Audience | Focus | Output |
| Weekly (tactical) | SEO team, content team | Rankings changes, traffic anomalies, technical issues, newly indexed pages | Dashboard + 5-bullet weekly note |
| Monthly (strategic) | Marketing leadership, department heads | Full performance across KPI hierarchy, content wins/losses, recommendations | Structured report (template above) |
| Quarterly (deep dive) | Executive team, board if applicable | Trend analysis, strategy review, competitive repositioning, AI visibility assessment, ROI analysis | Comprehensive presentation |
| Ad-hoc (incident) | Relevant stakeholders | Algorithm updates, major traffic shifts, migrations, crises | Incident report with diagnosis + plan |
Dashboard layer
For ongoing visibility, build a dashboard (Looker Studio, GA4 native, or a BI tool) with:
- Rolling 12-month trend for priority KPIs
- Branded vs non-branded segmentation
- Top pages by revenue (not just traffic)
- Top queries by click volume and by impression volume (two different lists)
- CWV field scores
- Referring domain velocity
- AI visibility panel with sample sizes surfaced
Update weekly; review monthly in the structured report.
Output format
See the report structure template above. For each report, customize the sections based on what the audience can act on. Executive reports drop the technical detail; SEO team reports drop the business framing.
Example — monthly SEO report for a mid-market SaaS
Period: March 2026 Audience: VP of Marketing + Head of Growth
Executive summary
Organic revenue grew 18% month-over-month, driven by 12 new product analytics cluster pages published in Q1 now ranking on page 1. AI visibility on ChatGPT rose from 12% to 19% after rewriting the pricing page to server-side render. Core Web Vitals regressed on mobile (LCP 3.8s vs 2.4s target) after a blog redesign — fix planned for sprint 14.
Business impact
| Metric | Mar 2026 | Feb 2026 | YoY | Target |
| Organic revenue (multi-touch) | $187K | $158K | +34% | $200K |
| Organic demo requests | 142 | 119 | +41% | 150 |
| Organic sessions (non-branded) | 89K | 82K | +22% | 100K |
⚠️ Multi-touch data-driven attribution via GA4. GSC impression data ~75% filtered — trend reliable; absolute impression count understated.
Search visibility
- Ranking distribution: 34 keywords in top 3 (+6), 127 in top 10 (+14), 310 in top 20 (+23)
- Branded traffic: 38% of organic sessions (stable)
- Non-branded traffic: 62% (up from 58% last month — the 12 new cluster pages are the primary driver)
- AI Overviews present on 28 of 50 priority keywords — we are cited in 7 of them
- New keywords (first rank): 47 (strong)
- Lost keywords: 12 (normal churn)
AI search visibility
| Platform | Visibility % | Δ vs Feb | Sample size |
| ChatGPT | 19% | +7% | 15 prompts × 60 runs |
| Perplexity | 24% | +2% | 15 prompts × 60 runs |
| AI Overviews | 14% | +6% | 15 prompts × 60 runs |
| AI Mode | 8% | +1% | 15 prompts × 60 runs |
⚠️ AI metrics are probabilistic. Replication studies: <1% chance of same brand list on same prompt. Figures represent patterns across 60 runs per prompt per platform. ChatGPT gain corresponds to pricing page SSR launch on March 11.
Content performance
- Top performers:
/product-analytics-guide(pillar, 8.2K sessions, 22 demos),/amplitude-vs-mixpanel-vs-posthog(4.8K sessions, 31 demos — highest converting),/event-scoping-framework(3.1K sessions, 18 demos) - Decaying: 3 old category pages losing rankings steadily — queued for Q2 refresh
- Refresh priorities:
/what-is-product-analytics(ranked #8, falling from #3 over 3 months)
Technical health
- CWV (mobile, field data): LCP 3.8s ⚠️ (target ≤2.5s), INP 180ms ✅, CLS 0.08 ✅
- LCP regression traced to blog redesign — new hero image treatment is 2.1MB unoptimized. Fix in sprint 14.
- Indexation: 924 indexed (+12), 18 new Coverage errors (all thin tag archives, queued for noindex)
- AI crawler access: GPTBot, PerplexityBot, ClaudeBot all hitting site normally (~40/day, ~25/day, ~10/day respectively)
Recommended actions
| Priority | Action | Impact | Effort | Timeline |
| Critical | Fix LCP regression (compress blog hero images, preload) | High — CWV is page experience signal | Low | Sprint 14 |
| High | Refresh /what-is-product-analytics with updated stats, ski ramp structure | High — recovery of top-3 position | Medium | April W2 |
| High | Publish Q2 benchmark data study | High — PR asset + linkable content | High | End of April |
| Medium | Server-render remaining 4 JS-heavy landing pages | Medium — unlocks AI visibility on those pages | Medium | Sprint 15 |
| Medium | Add FAQPage schema to top 10 commercial pages | Medium — rich result eligibility | Low | April W1 |
Appendix
- Data sources: GSC, GA4 (DDA attribution), rank-tracking tool, AI visibility tracker
- Attribution: GA4 data-driven; may differ from GA4 last-click by ~40% for organic
- AI measurement: 15 prompts × 60 runs across ChatGPT, Perplexity, AI Overviews, AI Mode. Run on 3/28-3/30. Re-measured after pricing page SSR launch.
- Known limitations: GSC ~75% impression filtering, AI probabilistic, CWV field data 28-day lag
- Period: March 1-31, 2026. Compared to February 1-28, 2026 and March 1-31, 2025.
Guidelines
- Lead every report with the GSC caveat. Stakeholders consistently misinterpret SEO data because they don't know it's ~75% filtered. State it every time.
- Always segment organic traffic by intent. Aggregate organic traffic is near-useless as a KPI — it conflates transactional wins with informational zero-click impressions.
- Use the GSC branded query filter (November 2025+) to separate branded and non-branded. This is one of the biggest analytics improvements in years and most teams still don't use it.
- Retire average keyword position. Averaging positions across different queries is statistically incoherent. Use ranking distribution (top 3, top 10, top 20) instead.
- Report AI visibility with sample sizes every time. "Visibility 18% across 60 runs" not "Visibility 18%." The second form invites false precision.
- Never collapse AI visibility into a single score. 86% of top-cited sources are platform-unique. A "unified AI score" hides the actionable signal.
- Multi-touch attribution reveals the real organic contribution. Last-click typically under-reports organic by 30-60%. If your stakeholders think SEO is underperforming, the attribution model is often the cause.
- Zero-click is the new normal. 58.5% US, 77.2% mobile searches end without a click (third-party search studies). Don't report declining clicks as declining performance if impressions and rankings are stable.
- Algorithm update impact takes 2-4 weeks to stabilize. Don't recommend reactive changes within the rollout window — wait for the dust to settle.
- Content decay is a leading indicator of traffic loss. A page dropping from #3 to #8 over 90 days predicts a traffic drop; catch it early through content performance tracking.
- Track referring domains and brand mentions together — AI visibility depends on both. Mentions without links still feed AI signal.
- Build for decisions, not dashboards. Every metric in every report should answer the question "what would we do differently if this number changed?" If nothing, remove the metric.
- The report is for the reader. Executives want the one-page summary with the one recommendation. SEO specialists want the detail. Build two versions if needed.
- Flag confidence levels on every recommendation. "Data-backed," "directional," or "experimental." This builds trust with stakeholders who will come back to hold you accountable.
- Reports are a trust instrument. Over-claiming on good months and hiding bad months destroys the discipline. Show both, contextualize both, and let the trend do the talking.
Reference files
Read these when the task warrants:
references/ai-metrics-reporting.md— Detailed guidance on reporting AI visibility metrics responsibly, including sample-size tables, caveat language, and common stakeholder pitfalls. Read when building an AI visibility section of a report.references/algorithm-update-playbook.md— Diagnostic playbook for responding to Google core updates, AI platform shifts, and traffic anomalies. Read when you suspect an algorithm-driven impact.
Cross-skill handoffs
- ← all other SEO skills — Receive baselines, KPIs, and priority actions for tracking.
- → seo-keyword-research — Flag keywords with declining performance for re-evaluation.
- → seo-content-strategy — Flag content with decay for refresh prioritization.
- → seo-technical-audit — Escalate CWV regressions, indexation issues, crawl anomalies.
- → aeo-ai-search-visibility — Feed visibility baseline data for strategy iteration.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

