Window: 21 Aug – 17 Sep 2026 (28 days) · pulled live 19 Sep 2026
Search Console, GA4, Clarity and the live database, read across both the gym and the shop. Real growth through mid-August, a plateau since, a real day-pass sale, and two tracking gaps closed the same day.
This is not a flat month. Weekly Search Console clicks for unrealgymthailand.com roughly 5x'd from late May to mid-August, peaked at 424 clicks in one week, and have held in a band around 380–420 clicks a week for the five weeks since. Growth stalled, it did not drop.
Growth phase (86 → 424 clicks/wk, late May to mid-Aug) in ink, plateau (five weeks around 380–420/wk) in amber. Week of 17 Sep is partial and excluded. Source: Search Console weekly aggregation, sc-domain:unrealgymthailand.com.
28-day window vs prior 28 days
Cross-check: the 14 Sep automated SEO monitor, on its own rolling window ending 11 Sep, read 7d clicks at 406 vs 366 prior (+11%) and 28d clicks at 1,625 vs 1,323 prior (+23%). That window still overlapped the peak, so it read as growth; this pulse's fixed 21 Aug–17 Sep window is the one that isolates the plateau. Different windows, not conflicting evidence.
Three real people paid for a day pass this window, confirmed twice over: the same three purchases show up in GA4 as purchase events and as rows in the database. That is the win. The funnel that gets people there is the problem.
Confirmed paid day passes, this window
The path that leaks, gym → shop
The fixed Day Pass control routes to the /day-pass-bangkok info page, not the shop, so 947 counts info-page arrivals only. Just 142 continued on to the shop's own checkout link — that was the funnel's weakest joint, fixed and verified live the same day (section 05).
Zero product orders is expected, not a failure: every buy button is "Notify me" while Ton finishes the physical stock count that flips products live one by one. The real gap is that the list meant to capture that pre-launch demand is not capturing anything.
Real shop traffic vs. what got captured
Notify me / waitlist capture, 28 days
0 new waitlist rows from those 1,527 visits (shop_page_views, first-party, no consent gate). The Notify me modal fired its click event once in the whole window.
Read shop volume from the right source
The shop's own GA4 sees only 48 sessions in 28 days against 1,527 real page views — a 17.5:1 undercount caused by its consent gate. That is correct behaviour, not a bug, and it stays. It just means shop volume is read from shop_page_views or Clarity, never from shop GA4 alone.
Two live searches this morning both surfaced UnReal Gym prominently and favourably in the AI-synthesized answer, citing the gym's own pages plus independent third parties. This corroborates a real GA4 channel, not a one-off.
Cited live this morning, answering "best gym in Bangkok 2026" and "UnReal Gym Bangkok Sukhumvit review":
Watch item: Clarity's own AI-citation count dropped Single-digit counts
Separate from the GA4 channel above: Clarity's own AI Visibility tab tracks how often its crawl finds UnReal cited in an AI answer. The 14 Sep automated monitor read that count fall from 9 citations (previous run) to 1, share of authority from 30% to 16.67%, narrowing to a single cited page (/day-pass-bangkok). Both weeks are single-digit absolute counts, so this is a watch item for the next run, not a confirmed regression.
Both findings below shipped and were verified live the same day.
Confidence and rough upside per item, most valuable first.
Hygiene
QA rows still sitting in production: one stale tracking-audit lead (18 Aug), 2–4 healthcheck day-pass rows, all 8 shop_orders from the 8 Aug stock-gate test. Safe to delete, excluded from every number above. A JSON-LD reader crash on the shop (r["@context"].toLowerCase) is cosmetic but real, worth a small polish fix.