1 Executive Summary
April Bookings
42
-59.6% MoM
April CPO
¥84,407
+160% MoM
Female Installs
38
-96% vs Feb
Web CVR
0.18%
1/7th of App (1.22%)
CRITICAL ALERT
Root cause of booking decline identified: Install campaign budget was eliminated (35% → 1% of total spend). The halving of CPC is NOT an improvement—it reflects a fundamental shift in optimization objective from Install to Click/Follower. Female installs, the only proven leading indicator for bookings (r=0.70), have collapsed from ~1,000/month (Feb) to 38 (Apr).
KEY INSIGHT
Every major performance metric degradation traces back to a single cause: the reallocation of budget away from install-optimized campaigns. Restoring install campaign spend is the highest-leverage action available.
2 Booking Volume Grouping Analysis
Days grouped by booking count reveal stark efficiency differences. The data below covers the full Sep 2025–Apr 2026 period.
| Group |
Bookings/Day |
Days |
% of Days |
Avg Bookings |
Avg Cost/Day |
CPO |
CVR |
ROAS |
| A |
0 bookings |
75 |
31.5% |
0.0 |
¥78,243 |
∞ |
0.00% |
0.0% |
| B |
1–2 bookings |
112 |
47.1% |
1.4 |
¥95,617 |
¥68,298 |
0.82% |
30.4% |
| C |
3–4 bookings |
38 |
16.0% |
3.4 |
¥117,832 |
¥34,657 |
1.71% |
56.2% |
| D |
5+ bookings |
13 |
5.5% |
5.8 |
¥160,405 |
¥27,656 |
2.48% |
85.8% |
KEY FINDING
D-group days (5+ bookings) achieve CPO ¥27,656 and ROAS 85.8%—3x more efficient than B-group days. The paradox: spending less total budget doesn't improve efficiency. Days with higher spend AND higher bookings are the most efficient, because the incremental cost of each additional booking decreases sharply above 3 bookings/day.
3 Female Install Lag Correlation
We tested correlation between 4 female ad metrics (Reach, Impressions, Clicks, Installs) and future bookings at lags of 0–21 days. Only Install shows meaningful positive correlation.
| Metric |
Peak r |
Optimal Lag |
Direction |
Interpretation |
| Female Install |
+0.70 |
+12 days |
POSITIVE |
Strong positive — the ONLY positive leading indicator |
| Female Click |
-0.77 |
+5 days |
NEGATIVE |
More clicks = fewer bookings (click-optimized = wrong audience) |
| Female Reach |
≈0.00 |
N/A |
NONE |
No correlation at any lag |
| Female Impressions |
-0.15 |
+3 days |
WEAK NEG |
Negligible |
Rolling 7-Day Female Installs by Booking Group
| Lag Window |
D-Group (5+) |
C-Group (3-4) |
B-Group (1-2) |
A-Group (0) |
D/A Ratio |
| -14 to -20 days prior |
57.2 |
32.1 |
18.4 |
8.9 |
6.4x |
| -7 to -13 days prior |
48.6 |
28.7 |
15.2 |
10.1 |
4.8x |
| -1 to -6 days prior |
35.4 |
22.3 |
14.8 |
11.6 |
3.1x |
CRITICAL
Female Install is the ONLY positive leading indicator for bookings. More reach does NOT mean more bookings. More clicks actually correlate with FEWER bookings. The install-to-booking pipeline has a ~12-day lag, meaning today's install cuts will fully manifest as booking declines ~12 days from now.
4 Monthly Trend: February → March → April
| Metric |
February |
March |
April (to 26th) |
Feb → Apr Δ |
| Bookings |
104 |
104 |
42 |
-59.6% |
| Female Installs |
~1,000 |
248 |
38 |
-96.2% |
| Female CPC |
¥152 |
¥98 |
¥74 |
-51.3% |
| Female Reach |
89,200 |
124,500 |
98,700 |
+10.7% |
| Female Clicks |
6,580 |
10,200 |
13,500 |
+105% |
| CPO |
¥32,455 |
¥32,455 |
¥84,407 |
+160% |
| Install Campaign % |
35% |
12% |
1% |
-34pp |
| Follower Campaign % |
0% |
15% |
28% |
+28pp |
BUDGET SHIFT ANALYSIS
Install campaigns went from 35% to 1% of total budget. Follower campaigns grew from 0% to 28%. CPC fell 51% because click/follower optimization targets cheaper (but less qualified) traffic. Clicks doubled while installs crashed 96%—a textbook case of optimizing for the wrong metric.
5 Meta Ad Segment Analysis
5a. Gender Segmentation
| Day Type |
Female CPC |
Male CPC |
Premium % |
Note |
| B-Group Days (1-2 bookings) |
¥118 |
¥104 |
+13.5% |
Female audience costs more on low-volume days |
| D-Group Days (5+ bookings) |
¥96 |
¥91 |
+5.5% |
Gender gap nearly disappears on high-booking days |
EVIDENCE
The female CPC premium shrinks from 13% to 5% on D-group days. High-volume days achieve better female targeting efficiency, suggesting the algorithm finds better female audiences at scale.
5b. Age Segmentation
| Age Group |
CPC |
Install Rate |
Booking Share |
Assessment |
| 18-24 |
¥95 |
2.1% |
8% |
Low volume, low cost |
| 25-34 |
¥78 |
3.8% |
42% |
BEST SEGMENT Cheapest CPC, highest conversion |
| 35-44 |
¥112 |
2.9% |
31% |
Good volume, moderate cost |
| 45-54 |
¥134 |
1.7% |
14% |
Expensive, low efficiency |
| 55+ |
¥148 |
0.9% |
5% |
Not recommended for scale |
5c. Country Segmentation
| Country |
CPO (Cost/Purchase) |
Install-to-Purchase |
Booking Share |
Flag |
| Australia |
¥186,000 |
4.2% |
18% |
BEST ROAS Half the cost of US |
| United States |
¥362,000 |
2.1% |
35% |
Largest volume, but expensive |
| United Kingdom |
¥278,000 |
2.8% |
15% |
Moderate efficiency |
| Canada |
¥— |
0.0% |
0% |
WARNING Installs but zero purchases |
| Other |
¥245,000 |
1.5% |
32% |
Mixed bag |
5d. Placement Analysis
| Placement |
CPC |
CTR |
Install Rate |
Efficiency |
| Stories |
¥27 |
1.8% |
2.4% |
3x CHEAPER than Feed |
| Reels |
¥42 |
2.1% |
2.8% |
Strong performer |
| Feed |
¥168 |
0.6% |
3.1% |
Highest install rate but very expensive reach |
| Explore |
¥89 |
0.9% |
1.9% |
Mid-range |
6 Creative Analysis
Top Female Install Creatives (All Currently Stopped)
| Creative |
Female Installs |
Female % |
CPI |
Active Period |
Status |
| LOST Brenden |
1,860 |
62% |
¥285 |
17 weeks |
STOPPED |
| Yanaka Tour Ad |
723 |
74% |
¥312 |
15 weeks |
STOPPED |
| Secret Tokyo Experience |
548 |
68% |
¥298 |
12 weeks |
STOPPED |
| Omakase Discovery |
412 |
65% |
¥341 |
8 weeks |
STOPPED |
CREATIVE LIFESPAN
Tour-specific creatives (Yanaka, Secret Tokyo) sustain performance for 15–17 weeks before fatigue. Generic creatives last only 4–8 weeks. The top-performing creatives by female install volume have all been deactivated due to the campaign mix shift, not due to performance degradation.
7 User Journey Analysis
7a. Account-to-Booking Speed
FINDING
46–51% of users book within 0–1 day of creating their account. The app experience is highly conversion-optimized for immediate bookers. However, this also means delayed-install users (from click/follower campaigns) face a steep drop-off.
7b. Repeat Booking Behavior
| Day Type |
Repeat Bookers % |
First-Time Bookers % |
Implication |
| D-Group Days (5+) |
44.6% |
55.4% |
High-volume days driven by loyal users returning |
| B-Group Days (1-2) |
26.0% |
74.0% |
Low-volume days rely on new user acquisition |
7c. Top Tour Combinations
Omakase + Secret Tokyo together account for 26% of March bookings. These are the gateway experiences that drive initial engagement and repeat purchases.
7d. Booking Time Distribution (JST)
| JST Time |
US Eastern |
Booking Share |
Insight |
| 08:00–13:00 JST |
6:00–11:00 PM EST (prev day) |
42% |
US evening browsing → booking for Japan morning |
| 13:00–18:00 JST |
11:00 PM–4:00 AM EST |
22% |
Mixed US late-night / AU morning |
| 18:00–23:00 JST |
4:00–9:00 AM EST |
24% |
US morning / EU afternoon |
| 23:00–08:00 JST |
9:00 AM–6:00 PM EST |
12% |
US daytime (working hours, low intent) |
7e. Cancellation Rates
| Day Type |
Cancel Rate |
Interpretation |
| D-Group Days (5+) |
22% |
Higher-intent users, more repeat bookers |
| B-Group Days (1-2) |
36% |
Lower intent, more impulsive/uncertain bookings |
8 Amplitude Funnel Analysis
Web Funnel vs App Funnel
| Step |
Web UU |
Web Drop |
App UU |
App Drop |
| Landing / Home |
541 |
— |
1,807 |
— |
| Tour Detail View |
287 |
-47% |
892 |
-51% |
| Date Selection |
26 |
-91% |
198 |
-78% |
| Payment |
8 |
-69% |
54 |
-73% |
| Booking Complete |
1 |
-88% |
22 |
-59% |
| Overall CVR |
0.18% |
1.22% |
CRITICAL BOTTLENECK
Web: Detail → Date Selection = 91% drop-off. This is the single largest leak in the entire funnel. App performs 6.8x better at this step (78% drop vs 91%). The web experience likely lacks the tour availability/calendar UX that the app provides natively.
Additional Amplitude Metrics
| Metric |
Value |
Trend |
| Top Viewed Tour |
Secret Tokyo |
#1 by view count across both web and app |
| Weekly App Registrations |
30/week |
Declining: 59 → 30/week over 4 weeks (-49%) |
| App DAU (7d avg) |
142 |
Stable but low |
9 Budget Optimization Analysis
| Daily Spend Bracket |
Avg Bookings/Day |
CPO |
ROAS |
Assessment |
| <¥50K |
0.4 |
¥125,000 |
15.2% |
TOO LOW Insufficient for algorithm learning |
| ¥50–80K |
1.1 |
¥63,636 |
28.7% |
Baseline viable |
| ¥80–100K |
2.4 |
¥37,500 |
48.6% |
SWEET SPOT Best efficiency |
| ¥100–160K |
2.6 |
¥50,000 |
36.4% |
DEAD ZONE 40-80% more spend, same bookings |
| ¥160–200K |
3.2 |
¥53,125 |
35.1% |
Transition zone |
| >¥200K |
4.8 |
¥41,667 |
42.3% |
VOLUME PLAY Needed for 4+ bookings/day |
BUDGET STRATEGY
¥80–100K/day is the efficiency sweet spot (ROAS 48.6%). The ¥100–160K range is a "dead zone"—you spend 40–80% more but get nearly the same booking volume. To achieve 4+ bookings/day consistently, daily spend needs to exceed ¥200K, but this should only be pursued with install-optimized campaigns running.
10 Top 10 Recommended Actions
-
Restore Install Campaign Budget to 30%+ of Total Spend
P0 — IMMEDIATE
The single highest-impact action. Female install is the only proven positive leading indicator (r=0.70). Current allocation of 1% is catastrophically insufficient.
-
Reactivate Top Female Install Creatives
P0 — IMMEDIATE
LOST Brenden (1,860 installs) and Yanaka Tour Ad (74% female ratio) were stopped due to campaign mix change, not performance degradation. Reactivate immediately.
-
Reduce Follower Campaign Budget from 28% to <10%
P0 — IMMEDIATE
Follower campaigns show negative correlation with bookings. The audience they attract does not convert. Reallocate to install campaigns.
-
Target ¥80–100K/day Install Spend as Minimum Floor
P1 — THIS WEEK
This bracket achieves the best ROAS (48.6%). Below ¥80K, the algorithm cannot optimize effectively. Avoid the ¥100–160K dead zone.
-
Concentrate on 25–34 Female Audience Segment
P1 — THIS WEEK
Cheapest CPC (¥78), highest install rate (3.8%), and 42% of all bookings. This is the core audience.
-
Increase Australia Targeting Weight
P1 — THIS WEEK
Australia CPO (¥186K) is half of US (¥362K). Consider excluding Canada entirely (installs but zero purchases).
-
Prioritize Stories/Reels Placements
P2 — THIS MONTH
Stories CPC ¥27 is 3x cheaper than Feed ¥168. Shift creative production toward vertical video formats optimized for Stories and Reels.
-
Fix Web Date Selection Drop-off (91% Loss)
P2 — THIS MONTH
The web funnel loses 91% of users at the Detail → Date Selection step. Implement availability calendar, social proof ("X booked today"), and simplified date picker. App is 6.8x better at this step.
-
Develop 2–3 New Tour-Specific Creatives
P2 — THIS MONTH
Tour-specific creatives last 15–17 weeks vs 4–8 for generic. Focus on Secret Tokyo and Omakase (26% of bookings combined). Create female-centric narratives.
-
Schedule Ad Delivery for US Evening Hours (6–11PM EST)
P2 — THIS MONTH
42% of bookings occur at JST 08:00–13:00 (= US evening). Concentrate ad delivery during this window to capture peak intent.