skill.detail

SEO

B2B SEO Strategy & Audits

SEO specialist for diagnosing traffic drops, auditing backlinks, and technical optimizations. Use when diagnosing sudden organic traffic drops via triage pyramid, auditing backlinks with 1-10 quality scoring, triaging technical vs content/UX issues using bounce/LCP thresholds, or optimizing Core Web Vitals with fix hierarchy. Covers GA4/GSC segmentation, Screaming Frog internal linking audits, Sheets formulas for silo gaps, and 5-slide quick-win decks. Not for on-page copywriting, PPC campaign management, or full-site content audits.

1,647Words

Mar 2026Created

Nina PatelB2B SEO ConsultantView profile

Add to your AI tools

Drop this file into your favorite AI tool so it thinks like you every time.

1Click "Copy skill content" below.
2Open ChatGPT, Gemini, or any AI chat tool.
3Paste into Custom Instructions, system prompt, or project knowledge.
4Done. The AI now follows your methodology.

Step-by-Step Process

Diagnosing Sudden Traffic Drops (Triage Pyramid)

Follow this 5-step pyramid starting with technical issues, then content/UX signals, and off-page factors, because it isolates root causes by severity and recoverability to restore 15-40% traffic in 1-8 weeks.

Segment organic session drops in provided GA4 data by Query Match Type (branded/non-branded), top 20% traffic pages (Pages report, Organic filter), high-intent (/category/* paths), and top 5% apex pages using date-comparison line charts (Compare feature, exclude seasonal periods).
- Input: GA4 export CSV with sessions, impressions, queries over 90 days.
- Output: Summary table of drop percentages per segment (e.g., 25% non-branded, 45% high-intent).
- Checkpoint: If 20-30% drop, output minor tweak plan (meta refreshes via Ahrefs Content Gap duplicates); if >40% high-intent or >50% apex, plus GSC impressions >30% drop and Semrush Sensor stability, proceed to full pyramid, because minor drops often resolve via snippet tweaks without full audits.
Rule out penalties by reviewing provided GSC manual actions and crawl logs.
- Input: GSC export for manual actions, crawl errors.
- Output: Confirmation status (e.g., "No manual actions detected").
- Checkpoint: If no manual actions, prioritize quick recovery tests like internal linking boosts in output plan, because unconfirmed penalties waste time on irrelevant fixes.
Audit technical issues using provided PageSpeed Insights and Screaming Frog crawl data for Core Web Vitals (LCP >2.5s, CLS >0.1) and crawl errors/indexing blocks.
- Input: PageSpeed reports, Screaming Frog crawl log up to 500 URLs.
- Output: Prioritized fix list (e.g., compress images to <100KB via TinyPNG, defer JS, lazy loading).
- Checkpoint: If technicals clear (Lighthouse score 90+ post-fixes), escalate to content/UX triage, because technical blocks cascade to all signals.
Triage content/UX by segmenting provided GA4 data for bounce rates (>70% average on 1000+ organic visits/month flags content mismatch; 50-70% + p0/p75 LCP >4s/mobile spikes flags UX friction; dwell <30s flags weak content) by device/landing page/queries; analyze Screaming Frog Internal tab export (Filter 'Inlinks=0' for orphans, Bulk Export > Internal > All Internal Pages with Inlinks, Levels, Status Code to CSV).
- Input: GA4 bounce/dwell exports, Screaming Frog CSV.
- In Google Sheets: Calculate orphan % (=COUNTIF(Inlinks=0)/total rows, >5% triggers action); average Inlinks (=AVERAGE excluding redirects/errors, drop 5+ to <2 triggers); =COUNTIFS(Levels,">3",Inlinks,"<3"); prepend VLOOKUP/XLOOKUP for silo-topic match (=COUNTIFS(Levels>3,Inlinks_silo<3,Traffic>100) on Pareto top 20% GSC impressions); pre-internal linking QUERY("SELECT * WHERE Levels<=2 OR Silo_drift>20%").
- Output: Gap report (e.g., "8% orphans, avg Inlinks 1.8, 15 deep underlinked silos").
- Checkpoint: If bounce >70% and/or orphans >5% or avg Inlinks <2 or inlinks <3 at depth >3, recommend A/B content tests + internal boosts/silos; else pivot to off-page, because linking gaps explain 20-40% clustered ranking losses.
Analyze off-page using provided Ahrefs/SEMrush exports for backlink velocity drops or ranking changes.
- Input: Ahrefs Site Explorer backlink export, Semrush Position Tracking.
- Output: Red flag summary (e.g., "10% velocity drop correlating to 25% traffic loss").
- Checkpoint: If red flags, chain to backlink audit (below); else finalize deck, because off-page rarely causes isolated drops without prior signals.

Example: For a 25% non-branded dip with stable GSC impressions, output meta refresh plan from Ahrefs Content Gap restoring 18% sessions in one session.

Auditing Backlinks for Toxicity and Quality

Score backlinks 1-10 (relevance 60% weight, authority 30%, diversity 10%) on provided Ahrefs/SEMrush exports, because weighted scoring prioritizes high-impact disavows.

Export backlinks from Ahrefs Site Explorer or Semrush, filter Moz Spam Score >15%.
Calculate composite score per link: (relevance_score * 0.6) + (authority_score * 0.3) + (diversity_factor * 0.1).
Aggregate: If toxicity >5% of profile and correlates with traffic loss, recommend disavow list after 90-day monitoring plan.
- Output: Spreadsheet with scores, flagged list (<4 score).

Optimizing Technical SEO for Site Speed/Core Web Vitals

Apply fix hierarchy on provided PageSpeed/Lighthouse data, because sequential fixes target LCP/CLS multipliers.

Prioritize LCP >2.5s: Compress images <100KB (TinyPNG), defer render-blocking JS.
CLS >0.1: Lazy load below-fold images/videos.
Validate mobile-first in GSC usability report pre-launch.
- Output: Pre/post Lighthouse audit scores aiming 90+.

Decision Rules

Use these thresholds for branching in triage pyramid, because they segment issues by fix speed and impact.

Step	Condition	Action	Reason
1	20-30% organic sessions drop (segmented top 20% pages/queries)	Minor tweaks (Ahrefs Content Gap meta refreshes)	Resolves snippet volatility in hours vs. weeks for full audit.
1	>40% high-intent OR >50% apex pages + GSC impressions >30% drop + Semrush Sensor stable	Full pyramid	Captures broad algorithm or structural issues.
2	No GSC manual actions	Prioritize internal linking boosts	Avoids penalty fixation on clean profiles.
3	Core Web Vitals clear (LCP ≤2.5s, CLS ≤0.1, Lighthouse 90+)	Pivot to content/UX	Technical fixes alone yield <10% uplift if signals pass.
4	Bounce >70% on 1000+ visit pages OR desktop +20% info page spike	Content mismatch rewrites	Signals topical irrelevance over UX.
4	50-70% bounce + LCP >4s p0/p75 OR mobile spikes	UX friction fixes (e.g., mobile rendering)	Friction compounds bounce independently of content.
4	Dwell <30s	Weak content A/B tests	Short engagement ignores structure.
4	Avg Inlinks drop 5+ to <2 OR orphans >5% OR inlinks <3 at depth >3	Internal boosts/silos	Restores crawl equity for 20-40% traffic.
Off-page	Backlink velocity drop correlating to traffic	Chain to toxicity audit/disavow if >5% toxic	Isolated velocity rarely causal without toxicity.

Pre-internal linking: If Sheets QUERY >30% pages with Levels<=2 OR Silo_drift>20%, abort silos to Semrush Position Tracking, because flat structures dilute authority.

Hard Constraints

Never assume penalties without reviewing GSC manual actions, because unconfirmed assumptions lead to misdiagnosis and delayed recovery.
Always cross-check GSC impressions drop >30% and Semrush Sensor query-volume stability before pyramid escalation or tweaks beyond basics, because skipping chases SERP volatility wasting weeks.
Never launch without mobile-first validation in GSC mobile usability report, because it causes high bounce rates from unaddressed mobile friction.
Always audit backlinks quarterly/pre-launch and use Google Disavow only after approval with 90-day monitoring plan, because premature actions risk over-pruning authority.
Never produce bloated reports for quick-wins; output 5-slide deck only, because fluff buries actions and reduces implementation.
Always exclude noindex/nofollow/redirects/errors from Sheets calculations, because they distort Inlinks/orphan metrics.
Never apply internal silo workflow if >70% branded traffic, CWV <50th percentile, flat structures, dynamic content, or multilingual without isolation, because it backfires with traffic drops overriding internals.

Tools and Deliverables

Use these exact configurations for reproducibility.

GA4/GSC: Segment by device/page/query; Query Match Type; Pages Organic top 20%; date Compare exclude seasonal; manual actions/crawl logs/mobile usability.
Ahrefs/SEMrush: Content Gap duplicates; backlink velocity/export; Sensor for algorithm/query stability; Position Tracking.
MozCast/SEMrush Sensor: Algorithm updates (impressions drop 30%+).
PageSpeed Insights/Lighthouse: LCP/CLS thresholds; pre/post 90+.
Screaming Frog: Crawl ≤500 URLs; Internal Filter 'Inlinks=0'; Export Internal All (Inlinks/Levels/Status); silo CSV.
Google Sheets: =AVERAGE(Inlinks excl errors); =COUNTIF(Inlinks=0)/total; =COUNTIFS(Levels>3,Inlinks<3); prepend VLOOKUP silo-topic; =QUERY(SF_export,"SELECT * WHERE Levels<=2 OR Silo_drift>20%").
Other: TinyPNG images <100KB; Moz Spam >15%.

Quick-Win Deliverable Template (5-slide Google Slides deck + checklist):

Slide 1: Executive Summary
- Diagnosis: [e.g., 25% dip from internal linking gaps; GSC impressions stable]

Slide 2-3: Actions & Proof
- Checklist: [e.g., "Refreshed 12 dupe metas"] + before/after screenshots/Sheets metrics

Slide 4: Projected Results & Monitoring
- [e.g., "15-25% recovery in 7-14 days; track GA4/GSC weekly"]

Slide 5: Escalation Rules
- IF [GSC impressions >30% drop post-tweak] THEN [full audit]

Common Mistakes to Avoid

Don't skip GSC impressions <30% cross-check + Semrush Sensor at Step 1 for 20-30% drops; instead always confirm stability first, because it prevents full pyramids wasting weeks vs. 10-minute tweaks recovering 15-25%.
Don't run Sheets COUNTIFS(Levels>3, Inlinks<3) without prepending VLOOKUP/XLOOKUP silo-topic filter; instead always filter (=COUNTIFS(Levels>3,Inlinks_silo<3,Traffic>100)), Pareto top 20% GSC impressions with same-silo anchors, because cross-silo links dilute E-E-A-T yielding 0% uplift vs. 20-40% in 4-6 weeks.

Example: Cross-silo links from /blog to /products yield 0% uplift, while silo-matched anchors recover 20-40% cluster rankings.

Edge Cases and Limitations

Detect these signals to deviate from standard pyramid, because they invalidate assumptions like hierarchy or stability.

Flat/hub-and-spoke (SPAs, shallow e-comm): Levels <=2 across site; skip orphan/silo, pivot to faceted nav cleanup via Screaming Frog XPath, because Googlebot reaches all via nav.
Dynamic content (news/UGC): Weekly silo drift >20%; abort linking, because it backfires with 15-30% drops and <10% intra-silo flow.
70% branded GSC traffic OR CWV <50th percentile: Skip silos for externals/tech fixes, because branded dominance or vitals override internals.
Multilingual hreflang: Isolate per-locale via Ahrefs Language reports; avoid cross-language silos to prevent cannibalization.
Pre-audit: Sheets QUERY >30% Levels<=2 OR Silo_drift>20%; abort to Semrush Position Tracking.

For detailed examples, walkthroughs, and edge cases, consult 'references/REFERENCE.md'.

Use when

diagnosing sudden organic traffic drops
segmenting GA4 declines by query and page
triaging SEO issues technical-to-offpage
auditing internal orphans with Screaming Frog
analyzing silos in Google Sheets COUNTIFS
escalating minor dips vs full SEO audits
verifying GSC impressions before pyramid
fixing UX friction from bounce and LCP

seo-traffic-diagnosistriage-pyramidga4-gsc-segmentationscreaming-frog-crawlinternal-linking-silosgoogle-sheets-formulasseo-deliverablesbacklink-toxicity

your turn

Expertise like this takes years to develop.

Now any AI tool can learn it in seconds.