FORENSIC VERDICT + RECOVERY PLAYBOOK — itblogpros.com

Synthesized June 2026 from four research streams + independent verification pass. Confidence levels stated per section.

1. Why itblogpros was hit — VERDICT (confidence: HIGH)

Primary cause: Google's Helpful Content sitewide classifier — specifically the improved December 2022 classifier, whose rollout completed January 12, 2023. That date is the single best match for a decline "starting early 2023," and the HCU is the only system in the window whose documented mechanics produce a gradual slide rather than a cliff: the classifier runs continuously, weights suppression by the volume of unhelpful content, and applies/lifts over months — "the good content on your site will also be affected". Google's definition of unhelpful content reads like the site's historic profile: "lots of content on many different topics in hopes some of it performs well" (UK ISPs + US ISPs + laptops + area codes + Tiscali + DirecTV sprawl) and "summarizing what others say without adding much value" (7+ templated reset-router pages).

Compounded by a spring-2023 triple punch aimed at exactly this site archetype:

The Aug 2023 core ratcheted the suppression down further, and the Sept 2023 HCU was the finisher, not the trigger. Glenn Gabe's ~700-site reviews study shows the reviews system and HCU converge on the same site-level quality judgment — a spring-2023 reviews hit presaged the September HCU hit.

Why AI/scaled competitors outranked you in 2023 specifically: they manufactured the exact signal Google had just started rewarding. robotpoweredhome.com opened every article with a fabricated first-person anecdote ("I was recently trying to help my father pay his Verizon phone bill…") — synthetic "Experience" at scale, during the 9-month window between E-E-A-T's expansion (Dec 2022) and the Sept 2023 HCU learning to detect it. Same archetype as itblogpros, executed harder, with fake experience bolted on. It was arbitrage on a classifier lag — not a better site.

Caveat: without date-stamped GSC data this is probabilistic — but the profile-to-target match is unusually clean, and the verification pass confirmed every load-bearing date and mechanic.

2. The robotpoweredhome lesson (confidence: HIGH)

What happened: Registered April 2020, anonymous personas ("brainchild of Mathew and Charlotte," all posts bylined "Matt," no bios — verified via RDAP and Wayback), ~586 posts blanketing your exact keyword matrix (router lights × every US ISP, reset templates, bill-pay-without-login), Mediavine + Amazon monetized. Its last legitimate post is dated September 27, 2023 — inside the Sept 14–28 HCU rollout. The operators abandoned it within two weeks of the update landing. By June 2026: absent from every SERP it owned, ~19K visits/mo and falling, 93.8% bounce, StatShow worth $10, and since Feb 2026 it's a multilingual casino/crypto spam shell (still bylined "Matt") after an apparent quiet domain sale.

Three strategic conclusions:

  1. Your 2023 loss was a cohort event, not a head-to-head defeat. RPH didn't beat you with quality; it rode the same thin-affiliate pattern harder until Google reclassified the entire archetype. Its collapse and your gradual slide are the same event viewed from two altitudes.
  2. The vacated SERPs are re-winnable. Current top-10 for RPH's old money queries ("spectrum router red light") is mid-tier beatable sites — PCGuide, ConnectCalifornia, GetBillSmart, ModemGuides — not locked-up big brands.
  3. The one fatal move is re-adopting its playbook. Google's classifier demonstrably learned to detect template anecdotes and synthetic experience; HCU-cohort sites that kept the pattern stayed dead. The only viable experience signal now is verifiable experience: real name, real photos, real testing. Your June 2026 author and consolidation work is the correct anti-pattern — RPH is the control group proving why.

3. The survivor pattern (confidence: MEDIUM — observed, causality inferred)

Six practices shared by every verified survivor (ispreview, thinkbroadband, broadbandsavvy, stevessmarthomeguide, routersecurity.org, routerleds) — and inverted by every casualty:

  1. One nameable topical identity, zero sprawl. "The router-LED site." "The router-security site." For itblogpros: the home-WiFi troubleshooting site — finish pruning everything else. (Backed by leaked siteFocusScore/siteRadius attributes and a documented 23% lift from pruning.)
  2. At least one non-article asset a scaled site can't fake — calculator, original photo library, dataset, course, forum. Cheapest fit: a "which light is your router showing?" interactive picker + photographed LED states.
  3. A named human with a verifiable OFF-site footprint — Guardian bylines (Paton), Computerworld columns (Horowitz). The Graeme Messina credential build-out mirrors this exactly; keep compounding it.
  4. Zero-to-two affiliate links on troubleshooting pages, late, disclosed, never the page's purpose. No survivor runs Amazon "best X" roundups — those SERPs are now exclusively big media + ISPs. Cede them.
  5. Per-page information gain: routerleds covers 20+ brands but every page has device-specific original photos and brand-specific steps — never find-replace templates. It outranks against official ISP help pages on this alone.
  6. Visible maintenance: original + updated dates, changelogs, steady cadence.

4. Bing-first playbook — ranked actions (confidence: HIGH)

You're at positions 3–5 with ~430 clicks/mo and IndexNow already wired. CTR is itself a Bing ranking signal and Bing shows meta descriptions mostly verbatim (vendor-sourced, but consistent across guides) — so CTR work is double-leverage and needs zero new content.

  1. Rewrite titles + meta descriptions on the top-20 Bing queries. Bing is literal: titles 55–60 chars containing the exact query phrasing ("WiFi shows 2 bars," not a paraphrase); descriptions 150–160 chars with keyword + concrete promise. Highest ROI move available, this week.
  2. Audit sitemap lastmod honesty. Must be ISO-8601 from per-post front-matter/git dates, not build time — Bing says accurate lastmod directly controls how fast updates reach search and AI answers, and a 241-page build-date stamp is a false freshness signal.
  3. Expand the proven clusters using BWT Keyword Research (impressions-based, 24 months of data, country filter for the UK/US split): router lights, bill pay, signal bars. RPH's keyword matrix is a free shopping list of vacated, mid-tier-held SERPs.
  4. Genuine 30–60-day refresh cycle on top earners — real content/data changes at code level (Bing detects date-only edits as nothing). This also counters the 97% Copilot-citation decay over two months.
  5. Add 40–60-word capsule answers under question-H2s + FAQ/HowTo schema to the how-to templates — feeds both Bing rich results and Copilot citation likelihood.
  6. Light social distribution (LinkedIn/Pinterest/Facebook) — social engagement is an explicit Bing factor (practitioner-sourced; cheap, do it incidentally).
  7. Check AI Performance grounding queries monthly as a keyword-research layer — but treat citations as brand presence only: 48,000 citations produced 14 clicks.

Skip: MSN Partner Hub (invite-only, prohibits affiliate content), Bing Places (local-only), citation-chasing as traffic. Bonus: Bing rank syndicates free to Yahoo, DuckDuckGo, Ecosia, and ~87% of ChatGPT search citations matched Bing top-10 (declining as OpenAI's own index ramps, but still material).

5. Google recovery — honest odds + plan (confidence: HIGH on mechanics, MEDIUM on probability)

Odds, stated plainly: The regime changed in your favor — the June 2025 core brought HCU sites "surging back from the dead" including full recoveries, after base rates of 129/130 still declining (early 2024). But recoveries didn't correlate cleanly with remediation (DMARGE spent $200K+ for nothing; HouseFresh fully recovered via brand + Google-side change), and Google told publishers Sept-2023 traffic "would not return" — AI Overviews now absorb exactly your how-to queries. Realistic call: ~50% chance of meaningful partial recovery (a fraction, not a multiple, of pre-2023 traffic) within 1–3 core cycles (6–18 months); full recovery <10%, ever. One genuine positive signal: strong Bing performance means users are satisfied by the content — this is classifier-level suppression, not content failure, and that's the recoverable kind.

Plan, in priority order (weighted by evidence strength):

  1. Brand demand + engagement — the heaviest lever. Moz: HCU losers averaged Brand Authority 37 vs 50–52 for winners. Anything that makes people search "itblogpros" or return directly: YouTube router-fix videos, genuine Reddit/forum presence. This is what separated HouseFresh from DMARGE.
  2. Finish the topical prune to pure home-WiFi/broadband (UK+US). The 78-page consolidation was right; complete it — area codes, Tiscali, DirecTV, laptops all gone or 301'd. Consolidate (with 301s) rather than mass-delete — Google calls deletion "a last resort".
  3. First-hand evidence on the money pages: original photos of actual router light states, hands-on test results — the strongest content-level winner trait in Shepard's analyses, and it differentiates the router-lights cluster on Bing simultaneously.
  4. Differentiate or consolidate the remaining reset-router templates — per brand, with device-specific photos and steps (the routerleds model), or merge them.
  5. Keep the E-E-A-T/author work as supporting infrastructure — necessary, never sufficient; no documented recovery was driven by bios alone.
  6. Do NOT buy or build artificial links — Pickr recovered with zero link building; a high link-to-brand ratio is a loser marker.

Operating rule: ship in big batches, not trickles (Gabe: "tackle big problems in big ways — minor changes will probably not help"), then judge results only at core-update boundaries — reassessment takes months, often 6+, sometimes multiple cycles. Between updates, Bing is the business; Google recovery is the call option.

The single sentence to operate by: Become the verifiable, photographed, single-topic router-troubleshooting authority that RPH faked — compound Bing now, and let Google's recalibration find a site that no longer matches the 2023 loser profile.