The learning journey

The bubble-and-burst story ↻Later revised

Six agents pulled GA4, GSC, and live-site data. First verdict: both sites' organic footprint was ~95–99% informational "People Also Ask" blog traffic from a mass 2025 publishing campaign (TC: 141 posts Jun 23–Sep 11 2025; Dallas: 310 posts Apr 24–Sep 26 2025), and a mid-May-to-mid-June 2026 algorithmic reassessment deflated that bubble roughly in half.

• Held up: commercial/local organic was always negligible (TC ~1.7 clicks/day at peak; Dallas ~2–5/day, head term on page 2 since Nov 2025) — so lost calls could not be explained by lost commercial web clicks.
• Held up: two site defects — the Dallas tel-link hijack (GA call tracking dead since Oct 2024) and a broken quote form in June (fixed July).
• Later revised: the size of the June damage, the "TC call_to_click −80% YoY" claim, and the May core-update attribution.

Hit map, entity audits, tracking sweep

Five more agents: a 35-property GSC hit map, Houston/Tampa deep-dives, a sibling/entity audit, and a 39-site tracking-snippet sweep.

• Hit map: network −22.5% clicks/day spring→summer; 89.5% of absolute loss from /blog URLs; blog-heavy sites −26.6%, blog-light −7.7%. ↻Revised in R3
• Region-entity as cause of the June drop: region-named sites averaged +2.8% vs city-named −13.9%; central-ohio +105%. ✕Rejected
• TC region-entity as cause of pre-existing commercial weakness: "treasure coast" ≈ 2 commercial queries / ~100 impressions in 5.5 months; schema pins the business to Palm Bay. ✓Supported
• Fort Worth "whiteout" of Dallas: zero town overlap, zero cross-bleed. Real story: Dallas is a thin FW-cloned site (service pages still slugged *-fort-worth) in the network's biggest demand pool — 50.9k commercial-geo impressions. ✕Cause rejected, symptom real
• Missed aggravator found: fort-pierce duplicates TC's entire 42-town service-area map URL-for-URL and shares the same street address — three FL sites, one NAP. ✓Verified in DOM/schema
• Tracking sweep: the tel-hijack is unique to Dallas (1/39); five sites are analytically blind; florence.aaacwildliferemoval.com is NXDOMAIN (real franchise at centralky). ✓Confirmed

The adversarial re-audit that changed the verdict ✓The pivotal round

An adversarial re-audit used GA4's 2024 pre-bubble year as the seasonal baseline — GSC's 16-month retention had hidden it, and 2025 (the control used earlier) was contaminated by the bubble. Summer normally runs ~25% below spring (2024 median ratio R = 0.75 across 9–10 GA4 properties; GSC and GA4 agree within 4 points where both exist). 2026's ratio (0.72) is within 4% of the 2024 norm at the median site.

R3 also produced the single most important finding of the investigation: the ~20×-traffic / zero-incremental-calls comparison. And it opened a contradiction: if TC total call events are at an all-time high, why did the franchisee report a bad two weeks?

The flat aggregate was hiding a real decline ✓Both prior views partly right

Decomposing TC's summer organic call events by engine and landing-page class (windows Jun 1–Aug 13 each year) resolved the "44 / 123 / 35" call divergence:

• The 2025 "spike" was a Yahoo artifact: Google-organic calling sessions are flat — 31 / 29 / 29 across 2024/2025/2026. 82 of 2025's 123 events came from a four-month Yahoo/Bing burst, out-of-market (Miami 82, Fort Myers 21, Orlando 8 sessions), converting at 14.9% vs Google's 5.1% on the same page and window.
• Under the flat aggregate: TC commercial sessions −65% vs 2024 (841→295), service-area calls 20→12 — purely traffic-driven; conversion rates held (~10–11% all three years).
• Call economics: 1 call per ~9 service-area sessions vs 1 per ~1,650 blog sessions.
• So R3's "nothing happened to TC calls" was right about the aggregate and wrong in composition — a genuine, non-seasonal commercial visibility decline is being masked. This re-elevated the entity/cluster findings.

Bing misclassification confirmed; Fort Pierce cleared as the cause

• The Yahoo burst = untagged Microsoft Ads (owner's hypothesis confirmed): yahoo/organic ran 1–9 sessions/mo for 16 months, jumped to 38/114/130/181 (May–Aug 2025), then died the exact month bing/cpc tagging appears (Aug 2025: 11 sessions; Sep: 103). Corrections: TC Bing ads ran from ~May 2025, not "since Jan 2026"; paid masking of TC organic began May 2025; the 14.9% burst CTC rate matches the 10–14% bing/cpc signature — one consistent, still-unexplained high-rate paid pattern. ✓Confirmed
• Fort Pierce duplication is fragility, not the 2026 cause: the 165 duplicate town pages date to Jan 2022 and coexisted with TC's good 2024 commercial performance; FP's 2026 growth is old blog posts, not town pages capturing TC traffic. Dedup remains worth doing — it costs nothing and removes doorway exposure — but it is necessary-not-sufficient. ↻Expectation reset