2026-08-03

TL;DR

• UK is the day's only mover (avg Δ1d −0.6; every other market flat-to-up). t3 183→158, t10 338→319, ranked 347→338. This is the deepest UK compression of the month and the first to push the ranked count down — my standing escalation trigger fired.
• Investigation resolves it benign-but-watch: of the 183 terms that were top-3 yesterday, 0 fell out of top-20 (140 held t3, 43 slid within top-20). The −9 ranked is a separate group of 10 city-head terms cliffing from the 7–11 contention zone straight out; only 1 term (sheffield) came in.
• Not technical (all pages 200, clean canonical, no noindex through the 08-04 snapshot), not a competitor takeover (vivastreet #1 / adultwork #2 frozen; uk.skokka.com climbs locally on southend 12→7 and manchester 18→9 but is flat overall 8.8→9.0).
• AT (t3 209→217) and DE (155→166) both rebounded today — the rolling top-3 reshuffle reversed in those markets exactly as forecast. UK is the phase-shifted laggard having its dip while the others recover.
• DE Hannover case de-escalating: huren hannover (vol 18,100) OUT→17→8, nutten hannover 8→7 — the escalated slide is reversing.
• Verdict: amber. UK likely reverses on the month-long pattern; escalate to red only if these terms stay out 2+ days or more stable heads follow.

What moved and why

UK city-head compression, deepest instance (reopened — confidence: likely reversible). The board fell across t3/t10/ranked in one day, UK-only. Two distinct things are happening and it's worth separating them:

1. The top-3 cohort reshuffled within top-20 (benign, same as the four prior dips). Dispersion of the 183 terms that were top-3 on 08-02: 140 still t3, 39 at 4–10, 4 at 11–20, 0 out. Zero visibility lost from the top-3 group — pure #1-3↔#4-10 churn, identical signature to 07-29→30 which reversed.

2. A separate group of 10 boundary terms (rank 7–11 yesterday) cliffed fully OUT of top-20 — hertfordshire, hemel hempstead, northampton, cardiff, southend, canary wharf, wolverhampton, london (/central-london/), luton, manchester-2. This is the −9 ranked and the new element. About half are habitual oscillators (cardiff OUT on 08-01, london OUT 07-31/08-01, luton OUT 07-31, manchester OUT 07-29); the other half were genuinely stable page-1 (southend rock-steady 4–6 all fortnight, wolverhampton 4–7, hertfordshire 3–7, hemel 2–9) and cliffed clean.

Why I don't read this as a real authority loss despite the trigger firing:

• Pages healthy. All four stable-head pages return 200 with correct self-canonical and empty robots through both the 08-03 and 08-04 snapshots. No deindex, no relocation.
• No competitor cause. The top-2 incumbents are frozen on every term I checked. uk.skokka.com's localized climb (southend 12→7, manchester 18→9) is a symptom of the same recompression at the mid-page-1 boundary, not a cluster takeover — its overall UK avg is flat (8.8→9.0).
• UK-only. Five other markets flat-to-up. A domain-wide technical or a global algo update would hit all six; it didn't. The UK-only footprint is itself evidence against both.
• AT/DE reversed today. The rolling top-3 reshuffle bounced back in the two markets that were mid-dip yesterday. UK, which had recovered to 183 on 08-02, took its turn to dip. The rotation is behaving exactly as the case describes.

So the honest read: the deepest instance of a month-long, always-reversing UK SERP recompression, which this time squeezed ~10 contention-zone terms past the top-20 line rather than holding them inside it. Sits in the still-unconfirmed late-July→Aug Google small-wave volatility.

DE Hannover slide reversing (de-escalating — confidence: likely). The 08-02-escalated head, huren hannover (vol 18,100), went OUT→17→8 — back into top-10. nutten hannover 8→7, huren in hannover steady 8. Only sex in hannover is still OUT, flapping between /sex/hannover/ and /escort/hannover/ (the known cannibalization). The internal-link lift I recommended looks less urgent now that the priority head recovered on its own.

IT capture holding (confidence: confirmed). IT at period-high depth (ranked 172, t10 116), 7 top-3 terms — all EA-vacated cities (escort italia, potenza, laquila, legnano, sassari, cremona). The escort-advisor migration vacuum is still being converted as forecast.

Investigations

• UK ledger. 10 out / 1 in (sheffield 18). One-sided −9 day, not offsetting churn — that's what made me treat it as more than baseline.
• Snapshots. Pulled status/robots/canonical history for the four stable OUT pages (southend, wolverhampton, hertfordshire, hemel). All 200, clean, unbroken back to 07-16. Ruled out technical.
• Competitor_ranks on 5 OUT terms, 08-02 vs 08-03. Incumbents frozen; skokka's only meaningful gains are localized (southend, manchester). Ruled out a competitor takeover.
• Dispersion of the 08-02 top-3 cohort: 0 of 183 fell out of top-20. This is the single most important number today — it says the t3 −25 carries no visibility loss.
• Cross-market. AT/DE trajectories confirm both rebounded; UK is the lone faller.
• Dead end / note: GSC impressions (the ideal way to separate tracker flux from real click loss on the 10 OUT terms) remain unavailable in this env — no gsc-monitor DB on disk. Substituted rank dispersion, which answers the top-3 question directly but can't confirm the 10 OUT terms; that's why the watch stays sharp rather than closing green.
• huren münchen (vol 27,100) 5→OUT after a week at 5 — flagged because of the volume, but it's a single München-head blink (page presumed fine; this head closed as return-on-jitter before) on a DE board whose t3 rebounded today. Light note, not a front.

Opportunities

• UK reclaim, once flux settles. The 39 terms that slid t3→4–10 today are all striking-distance to reclaim top-3 the moment the recompression relaxes (it has four times this month). No action needed now — just don't touch the template mid-flux.
• IT EA-vacated cities remain the single highest-ROI action on the board — internal-link + freshness push on the rank 4–7 cities (palermo/perugia/firenze bounce 3↔6-7; a lift nails them to #3). Window is decaying as EA rebuilds under .net; bank gains this week.
• BR still compounding (ranked 109, Δ30d +1.2) — the /br/acompanhantes/ internal-link build remains an open, receptive target.

Recommendations

1. Do NOT touch the UK escort city-page template. Four prior compressions this month all reversed untouched; the pages are healthy and the cause is external SERP flux. (posture, not action)
2. Watch the 10 UK OUT terms daily. If they return in 1–3 days → close green as the fifth reversal. If they stay out 2+ days, or more stable heads (currently top-3) cross below 20 → escalate red and treat as real visibility loss.
3. IT internal-link push on rank 4–7 EA-vacated cities — highest-ROI action, window closing.
4. Standing UK anchor hygiene (low urgency): point city-head anchors at canonical city URLs (/cardiff/ not /cf11/, /central-london/ not /london-2/) to reduce the sibling-URL flapping that amplifies these boundary swings.

Watchlist

Updated in watchlist.md: UK case reopened (deepest compression, ranked −9, benign composition, sharp watch with a red-tomorrow trigger); DE Hannover de-escalated (huren hannover recovered to 8); rolling top-3 reshuffle noted AT/DE rebounded; IT capture holding; BR compounding.