Skip to main content
skill CX Codex

seo-traffic-diagnosis — Claude Code skill: what it does

seo-traffic-diagnosis diagnoses drops, stalls, or sudden spikes in organic search traffic using Ahrefs MCP, Search Console, and site analytics.

MIT 290 installs ★ 1,107 updated 2026-10-19

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add rampstackco/claude-skills --skill seo-traffic-diagnosis -a claude-code
Open on GitHub

What seo-traffic-diagnosis does

This skill guides the agent through a five-layer root cause framework. It starts by ruling out analytics tracking bugs, weekday fluctuations, and seasonality by cross-checking Search Console clicks against organic sessions. It then isolates the issue across devices, countries, landing pages, and branded versus non-branded queries.

Next, the diagnosis inspects page-level metrics and technical factors. It checks SERP layout shifts, keyword rank changes via Ahrefs, canonical tags, robots.txt, redirect loops, and recent code deployments. Without active access to Ahrefs MCP, Google Search Console, and your analytics source, the skill cannot verify the root cause.

If internal checks pass, the workflow reviews external factors such as Google algorithm update dates and competitor rank movements. It turns findings into a single-sentence hypothesis backed by an actionable fix list.

What it helps with

  • Pinpoint why organic search sessions dropped after a site deploy
  • Investigate flat search traffic despite steady content production
  • Audit traffic loss following domain migrations or platform shifts
  • Assess SERP ranking losses after a search engine algorithm update

Example prompts for the agent

  • Organic traffic dropped by 30% after Tuesday's deploy, run a diagnosis across Search Console and analytics data
  • Blog search visits have stalled for three months, check non-branded query trends and SERP changes
  • A core ranking dropped on our key landing page, run page-level and technical diagnostics via Ahrefs MCP

How to install seo-traffic-diagnosis

1Skills CLI (Vercel)

npx skills add rampstackco/claude-skills --skill seo-traffic-diagnosis -a claude-code

2Manually: put the files in place

~/.claude/skills/seo-traffic-diagnosis/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/seo-traffic-diagnosis/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about seo-traffic-diagnosis

How to install seo-traffic-diagnosis in Claude Code?

Run npx skills add rampstackco/claude-skills --skill seo-traffic-diagnosis -a claude-code. The skill will be available at ~/.claude/skills/seo-traffic-diagnosis/SKILL.md.

What does seo-traffic-diagnosis do?

It applies a 5-layer methodology to isolate organic traffic anomalies, checking data accuracy, segmentation, page positions, technical deploy errors, and external algorithm shifts.

When should you avoid using this skill?

Do not use it for routine reporting, standalone backlink reviews, content gap discovery, or diagnosing broken analytics tracking tags.

Source text

SKILL.md · MIT
---
name: seo-traffic-diagnosis
description: "Diagnose organic traffic changes (drops, stalls, or unexpected wins) using Ahrefs MCP plus Search Console data. Use this skill when traffic suddenly dropped, has been flat despite investment, after an algorithm update, after a migration or deploy, or when a competitor seems to be taking share. Triggers on traffic dropped, traffic decline, traffic stalled, organic decline, lost rankings, why is traffic down, algorithm update, post-migration traffic loss, traffic diagnosis. Also triggers when stakeholders are panicking about an organic traffic number they cannot explain. A panic about tracking, event, or dashboard setup is a measurement problem, not a traffic problem: route those to `analytics-strategy`."
category: seo-audit-suite
catalog_summary: "Diagnose drops, stalls, or wins via 5-layer root cause analysis"
display_order: 5
---

# SEO Traffic Diagnosis

Diagnose why organic traffic moved (down, flat, or unexpectedly up) using Ahrefs MCP combined with Search Console and analytics data. Stack-agnostic. Produces a root-cause diagnosis and an action plan.

---

## When to use

- Organic traffic dropped sharply
- Organic traffic has been flat for months despite content investment
- After a known Google algorithm update
- After a migration, replatform, or domain change
- After a deploy that touched routing, redirects, or rendering
- When a competitor is visibly taking organic share
- When a single page dropped from a ranked position
- When stakeholders need an explanation, fast

## When NOT to use

- Routine performance reporting (use `analytics-strategy`)
- Pre-emptive content planning (use `seo-content-gap-audit`)
- Backlink-only investigations (use `seo-backlink-audit`)
- Technical issue triage outside of traffic concerns (use `seo-site-health-audit`)

---

## Required inputs

- Description of the symptom (what changed, when, magnitude)
- Date the change started (best estimate)
- Recent SEO history: deploys, migrations, content changes, link campaigns
- Access to Ahrefs MCP, Search Console, and analytics
- Confirmation the change is real (not a tracking artifact)

---

## The framework: 5 layers of diagnosis

A traffic change has one or more root causes. Move through the layers in order. Stop when you have enough evidence.

### Layer 1: Confirm the change is real

Before diagnosing, rule out:

- Tracking gaps (analytics outages, tag manager issues)
- Bot traffic changes
- Reporting comparison errors (different date ranges, wrong segment)
- Seasonality (compare year-over-year, not just month-over-month)
- Holiday or weekday effects

Cross-check Search Console clicks against analytics organic sessions. Significant divergence often points to a tracking issue, not a real traffic change.

### Layer 2: Localize the change

Where is the change happening?

Segment by:

- Country and language
- Device (mobile, desktop, tablet)
- Page or section (homepage, blog, product, category)
- Query type (branded vs non-branded)
- Landing page

A change in one segment requires different diagnosis than a change everywhere.

| Pattern | Likely cause |
| --- | --- |
| One country dropped | Local algorithm update, hreflang issue, geo redirect issue |
| Mobile dropped, desktop flat | Mobile usability or page speed regression |
| One section dropped | Topical algorithm update or section-specific quality issue |
| Branded queries dropped | Brand-level issue: site outage, reputation, manual action |
| Non-branded dropped | Algorithmic ranking issue |
| Single page dropped | Page-level issue: content, technical, or competitive |
| Sitewide dropped | Sitewide issue: penalty, technical, migration, or algorithm |

### Layer 3: Page-level analysis

For affected pages, audit:

- Position changes per ranked keyword (Ahrefs Rank Tracker history)
- SERP composition changes (more ads, AI overviews, featured snippets, video)
- Click-through rate changes
- Index status (Search Console coverage)
- Crawl errors and accessibility
- Recent content changes
- Internal link changes
- Backlink changes (lost links, redirect chains)

A page can lose traffic without losing rank if SERP composition changed.

### Layer 4: Technical analysis

Did anything break technically?

Check:

- Robots.txt changes
- Canonical tag changes
- Meta robots changes (accidental noindex)
- Redirect chains and loops
- Render issues (especially for JS-heavy frameworks)
- Site speed regressions
- Hreflang errors
- Sitemap freshness
- HTTP status codes (4xx, 5xx spikes)
- Server log evidence of crawl behavior changes

Recent deploys are the prime suspect. Compare deploy dates to traffic change dates.

### Layer 5: External analysis

If layers 1-4 do not explain the change, look outward.

- Algorithm update calendar (cross-reference timing)
- Competitor moves (new content, new SERP features they captured)
- Industry trend (declining search demand for the topic)
- Manual action (Search Console security and manual actions)
- Negative SEO (sudden link velocity changes)

External-factor diagnosis benefits from competitive context: did your traffic drop while competitors held steady (suggests an algorithm-specific issue), or did the entire vertical lose ground (suggests a user-behavior shift)? Similarweb shows competitor traffic trends; Ahrefs shows competitor SERP movement; pairing both surfaces whether the issue is yours alone or the category's.

---

## Workflow

1. **Confirm the symptom.** Get exact dates, magnitude, segment if known.
2. **Validate the data.** Layer 1 checks. Rule out tracking and seasonality.
3. **Localize.** Layer 2. Segment until the pattern is clear.
4. **Page-level dive.** Layer 3 on the most affected pages.
5. **Technical check.** Layer 4. Recent deploys, robots, canonicals, redirects.
6. **External check.** Layer 5. Algorithm updates, competitors, industry.
7. **Build the hypothesis.** State the cause as a single sentence.
8. **Validate the hypothesis.** Find the evidence that confirms or refutes it. See [`references/diagnosis-checklist.md`](references/diagnosis-checklist.md).
9. **Action plan.** Specific fixes mapped to specific evidence.
10. **Communicate.** Write up the diagnosis. Stakeholders want clarity, not exhaustive analysis.

---

## Failure patterns

- **Jumping to algorithm update.** "It must be the algorithm" is the lazy answer. Eliminate technical and page-level causes first.
- **Solving the wrong problem.** A drop diagnosed as "content quality" when the real cause was an accidental noindex on a deploy. Validate the hypothesis before fixing.
- **No baseline for "normal."** Without a baseline, every fluctuation looks alarming. Establish what normal noise looks like before reacting.
- **Treating one page as the site.** Site-wide and page-level diagnoses are different. Confirm scope first.
- **Ignoring branded vs non-branded.** A drop in branded queries means a brand-level problem. A drop in non-branded means an SEO problem. Different teams own them.
- **Comparing wrong date ranges.** Comparing 28 days to the previous 28 days during a holiday distorts the picture. Use year-over-year for seasonal businesses.
- **Stopping at correlation.** A deploy and a drop on the same day is a strong correlation, not proof. Find the mechanism.
- **Single-source diagnosis.** Ahrefs sees position. Search Console sees clicks and queries. Analytics sees behavior. Logs see crawl. Use them together.
- **Premature reassurance.** Telling stakeholders "it is just an algorithm update, will recover" without evidence sets up a worse conversation later.

---

## Output format

A diagnosis document with:

1. **Summary.** What changed, when, magnitude, root cause in one paragraph.
2. **The symptom.** Charts and segment breakdowns.
3. **Layer-by-layer findings.** What each layer ruled in or out.
4. **Root cause hypothesis.** Single statement with evidence.
5. **Action plan.** Ordered fixes with owners and timelines.
6. **Recovery forecast.** Realistic expectations on timeline and ceiling.
7. **Monitoring plan.** What to watch for confirmation of recovery.

Length: 4-10 pages. Stakeholders read this fast.

---

## If required data is unavailable

This skill's output depends on data, measurements, or tool results it cannot generate on its own. When a required input, tool, or data source is unavailable or unverifiable, the sanctioned output is the deliverable with the gap stated: what was needed, what was actually obtained or verified, and which parts of the output are affected. Fabricating, estimating, or interpolating a required number to complete the deliverable is never sanctioned. A stated gap is a complete answer.

---

## Reference files

- [`references/diagnosis-checklist.md`](references/diagnosis-checklist.md) - Layer-by-layer diagnostic checklist with the specific data to pull at each layer and how to interpret each signal.

© rampstackco · MIT · rampstackco/claude-skills

Similar skills

All skills for SEO and GEO →

From the same set102

The whole set →
content-brief-authoring The content-brief-authoring skill helps shape a brief for one piece of content, from search intent and audience to outline and success criteria. Use it to guide a writer or review an existing brief. CX Codex +1 292installs jtbd-framing jtbd-framing is a Claude Code skill that applies Jobs-to-be-Done to product work. It activates for JTBD analysis, prioritization, and product positioning. CX Codex +1 291installs email-sequences The email-sequences skill plans and drafts email campaigns, from onboarding flows and newsletters to transactional messages and win-back sequences. CX Codex +1 291installs seo-site-health-audit The seo-site-health-audit skill reranks crawler results (Ahrefs, similar tools) by actual impact on rankings instead of issue count. It helps you decide which technical fixes to tackle first when dev time is limited. CX Codex +1 291installs lead-magnet-design lead-magnet-design — a growth methodology skill to structure gated offers, checklists, and templates that qualify audience interest before collecting emails. CX Codex +1 289installs programmatic-seo Architect durable programmatic SEO systems from structured data without triggering search engine thin-content penalties. CX Codex +1 289installs seo-backlink-audit seo-backlink-audit reviews backlink quality via Ahrefs MCP: whether your profile is healthy, which links are risky, what you've lost, and how you stack against rivals. CX Codex +1 288installs seo-audit-orchestration seo-audit-orchestration runs a full SEO audit through Ahrefs MCP, coordinates focused checks, and turns their findings into one prioritized report. CX Codex +1 287installs

Project sponsors

Companies that keep this analytics open