Split Test Board
Every live split test — what we're testing, who's winning, how sure we are. All results count only the time both pages ran side-by-side in an equal split (traffic a control got while running alone is excluded). Revenue = Checkout Champ COMPLETE orders only, QA/internal orders excluded. Self-hosted page views updated 2026-09-18 12:16 (GMT+7). What do these numbers mean? A winner is called at 95% confidence + 100 paid orders per arm (day-14 stop-loss). Updated 2026-09-18, 12:16 (GMT+7). Daily buckets are CC accounting days.
Data window: complete Checkout Champ days only, ending 2026-09-17 (CC day) · refreshed 2026-09-18, 12:16 GMT+7 · next refresh 12:00 GMT+7 daily. A CC day ends at midnight New York = 11:00 GMT+7, so the newest complete day is always the previous CC day; today's orders are never counted.
Aether™ pendant
checkout Checkout page: v4 (new) vs v2.1a (control)
Day 16/14
v4 ahead
93% sure (need 95%)
KEEP RUNNING
STRONG LEAN
Solvane™ bracelet (neuropathy)
checkout Checkout page: v2 (new) vs v1 (control)
Day 16/14
v2 ahead
76% sure (need 95%)
KEEP RUNNING
EARLY
Schorl bracelet
checkout Checkout page: v2 (new) vs v1.1 (control)
Day 16/14
v2 ahead
85% sure (need 95%)
KEEP RUNNING
STRONG LEAN
Anti-Smoking Bracelet™
checkout Checkout page: v3 (new) vs v2 (control)
Day 16/14
v3 ahead
90% sure (need 95%)
KEEP RUNNING
STRONG LEAN
Pixiu Mantra Ring
checkout Checkout page: v5c vs v5b vs v5a (control) — one 3-way test, equal 33% split
Day 17/14
v5b ahead
89% sure (need 95%)
KEEP RUNNING
STRONG LEAN
VitalTac flashlight
checkout PayPal express: Legacy path, no email (v4) vs Direct path, email required (v2)
Day 3/14
null ahead
50% sure (need 95%)
KEEP RUNNING
STALE
VitalTac flashlight
checkout Checkout design: v3 (Pixiu-style) vs v2 (current)
Day 18/14
v3 ahead
86% sure (need 95%)
DRAW
STRONG LEAN
CC-hosted checkouts (CartClick)
Aether™ pendant checkout page
Checkout page: v4 (new) vs v2.1a (control)
16 complete days of data
Updated 10 minutes ago · 2026-09-18 12:06 GMT+7STRONG LEAN
new v4 checkout design challenging the proven v2.1a control (winner of the v2.1a-vs-v3a test) · window: 16 complete CC days — from 2026-09-02 21:34 (GMT+7) (both pages live together) through end of 2026-09-17. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v4 46% / v2.1a 54% — even ✓
18,575ad clicks (fb → funnel)
$2.59RPC — rev per ad click
$2.12cost per click
+$0.47margin per click
1.22×ROAS (tracked campaigns)
$75,172.62revenue — these pages (all routes)
RPC/ROAS use the tracker's own attributed revenue ($48,177.40) and clicks via ClickFlare "aether167" (visits=ad clicks) — one consistent universe. ⚠ These pages actually took $75,172.62 in the window — other campaigns/routes also feed them; RPC covers only the tracked ones. Per-arm decisions stay on the $ / checkout start columns.
v4 ⭐
| page views | collecting… |
| checkout starts | 615 |
| paid orders | 335 |
| form→sale rate | 54.5% |
| tried to pay (card/PayPal) | 71% |
| card approval (the bank) | 68% · 100 declined |
| $ / checkout start | $59.73 |
| AOV | $109.65 |
| revenue (window) | $36,733.21 |
| upsell take | 49% |
paid 335/100
open page ↗
v2.1a
| page views | collecting… |
| checkout starts | 722 |
| paid orders | 355 |
| form→sale rate | 49.2% |
| tried to pay (card/PayPal) | 64% |
| card approval (the bank) | 68% · 102 declined |
| $ / checkout start | $53.24 |
| AOV | $108.28 |
| revenue (window) | $38,439.41 |
| upsell take | 55% |
paid 355/100
open page ↗
Paid orders per day (v4 v2.1a)
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
09-16
09-17
v4 is ahead · we are 93.5% sure (a winner is called at 95%)
Why v4 is ahead
v4 makes $59.73 per checkout start vs $53.24 — a 12% gap. It wins on conversion: 54% of starts buy vs 49%, and its orders are bigger too ($109.65 vs $108.28). Why: bank approval is level (68% vs 68%) — the difference is the page, not the gateway; v4 persuades more people to actually press pay (71% vs 64%). Based on 690 paid orders across 16 complete days — not yet enough to call.KEEP RUNNING
confidence 93.5% of 95%
paid floor met.
Solvane™ bracelet (neuropathy) checkout page
Checkout page: v2 (new) vs v1 (control)
16 complete days of data
Updated 10 minutes ago · 2026-09-18 12:06 GMT+7EARLY
first-ever checkout challenger (v2) vs the original v1 · window: 16 complete CC days — from 2026-09-02 23:00 (GMT+7) (both pages live together) through end of 2026-09-17. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v2 54% / v1 46% — even ✓
$12,320.73revenue — these pages (CC orders)
—RPC (no click source wired)
v2 ⭐
| page views | collecting… |
| checkout starts | 97 |
| paid orders | 58 |
| form→sale rate | 59.8% |
| tried to pay (card/PayPal) | 79% |
| card approval (the bank) | 58% · 19 declined |
| $ / checkout start | $72.98 |
| AOV | $122.06 |
| revenue (window) | $7,079.51 |
| upsell take | 64% |
paid 58/100
open page ↗
v1
| page views | collecting… |
| checkout starts | 82 |
| paid orders | 48 |
| form→sale rate | 58.5% |
| tried to pay (card/PayPal) | 67% |
| card approval (the bank) | 79% · 6 declined |
| $ / checkout start | $63.92 |
| AOV | $109.19 |
| revenue (window) | $5,241.22 |
| upsell take | 50% |
paid 48/100
open page ↗
Paid orders per day (v2 v1)
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
09-16
09-17
v2 is ahead · we are 75.8% sure (a winner is called at 95%)
Why v2 is ahead
v2 makes $72.98 per checkout start vs $63.92 — a 14% gap. It wins on order size: $122.06 average order vs $109.19, with conversion about level (60% vs 59%). Why: card approval is close (58% vs 79%), so this is the page working rather than bank luck; v2 persuades more people to actually press pay (79% vs 67%); v2 also upsells better (64% vs 50%). Based on 106 paid orders across 16 complete days — not yet enough to call.KEEP RUNNING
confidence 75.8% of 95% · 48 of 100 paid in the smaller arm
at this pace the 100-paid floor is NOT reachable by day 14 — expect a no-difference call unless traffic scales.
Schorl bracelet checkout page
Checkout page: v2 (new) vs v1.1 (control)
16 complete days of data
Updated 9 minutes ago · 2026-09-18 12:06 GMT+7STRONG LEAN
new v2 checkout vs the v1.1 control · window: 16 complete CC days — from 2026-09-02 23:50 (GMT+7) (both pages live together) through end of 2026-09-17. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v2 42% / v1.1 58% — even ✓
8,528ad clicks (fb → funnel)
$1.76RPC — rev per ad click
$1.96cost per click
−$0.20margin per click
0.90×ROAS (tracked campaigns)
$17,981.09revenue — these pages (all routes)
RPC/ROAS use the tracker's own attributed revenue ($14,998.12) and clicks via ClickFlare "schorl145" (visits=ad clicks) — one consistent universe. Per-arm decisions stay on the $ / checkout start columns.
v2 ⭐
| page views | collecting… |
| checkout starts | 149 |
| paid orders | 88 |
| form→sale rate | 59.1% |
| tried to pay (card/PayPal) | 74% |
| card approval (the bank) | 70% · 22 declined |
| $ / checkout start | $54.67 |
| AOV | $92.57 |
| revenue (window) | $8,146.47 |
| upsell take | 51% |
paid 88/100
open page ↗
v1.1
| page views | collecting… |
| checkout starts | 205 |
| paid orders | 103 |
| form→sale rate | 50.2% |
| tried to pay (card/PayPal) | 65% |
| card approval (the bank) | 60% · 30 declined |
| $ / checkout start | $47.97 |
| AOV | $95.48 |
| revenue (window) | $9,834.62 |
| upsell take | 51% |
paid 103/100
open page ↗
Paid orders per day (v2 v1.1)
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
09-16
09-17
v2 is ahead · we are 85.2% sure (a winner is called at 95%)
Why v2 is ahead
v2 makes $54.67 per checkout start vs $47.97 — a 14% gap. It wins on conversion — 59% of starts buy vs 50% — even though its orders are smaller ($92.57 vs $95.48 AOV). Why: card approval is close (70% vs 60%), so this is the page working rather than bank luck; v2 persuades more people to actually press pay (74% vs 65%). Based on 191 paid orders across 16 complete days — not yet enough to call.KEEP RUNNING
confidence 85.2% of 95% · 88 of 100 paid in the smaller arm
at this pace the 100-paid floor is NOT reachable by day 14 — expect a no-difference call unless traffic scales.
Anti-Smoking Bracelet™ checkout page
Checkout page: v3 (new) vs v2 (control)
16 complete days of data
Updated 8 minutes ago · 2026-09-18 12:08 GMT+7STRONG LEAN
new v3 checkout vs the v2 control · window: 16 complete CC days — from 2026-09-02 22:20 (GMT+7) (both pages live together) through end of 2026-09-17. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v3 40% / v2 60% — ⚠ uneven — read with caution until it balances
16,095ad clicks (fb → funnel)
$2.52RPC — rev per ad click
$2.48cost per click
+$0.04margin per click
1.02×ROAS (tracked campaigns)
$32,450.14revenue — these pages (all routes)
RPC/ROAS use the tracker's own attributed revenue ($40,585.66) and clicks via RedTrack "trusthappytoday.com" (9 campaigns w/ traffic) — one consistent universe. ⚠ These pages actually took $32,450.14 in the window — other campaigns/routes also feed them; RPC covers only the tracked ones. Per-arm decisions stay on the $ / checkout start columns.
v3 ⭐
| page views | collecting… |
| checkout starts | 284 |
| paid orders | 161 |
| form→sale rate | 56.7% |
| tried to pay (card/PayPal) | 78% |
| card approval (the bank) | 66% · 60 declined |
| $ / checkout start | $49.08 |
| AOV | $86.58 |
| revenue (window) | $13,940.13 |
| upsell take | 45% |
paid 161/100
open page ↗
v2
| page views | collecting… |
| checkout starts | 425 |
| paid orders | 229 |
| form→sale rate | 53.9% |
| tried to pay (card/PayPal) | 70% |
| card approval (the bank) | 70% · 66 declined |
| $ / checkout start | $43.55 |
| AOV | $80.83 |
| revenue (window) | $18,510.01 |
| upsell take | 43% |
paid 229/100
open page ↗
Paid orders per day (v3 v2)
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
09-16
09-17
v3 is ahead · we are 89.8% sure (a winner is called at 95%)
Why v3 is ahead
v3 makes $49.08 per checkout start vs $43.55 — a 13% gap. It wins on conversion: 57% of starts buy vs 54%, and its orders are bigger too ($86.58 vs $80.83). Why: bank approval is level (66% vs 70%) — the difference is the page, not the gateway; v3 persuades more people to actually press pay (78% vs 70%). Based on 390 paid orders across 16 complete days — not yet enough to call.KEEP RUNNING
confidence 89.8% of 95%
paid floor met.
Pixiu Mantra Ring checkout page
Checkout page: v5c vs v5b vs v5a (control) — one 3-way test, equal 33% split
17 complete days of data
Updated 12 minutes ago · 2026-09-18 12:04 GMT+7STRONG LEAN
two challengers (v5c, v5b — both derivatives of the shipped v5a winner) against the v5a control in a single CC rotation · window: 17 complete CC days — from 2026-09-01 11:53 (GMT+7) (both pages live together) through end of 2026-09-17. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v5a 34% / v5c 33% / v5b 33% — even ✓
20,246ad clicks (fb → funnel)
$1.98RPC — rev per ad click
$1.79cost per click
+$0.19margin per click
1.11×ROAS (tracked campaigns)
$144,924.71revenue — these pages (all routes)
RPC/ROAS use the tracker's own attributed revenue ($40,026.54) and clicks via ClickFlare "pixiu47" (visits=ad clicks) — one consistent universe. ⚠ These pages actually took $144,924.71 in the window — other campaigns/routes also feed them; RPC covers only the tracked ones. Per-arm decisions stay on the $ / checkout start columns.
v5a
| page views | collecting… |
| checkout starts | 1187 |
| paid orders | 479 |
| form→sale rate | 40.4% |
| tried to pay (card/PayPal) | 64% |
| approval (the bank) | 63% · 280 declined |
| $ / checkout start | $40.79 |
| AOV | $101.08 |
| revenue (window) | $48,416.03 |
| upsell take | 36% |
paid 479/100
open page ↗
v5c
| page views | collecting… |
| checkout starts | 1130 |
| paid orders | 447 |
| form→sale rate | 39.6% |
| tried to pay (card/PayPal) | 62% |
| approval (the bank) | 64% · 254 declined |
| $ / checkout start | $40.14 |
| AOV | $101.47 |
| revenue (window) | $45,359.11 |
| upsell take | 33% |
paid 447/100
open page ↗
v5b ⭐
| page views | collecting… |
| checkout starts | 1152 |
| paid orders | 505 |
| form→sale rate | 43.8% |
| tried to pay (card/PayPal) | 65% |
| approval (the bank) | 67% · 244 declined |
| $ / checkout start | $44.40 |
| AOV | $101.29 |
| revenue (window) | $51,149.57 |
| upsell take | 35% |
paid 505/100
open page ↗
Paid orders per day (v5a v5c v5b)
09-01
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
09-16
09-17
v5b is ahead · we are 88.8% sure (a winner is called at 95%)
Why v5b is ahead
v5b makes $44.40 per checkout start vs $40.79 — a 9% gap. It wins on conversion: 44% of starts buy vs 40%. Based on 984 paid orders across 17 complete days — not yet enough to call.3-way comparison is descriptive — the ship decision comes from the registered pairwise tests (no valid simultaneous N-way probability yet).
KEEP RUNNING
3-way comparison is descriptive — the ship decision comes from the registered pairwise tests
paid floor met.
Self-hosted funnels (Trust Apollo)
VitalTac flashlight checkout page
PayPal express: Legacy path, no email (v4) vs Direct path, email required (v2)
3 complete days of data
Updated 2 days ago · 2026-09-16 12:11 GMT+7STALE
v4 = byte copy of v2 whose PayPal express uses Checkout Champ's Legacy path (no email before PayPal; order created at confirm from PayPal's payer data); sticky 50/50 (localStorage vt_cv) · window: 3 complete CC days — from 2026-09-16 (both pages live together) through end of 2026-09-16. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v2 0% / v4 0% — even ✓
$0.00revenue — these pages (CC orders)
—RPC (no click source wired)
v2
| page views | 2188 |
| checkout starts | 0 |
| paid orders | 0 |
| TRUE CVR (paid/views) | 0.0% |
| $ per page view | $0.00 |
| view → form start | 0% |
| form→sale rate | 0.0% |
| tried to pay (card/PayPal) | 0% |
| approval (the bank) | 0% · 0 declined |
| $ / checkout start | $0.00 |
| AOV | $0.00 |
| revenue (window) | $0.00 |
| PayPal orders | 0 |
paid 0/100
open page ↗
v4
| page views | 1875 |
| checkout starts | 0 |
| paid orders | 0 |
| TRUE CVR (paid/views) | 0.0% |
| $ per page view | $0.00 |
| view → form start | 0% |
| form→sale rate | 0.0% |
| tried to pay (card/PayPal) | 0% |
| approval (the bank) | 0% · 0 declined |
| $ / checkout start | $0.00 |
| AOV | $0.00 |
| revenue (window) | $0.00 |
| PayPal orders | 0 |
paid 0/100
open page ↗
null is ahead · we are 50.0% sure (a winner is called at 95%)
Why null is ahead
Not enough orders yet to explain a difference.KEEP RUNNING
confidence 50.0% of 95% · 0 of 100 paid in the smaller arm
no paid orders yet in the smaller arm.
VitalTac flashlight checkout page
Checkout design: v3 (Pixiu-style) vs v2 (current)
18 complete days of data
Updated 2 days ago · 2026-09-16 12:15 GMT+7STRONG LEAN
v3 = Pixiu-v5c design reskin of v2; DESIGN-only split, sticky 50/50 (localStorage vt_cv) · window: 18 complete CC days — from 2026-09-02 01:16 (GMT+7) (both pages live together) through end of 2026-09-16. Today is never counted; CC-hosted and self-hosted cards use this same rule.
Split check: traffic share v3 45% / v2 55% — even ✓
PayPal note: a PayPal abandon counts as a checkout start but not a pay-attempt, which deflates "tried to pay" — both arms carry identical payment options so the comparison stays fair (M-010); the PayPal-orders row shows the mix.
60,193ad clicks (fb → funnel)
$4.55RPC — rev per ad click
$3.79cost per click
+$0.76margin per click
1.20×ROAS (tracked campaigns)
$279,379.92revenue — these pages (all routes)
RPC/ROAS use the tracker's own attributed revenue ($273,939.81) and clicks via RedTrack "vitaltac" (18 campaigns w/ traffic) — one consistent universe. Per-arm decisions stay on the $ / checkout start columns.
v3 ⭐
| page views | 8812 |
| checkout starts | 1470 |
| paid orders | 1080 |
| TRUE CVR (paid/views) | 12.3% |
| $ per page view | $14.58 |
| view → form start | 17% |
| form→sale rate | 73.5% |
| tried to pay (card/PayPal) | 87% |
| card approval (the bank) | 74% · 145 declined |
| PayPal share of orders | 62% |
| $ / checkout start | $87.38 |
| AOV | $118.94 |
| revenue (window) | $128,452.51 |
| PayPal orders | 908 |
paid 1080/100
open page ↗
v2
| page views | 12365 |
| checkout starts | 1786 |
| paid orders | 1293 |
| TRUE CVR (paid/views) | 10.5% |
| $ per page view | $12.21 |
| view → form start | 14% |
| form→sale rate | 72.4% |
| tried to pay (card/PayPal) | 87% |
| card approval (the bank) | 74% · 182 declined |
| PayPal share of orders | 61% |
| $ / checkout start | $84.51 |
| AOV | $116.73 |
| revenue (window) | $150,927.41 |
| PayPal orders | 1093 |
paid 1293/100
open page ↗
Paid orders per day (v3 v2)
09-01
09-02
09-03
09-04
09-05
09-06
09-07
09-08
09-09
09-10
09-11
09-12
09-13
09-14
09-15
v3 is ahead · we are 85.7% sure (a winner is called at 95%)
Why v3 is ahead
v3 makes $87.38 per checkout start vs $84.51 — a 3% gap. It wins on order size: $118.94 average order vs $116.73, with conversion about level (73% vs 72%). Why: bank approval is level (74% vs 74%) — the difference is the page, not the gateway. Based on 2373 paid orders across 18 complete days — not yet enough to call.DRAW
both arms past 100 paid and only 3.4% apart at 86% confidence — more traffic will not separate them
Nothing changes until you say so. If you agree, reply:
To end it (self-hosted): set
end vitaltac-1-checkout-v2-v3, ship v3 (or name the simpler page) and free the slot for a real challenger.To end it (self-hosted): set
SPLIT_ARMS = ["v3"] in that funnel's sales router and redeploy through the safe-deploy gate. I can do this — but only when you reply with the explicit go.Decided — final results
Aether checkout v2.1a vs v3a
startspaidCVR$/startAOV
v2.1a 🏆87944851.0%$56.35$110.55
v3a2549838.6%$40.32$104.51
Pixiu checkout v3a vs v5a
startspaidCVR$/startAOV
v3a75426935.7%$35.34$99.06
v5a 🏆81137045.6%$48.27$105.81
Pixiu checkout v4a vs v3a
startspaidCVR$/startAOV
v4a118639833.6%$34.52$102.87
v3a 🏆182867336.8%$36.53$99.23
Larkspur™
Larkspur checkout v2 challenger vs v1 control (owner-added CC split 2026-09-02; first v2 order 10:24 EDT)
startspaidCVR$/startAOV
v2 🏆20914770.3%$46.53$66.16
v11015352.5%$36.40$69.36
InfiniteFi
Funnel path: straight to checkout (advco) vs via sales page (advsales)
startspaidCVR$/startAOV
advco 🏆77858675.3%$91.17$121.05
advsales19614473.5%$80.27$109.25
GhostWing drone
Checkout design: v3 (Pixiu-style) vs v2 (current) — both with PayPal
startspaidCVR$/startAOV
v2 🏆44931971.0%$124.31$174.97
v3896876.4%$112.40$147.11