LB24 Meta Sync 02.09.2026

+97%
Revenue / month
Pre€1.73M
Post€3.41M
Spendonly +48%
6.9x → 9.1x
MER — revenue per ad-euro
Pre€20.8M ÷ €3.0M = 6.9x
Post€47.7M ÷ €5.2M = 9.1x
202610.3x
€176 → €145
Cost per new customer
Volume+79%
3.7x
Brand clicks / month
Clicks5.6k → 20.5k
CPC−39%
Pre = the 12 months before the Meta join (Jul 24–Jun 25) · Post = everything since the join (Jul 25–Aug 26) · data through Aug 30, 2026.

1) MER trend over the last years

6.4x
2023
6.2x
2024
7.7x
2025
10.3x
2026 (Jan–Aug)
MER = ALL shop revenue ÷ ALL ad spend (Google + Meta + Bing), per month
0x4x8x12xJoin Jul 252023-24 level: 6.2x2023-01: Total ad spend EUR k 2312023-02: Total ad spend EUR k 1912023-03: Total ad spend EUR k 2222023-04: Total ad spend EUR k 1872023-05: Total ad spend EUR k 1912023-06: Total ad spend EUR k 1552023-07: Total ad spend EUR k 1772023-08: Total ad spend EUR k 2032023-09: Total ad spend EUR k 1932023-10: Total ad spend EUR k 2542023-11: Total ad spend EUR k 3092023-12: Total ad spend EUR k 2212024-01: Total ad spend EUR k 2432024-02: Total ad spend EUR k 1862024-03: Total ad spend EUR k 2072024-04: Total ad spend EUR k 1932024-05: Total ad spend EUR k 2082024-06: Total ad spend EUR k 1652024-07: Total ad spend EUR k 1722024-08: Total ad spend EUR k 1672024-09: Total ad spend EUR k 2092024-10: Total ad spend EUR k 2062024-11: Total ad spend EUR k 3042024-12: Total ad spend EUR k 2432025-01: Total ad spend EUR k 2692025-02: Total ad spend EUR k 2202025-03: Total ad spend EUR k 2812025-04: Total ad spend EUR k 3032025-05: Total ad spend EUR k 3242025-06: Total ad spend EUR k 3382025-07: Total ad spend EUR k 3832025-08: Total ad spend EUR k 4472025-09: Total ad spend EUR k 3832025-10: Total ad spend EUR k 4502025-11: Total ad spend EUR k 7712025-12: Total ad spend EUR k 3982026-01: Total ad spend EUR k 2922026-02: Total ad spend EUR k 2632026-03: Total ad spend EUR k 2822026-04: Total ad spend EUR k 2972026-05: Total ad spend EUR k 3152026-06: Total ad spend EUR k 2552026-07: Total ad spend EUR k 3412026-08: Total ad spend EUR k 35123-0123-0423-0723-1024-0124-0424-0724-1025-0125-0425-0725-1026-0126-0426-08

2) Meta ROAS, pre vs post

PRE (pixel-active)POSTΔ
Last-click ROAS (hardest metric)2.03x3.24x+60%
Triple-Whale-credited ROAS (TA+V)4.73x10.43x+120%
Meta-reported 7d-click ROAS5.51x10.52x+91%
CTR (Meta account)2.19%2.97%+36%
TW-credited Meta ROAS per month
0x6x12x18xJoin Jul 2523-0123-0423-0723-1024-0124-0424-0724-1025-0125-0425-0725-1026-0126-0426-08
Spend

3) Brand demand; people searching for brand keywords

4 209 → 21 886
Brand clicks / month
Δ5.2x
Window2024 vs Sep 25–Aug 26
€1.81 → €0.86
Cost per brand click
Δ−52%
Window2024 vs Sep 25–Aug 26
22.0 → 69.4
Brand clicks per 1 000 non-brand Google clicks
Δ3.2x
Window2024 vs Sep 25–Aug 26
€133k → €517k
Brand conversion value / month
Δ3.9x
Window2024 vs Sep 25–Aug 26
Brand clicks per month
014 22928 45842 687Join Jul 2524-0124-0424-0724-1025-0125-0425-0725-1026-0126-0426-08
Flat through three Google-only years; moves 2–6 weeks after the Jul/Aug-25 Meta ramp; stays up after Meta spend is halved. DE — where 90%+ of the Meta budget runs — grew +147% YoY (Jan–Aug 26 vs Jan–Aug 25).
Brand demand vs the Google baseline
04794141Join Jul 2524-0124-0424-0724-1025-0125-0425-0725-1026-0126-0426-08
Metric i
If Google drove the growth, these ratios would stay flat — instead they jumped 3.2x at the Meta ramp. Brand impressions: 6.6x.
Head-to-head test: brand searches follow META spend 1–8 weeks earlier (+0.34, p=0.007; BFCM excluded p=0.022) and do NOT follow Google spend (−0.05, p=0.64) — brand campaign budget controlled, reverse causality excluded (placebo leads negative). 0META spend → brand clicks, lags 1–8w+0.34*** · p = 0.007GOOGLE spend → brand clicks, lags 1–8w-0.05 · p = 0.641
The math, step by step
  • 1Data: 87 weeks (Nov 24–Aug 26). Per week: brand-search clicks, Meta spend, Google spend, brand-ad budget.
  • 2Question: when brand searches rise, what rose in the 1–8 weeks BEFORE — Meta spend or Google spend? Both compete in one model.
  • 3Scale: everything in logs, so coefficients read as elasticities: +0.34 means +10% Meta spend ≈ +3.4% brand searches a few weeks later.
  • 4Controls (alternative explanations removed): brand-ad budget (not "we simply bought more brand ads"), same-week spend, a general growth trend, Black-Friday weeks.
  • 5Result: Meta +0.34, p = 0.007 — less than a 1-in-100 fluke. Google −0.05, p = 0.64 — statistically zero. Whiskers in the chart = 95% confidence intervals (Newey-West errors, robust to autocorrelation).
  • 6Direction: the reverse test — do brand searches predict FUTURE Meta spend? — comes out negative. Spend drives searches, not the other way around.
  • 7Stress tests: all BFCM weeks removed → still significant (p = 0.022); re-estimated with an independent statistics library → identical numbers.

4) Direct traffic — people typing luxusbetten24.de

73k → 150k
Direct sessions / month
Δ2.0x
Window2025 H1 vs 2026
−48%
Meta spend, ramp → 2026
Sessions±0%
WindowJul–Dec 25 vs 2026
1.8k → 3.0k
Direct sessions per €1k Meta
Δ1.7x
Window2025 H1 vs 2026
+0.60, p = 0.0005
Spend → traffic, lags 1–8w
New vis.+0.59 · p = 0.03
4w-diffp = 0.07
Direct sessions per month (thousands) · bars = Meta spend
077155232Join Jul 252024-11: Meta spend EUR k x1.2 362024-12: Meta spend EUR k x1.2 312025-01: Meta spend EUR k x1.2 222025-02: Meta spend EUR k x1.2 242025-03: Meta spend EUR k x1.2 452025-04: Meta spend EUR k x1.2 472025-05: Meta spend EUR k x1.2 572025-06: Meta spend EUR k x1.2 542025-07: Meta spend EUR k x1.2 832025-08: Meta spend EUR k x1.2 1192025-09: Meta spend EUR k x1.2 912025-10: Meta spend EUR k x1.2 1092025-11: Meta spend EUR k x1.2 972025-12: Meta spend EUR k x1.2 702026-01: Meta spend EUR k x1.2 572026-02: Meta spend EUR k x1.2 462026-03: Meta spend EUR k x1.2 522026-04: Meta spend EUR k x1.2 532026-05: Meta spend EUR k x1.2 502026-06: Meta spend EUR k x1.2 472026-07: Meta spend EUR k x1.2 482026-08: Meta spend EUR k x1.2 4324-1125-0225-0525-0825-1126-0226-0526-08
Spend halves in 2026 — sessions stay at ~2x the pre level.
The budget-cut test: spend halved, direct traffic held
058115173€42k73k2025 H1 (pre-ramp)€95k150kJul–Dec 25 (ramp)€50k-48% vs ramp150k±0% vs ramp2026 Jan–Aug (after cut)
After the cut: spend −48%, sessions ±0% — still 2.0x the pre level. Bought traffic decays in days; this did not.
Direction of causality: traffic follows spend — not the reverse
-1.0-0.5+0.0+0.5+1.0+0.05+0.02same week+0.60***+0.59**AFTER spend (lags 1–8w)-0.37**-0.55*BEFORE spend (leads · placebo)
Direct sessionsDirect NEW-visitor sessions
The effect loads AFTER spend and turns negative BEFORE — reverse causality excluded. Whiskers = 95% CI (weekly OLS in logs, Newey-West; Google spend, trend, BFCM controlled).

5) Zero-click buyers naming Meta — the survey fingerprint

17.6% vs 2.7%
of zero-click buyers name Instagram/Facebook — 6.5x the TikTok recall placebo (n = 2,155, p ≈ 9e-34)
≈ €110k / mo
zero-click 2026 revenue naming Meta — invisible to any click attribution
Monthly detail
% of zero-click buyers naming Meta, by month · red line = TikTok placebo
0%8%16%24%TikTok recall placebo: 2.7%2025-09: % of zero-click buyers naming Meta = 16.4%2025-10: % of zero-click buyers naming Meta = 19.3%2025-11: % of zero-click buyers naming Meta = 17.5%2025-12: % of zero-click buyers naming Meta = 15.8%2026-01: % of zero-click buyers naming Meta = 15.6%2026-02: % of zero-click buyers naming Meta = 15.3%2026-03: % of zero-click buyers naming Meta = 18.2%2026-04: % of zero-click buyers naming Meta = 19%2026-05: % of zero-click buyers naming Meta = 16%2026-06: % of zero-click buyers naming Meta = 15.6%2026-07: % of zero-click buyers naming Meta = 18.6%2026-08: % of zero-click buyers naming Meta = 21.6%25-0925-1025-1125-1226-0126-0226-0326-0426-0526-0626-0726-08
Stable 15–22% every single month since the survey started — not a campaign artifact.

6) Meta clicks inside organic closers

9.8% → 18.2%
of organic closers carry an EARLIER Meta click (2025H1 → 2026, p = 4e-11)
61%
of Meta-touched orders close on another channel's click — last-click systematically undercounts Meta
Monthly detail
% of organic closers with an earlier tracked Meta click, by month
0%8%15%23%Join Jul 252025-01: % of organic closers with earlier Meta click = 5.3%2025-02: % of organic closers with earlier Meta click = 7.7%2025-03: % of organic closers with earlier Meta click = 10.4%2025-04: % of organic closers with earlier Meta click = 11.7%2025-05: % of organic closers with earlier Meta click = 9.1%2025-06: % of organic closers with earlier Meta click = 12.7%2025-07: % of organic closers with earlier Meta click = 11.3%2025-08: % of organic closers with earlier Meta click = 15.3%2025-09: % of organic closers with earlier Meta click = 15.4%2025-10: % of organic closers with earlier Meta click = 15.7%2025-11: % of organic closers with earlier Meta click = 16.3%2025-12: % of organic closers with earlier Meta click = 16.5%2026-01: % of organic closers with earlier Meta click = 16%2026-02: % of organic closers with earlier Meta click = 17.6%2026-03: % of organic closers with earlier Meta click = 20.1%2026-04: % of organic closers with earlier Meta click = 20.9%2026-05: % of organic closers with earlier Meta click = 19.4%2026-06: % of organic closers with earlier Meta click = 17.5%2026-07: % of organic closers with earlier Meta click = 17.5%2026-08: % of organic closers with earlier Meta click = 17.1%25-0125-0325-0525-0725-0925-1126-0126-0326-0526-0726-08
Tracking setup unchanged — the doubling is behavior, not measurement.

7) Double verified

−2.1%
TW revenue vs independent Shopify export (2026 Jan–Jul)
0.0%
ad spend vs TW export, matched windows, every year
€1 079 = €1 079
AOV pre = post — customers, not prices
pixel-active only
2024 pixel-dead months excluded everywhere
Full test table
ClaimTestEffectSignificance
Brand searches follow META spend (lag 1-8w)weekly OLS, HAC-8, brand budget controlled+0.34p = 0.007 · BFCM-excl. p = 0.022
Brand searches follow GOOGLE spendsame model, head-to-head−0.05p = 0.64 → rejected
Organic orders follow Meta (lag)weekly OLS, HAC-8; placebo leads negative+0.55p = 0.007
Organic traffic follows Meta (lag)weekly OLS, HAC-8+0.92p < 0.0001
Direct traffic follows Meta (lag)weekly OLS, HAC-8; 4w-differences p = 0.07+0.60p = 0.0005
Hidden Meta in zero-click orderssurvey share vs TikTok recall placebo17.6% vs 2.7%p ≈ 9e-34
Meta clicks inside organic closers2025H1 vs 2026, two-proportion z9.8% → 18.2%p = 4e-11
Efficiency break in the Meta erabreak scan at every month, trend-controlledMER +2.5xpeak t = 6.5–9.3 (Nov 25–Jan 26); no positive 2024 break (max t = 0.1)
Robust errors (Newey-West) · lag-vs-lead identification · placebo channels & dates · two sources per core number. Next step to sharpen further: 4-week geo-holdout / conversion-lift test.
Basis: 144,818 orders · 108,337 journeys · 12,529 surveys · full ad accounts. Sources: Triple Whale raw archive (journeys 2023-11–2026-08 · orders 2020-07–2026-08 · pixel attribution · post-purchase surveys · ad accounts 2021-11–2026-08, complete days through 2026-08-30) · Shopify order exports. Join date July 2025 · compiled 2026-08-31.