Of the two levers identified in the previous report — (1) Ads: secure volume, (2) CRM: boost conversion rate — Lever (1) (Install ad restart) has been executed and CPI efficiency is better than expected (¥400 vs. projected ¥880). However, Lever (2) (CRM initiatives) remains untouched after 10 days, and the CA-to-booking conversion rate continues to halve month-over-month (39.2% → 27.1% → 15.6%). Even with increased installs, users cannot be converted to bookings. The structure is "turning on the faucet while the bucket has a hole." Immediate action on Phase 1 CRM initiatives is the top priority.
| Phase | Description | Status | Impact |
|---|---|---|---|
| Phase 1 | CRM Optimization (Coupon UX, Push expansion, Email recovery) | ⚠️ Not Started | Expected: +26-31 bookings/month → Unrealized |
| Phase 2 | Install Ad Restart (ContentView optimization) | ✓ Started 4/29 | CPI ¥400 (less than half of projection). Installs +109%. However, zero booking conversions |
| Phase 3 | Messaging & Creative Testing (LP × Creative) | 🔄 Partially Started | Bar Hopping LP launched (4/30). Impact assessment pending |
| Phase 4 | Product Integration | — Not Started | — |
| Phase 5 | Service Alignment → Day 10 Bookings | — Not Started | — |
Because Phase 1 CRM initiatives remain unstarted, there is no mechanism to convert the users acquired through Phase 2 into bookings. The previous report positioned Phase 1 as "quick-win measures" designed to boost conversion rates before Phase 2 effects materialized (2-week lag). This sequencing delay is directly reflected in current results.
APP_INSTALLS CPI: ¥400 (less than half of projected ¥880). 277 installs over 7 days (39.6/day).
| Country | CPI |
|---|---|
| AU | ¥1,129 |
| US | ¥2,075 |
| NZ | ¥690 |
ENGAGEMENT budget significantly reduced from ¥1.13M to ¥94K — in line with report recommendations.
APP_INSTALLS campaign CPI efficiency significantly exceeded expectations. ENGAGEMENT budget reduction was also implemented in accordance with report recommendations.
| Metric | APP_INSTALLS (May) | OUTCOME_SALES (May) | OUTCOME_SALES (Apr) |
|---|---|---|---|
| Installs | 277 | 10 | 61 |
| CPI | ¥400 | ¥38,120 | ¥23,120 |
| In-App ContentView/Install | 141% | 1,390% | 903% |
| App Purchase | 0 | 2 | 7 |
| App Cart Add | 0 | 2 | 11 |
| App Search/Install | 91% | 420% | 402% |
APP_INSTALLS In-App ContentView rate is 141%, compared to 1,390% for OUTCOME_SALES — roughly 1/10th. The previous report's finding that "WEB Conv-sourced installs show 7x the in-app engagement rate of direct installs" is reproduced exactly. CPI efficiency is strong, but there is a significant quality gap in post-install in-app behavior.
Install ad restart was on 4/29. The previous report's correlation analysis showed an Install-to-Booking effect at 12-day lag (r=0.70). This means effects may begin to appear around 5/11-13.
However: Full-period AF Install-to-Booking correlation has shifted to r=-0.018 (near zero). Whether the post-period-only r=0.66-0.76 still holds requires verification.
Around 5/13 is the first decision point. If bookings do not increase by then, it is highly likely that APP_INSTALLS-sourced users "can be acquired in volume but do not convert to bookings." In that case, budget reallocation toward WEB Conv (OUTCOME_SALES) should be considered.
| Campaign | Installs | Sessions | Sess/Inst |
|---|---|---|---|
| [iOS]APP_Conversion | 116 | 419 | 3.6 |
| [Android]APP_Conversion | 149 | 542 | 3.6 |
| [WEB]Conversion_Main | 5 | 17 | 3.4 |
| ASA | 14 | 47 | 3.4 |
| Country | Installs | Sess/Inst |
|---|---|---|
| US | ~125 | 3.4-3.5 |
| AU | ~99 | 3.7 |
| CA | ~27 | 3.7 |
| NZ | ~10 | 3.8 |
Based on AppsFlyer Sessions/Install, the quality gap between APP_Conversion and WEB Conversion is minimal (3.4-3.7). However, Meta CSV In-App ContentView/Install rates show APP_INSTALLS at 141% vs. OUTCOME_SALES at 1,390% — a roughly 10x difference.
This discrepancy is due to different measurement layers: AppsFlyer measures sessions (app opens), while Meta measures ContentView (tour detail views). Sessions only require opening the app; ContentView requires active engagement. In other words, APP_Conversion users "open the app but don't browse tours" — suggesting an onboarding problem.
| Month | Traffic | CA | Bookings | Traffic→CA | CA→Booking | Traffic→Booking |
|---|---|---|---|---|---|---|
| 2026/03 | 12,833 | 265 | 104 | 2.1% | 39.2% | 0.8% |
| 2026/04 | 8,998 | 192 | 52 | 2.1% | 27.1% | 0.6% |
| 2026/05 | 2,202 | 45 | 7 | 2.0% | 15.6% | 0.3% |
CA-to-booking conversion rate is roughly halving every month. Traffic-to-CA rate remains stable at 2.0-2.1%, meaning the problem lies downstream of CA. The CA flow change (2/20) flagged in the previous report may be having a prolonged impact.
| Step | Mar | Apr | May (1-7) | Mar→Apr | Apr→May |
|---|---|---|---|---|---|
| View Home | 4,336 | 3,172 | 1,072 | -27% | -66% |
| View Tour Detail | 1,756 (40.5%) | 1,093 (34.5%) | 373 (34.8%) | -38% | -66% |
| Create Account | 184 (10.5%) | 116 (10.6%) | 30 (8.0%) | -37% | -74% |
| Complete Booking | 33 (17.9%) | 19 (16.4%) | 2 (6.7%) | -42% | -89% |
| Overall CVR | 0.76% | 0.60% | 0.19% |
* Percentages in parentheses are step-to-step conversion rates
View Tour Detail → TL Detail Select → Tap Booking → Confirmation → Complete Booking
| Step | Mar | Apr | May (1-7) |
|---|---|---|---|
| View Tour Detail Page | 1,697 | 1,073 | 367 |
| View TL Detail Select | 115 (6.8%) | 75 (7.0%) | 24 (6.5%) |
| Tap Booking | 41 (35.7%) | 33 (44.0%) | 7 (29.2%) |
| View Confirmation | 23 (56.1%) | 17 (51.5%) | 1 (14.3%) |
| Complete Booking | 7 (30.4%) | 5 (29.4%) | 0 (0%) |
Collapse point identified: Tap Booking → View Confirmation
Conversion rate collapsed from 56.1% → 51.5% → 14.3%. In May, 7 users reached Tap Booking, but only 1 reached Confirmation. 6 users dropped off at this step.
The CA flow change (2/20) moved account creation to just before booking, meaning a CA (registration) flow is likely inserted after Tap Booking. Unregistered users who tap "Book" are then asked to create an account and drop off at that point.
Three items to confirm with the development team:
| Objective | Cost | Installs | In-App CV/Inst | Purchase | CPI / CPO |
|---|---|---|---|---|---|
| APP_INSTALLS | ¥110,951 | 277 | 141% | 0 | CPI ¥400 |
| OUTCOME_SALES | ¥381,239 | 10 | 1,390% | 4 | CPO ¥95,310 |
| ENGAGEMENT | ¥94,326 | 1 | 0% | 0 | — |
| Total | ¥586,516 | 288 | — | 4 | CPO ¥146,629 |
OUTCOME_SALES is the only objective generating Purchases. The previous report recommended "adjusting ENGAGEMENT budget share from 44% to 10-15%." ENGAGEMENT was reduced from ¥1.13M to ¥94K (91.7% cut). However, the freed budget shifted to APP_INSTALLS, and overall booking efficiency has not improved.
| Country | Cost | Installs | CPI | Purchase |
|---|---|---|---|---|
| US | ¥300,876 | 145 | ¥2,075 | 3 |
| AU | ¥119,622 | 106 | ¥1,129 | 0 |
| CA | ¥72,304 | 25 | ¥2,892 | 0 |
| GB | ¥37,904 | 2 | ¥18,952 | 1 |
| Others | ¥55,810 | 10 | — | 0 |
The previous report recommended "AU priority → US." CPI efficiency favors AU (¥1,129) over US (¥2,075), but Purchases came only from US (3). AU has strong install efficiency but faces conversion challenges.
| Placement | Cost | CPC | Installs | Purchase | Previous Month CPC |
|---|---|---|---|---|---|
| Feed | ¥266,765 | ¥37 | 38 | 2 | ¥53 |
| Stories | ¥136,259 | ¥24 | 17 | 2 | ¥28 |
| Reels | ¥117,851 | ¥119 | 71 | 0 | ¥72 |
| FB Reels | ¥38,650 | ¥118 | 95 | 0 | ¥98 |
| FB Feed | ¥22,284 | ¥89 | 62 | 0 | ¥91 |
Reels CPC increased 65% from ¥72 to ¥119, while Feed CPC improved from ¥53 to ¥37. Purchases came only from Feed (2) and Stories (2). Reels/FB Reels/FB Feed generated many installs but zero Purchases — these are APP_INSTALLS campaign placements.
| Segment | Cost | Purchase | Cost per Purchase |
|---|---|---|---|
| female 35-44 | ¥49,118 | 2 | ¥24,559 |
| male 25-34 | ¥99,567 | 1 | ¥99,567 |
| female 25-34 | ¥72,116 | 1 | ¥72,116 |
Purchases came only from the 25-44 age range, consistent with the previous report's recommendation (25-54 targeting). However, sample sizes are too small for statistical conclusions.
| Media | Cost | Key Metric | CV | Verdict |
|---|---|---|---|---|
| Google Search General EN | ¥261,000 | CPC ¥1,067 | 0 | Pause recommended |
| TikTok | ¥72,000 | organicPost boost | 0 | Pause recommended |
| ASA | ¥40,000 | CPI ¥1,613 | 25 installs | Continue |
Google Search General EN (CPC ¥1,067) has extremely poor cost efficiency. TikTok organicPost boost also has zero CV and needs a continuation decision. ASA at CPI ¥1,613 is within acceptable range — recommended to continue.
| Campaign | Cost | Clicks | CPC | CV | Verdict |
|---|---|---|---|---|---|
| Brand EN | ¥121,309 | 438 | ¥277 | 2.0 | ✔️ Continue |
| General EN ② In-Market | ¥252,390 | 212 | ¥1,191 | 0 | ❌ Pause recommended |
| General EN ③ Experience | ¥107,434 | 96 | ¥1,119 | 0 | ❌ Pause recommended |
| General EN ④ Seasonal | ¥13,482 | 53 | ¥254 | 0 | ❌ Pause recommended |
| General EN ① Awareness | ¥2,365 | 2 | ¥1,182 | 0 | ❌ Pause recommended |
| General JP ② In-Market | ¥50,359 | 143 | ¥352 | 0 | ⚠️ Needs discussion |
| General JP ③ Experience | ¥44,820 | 104 | ¥431 | 0 | ⚠️ Needs discussion |
| General JP ④ Seasonal | ¥11,137 | 35 | ¥318 | 0 | ⚠️ Needs discussion |
| Search Query | Cost | Clicks | CPC | CV |
|---|---|---|---|---|
| private tour guide tokyo | ¥17,143 | 8 | ¥2,143 | 0 |
| private tours of japan | ¥15,690 | 1 | ¥15,690 | 0 |
| tokyo private tours | ¥14,238 | 4 | ¥3,560 | 0 |
| local guides in japan | ¥10,938 | 1 | ¥10,938 | 0 |
| tokyo bar tour | ¥9,243 | 7 | ¥1,320 | 0 |
| private tour guide in osaka japan | ¥7,098 | 3 | ¥2,366 | 0 |
| trip to japan | ¥6,405 | 2 | ¥3,202 | 0 |
| tokyo nightlife tour | ¥6,344 | 3 | ¥2,115 | 0 |
| shinjuku bar hopping tour | ¥6,111 | 2 | ¥3,056 | 0 |
| osaka tours by locals | ¥5,999 | 5 | ¥1,200 | 0 |
General EN total: ¥375,671 spent, 0 conversions. CPC of ¥1,100-1,200 is 4x that of Brand (¥277). High-CPC queries like "private tour" and "local guide" are driving up costs. Immediately pausing General EN frees ¥376K/month (¥12.5K/day) for reallocation to Meta APP_Conversion or CRM initiatives.
Previous report (post-period only): Install → BK r=0.66-0.76 (7-day lag)
Recalculation with full-period data:
| Metric | Same-Day Correlation | Optimal Lag | Optimal Correlation |
|---|---|---|---|
| View Home (APP) | r=0.574 | 0 days | r=0.574 |
| Create Account (APP) | r=0.410 | 3 days | r=0.423 |
| Page Viewed (WEB) | r=0.345 | 7 days | r=0.466 |
| Start Session (WEB) | r=0.308 | 6 days | r=0.419 |
| Ad Cost | r=0.343 | 1 day | r=0.351 |
| AF Installs | r=-0.018 | 5 days | r=-0.053 |
AF Installs booking correlation has shifted to r=-0.018 (near-zero to slightly negative). This is a significant deterioration from the post-period-only analysis (r=0.66-0.76). With full-period data, more installs do not guarantee more bookings. Meanwhile, View Home (r=0.574) and Create Account APP (r=0.41) maintain positive correlations.
This correlation shift may be influenced by the qualitative change in installs since April (shift from ENGAGEMENT-heavy to APP_INSTALLS). The post-period-only correlation needs to be recalculated to confirm whether a structural change has occurred.
Install ads restarted on 4/29. The 12-day lag peak effect is expected around 5/11-13.
| Scenario | Condition | Action |
|---|---|---|
| Bookings increase | Daily bookings ≥ 2.0 as of 5/13 | Consider expanding APP_INSTALLS budget (CA-to-booking conversion improvement still needed separately) |
| Bookings flat | Daily bookings < 1.5 as of 5/13 | Conclude APP_INSTALLS-sourced users do not convert to bookings. Concentrate budget on OUTCOME_SALES |
| Media | Decision | Rationale |
|---|---|---|
| Google Search General EN | Pause recommended | CPC ¥1,067, CV 0 |
| TikTok organicPost | Pause recommended | ¥38K, CV 0 |
| Google Brand EN + General JP | Room to continue | Branded search defense + JP market test |
| ASA | Continue | CPI ¥1,613, strong install performance |
Phase 1 (CRM optimization) from the previous report was designed as "quick-win measures." Ten days have passed without action, and this delay is a contributing factor to the booking slump. The following should be initiated this week:
| Phase | Original Schedule | Revised | Status |
|---|---|---|---|
| Phase 1 CRM | Week 1-2 (4/28-5/11) | Start this week → Key initiatives implemented by 5/15 | ⚠️ Delayed |
| Phase 2 Install | Week 2-4 (5/5-5/25) | In progress. Lag effect verification on 5/13 | ✓ In Progress |
| Phase 2b Install Quality | — (New) | Decision on 5/13. Compare APP_INSTALLS vs OUTCOME_SALES booking conversion | 🆕 Added |
| Phase 3 Creative Testing | Week 4-8 | Bar Hopping LP first. Other LPs from 5/19 onward | 🔄 Partially Started |
| KPI | Current (5/7) | Phase 1 Target | Post-Phase 2 | Final Target |
|---|---|---|---|---|
| Daily Bookings | 1.0 | 2.0 | 3.0-4.0 | 10 |
| Weekly Installs | 534 | Maintain | 250-500 | 500 |
| CA-to-Booking CVR | 15.6% | 25%+ | 30%+ | 40%+ |
| Coupon Utilization Rate | 24.3% | 35% | 40% | 50% |
| Push Bookings | 5/month | 15/month | 20/month | 50/month |