Conclusion: No correlation exists between ad-driven and organic bookings
Using Facebook Ads purchase events (MobileAppPurchases + WebsitePurchases) from AD_RAW, we found no correlation between days with ad-attributed bookings and organic booking volume. The previous "no correlation" conclusion remains unchanged.
The apparent co-occurrence observed on 4/16-17 was caused by a single user making multiple bookings, inflating the organic count. This is noise, not a structural pattern.
1. Analysis Premise
Hypothesis
Previously, the team concluded there was no correlation between ad-driven bookings and organic bookings. Recent weekend observations suggested ad bookings and organic bookings might be co-occurring. This analysis re-examines whether that conclusion has changed.
Methodology improvement: The previous version used Conversions(Ad) from _daily_data, which mixed install campaign CVs and Instagram profile visit CVs. This version isolates Facebook Ads Purchase events only from AD_RAW for accurate analysis.
2. Scale: FB Ad Booking Contribution
747
Total Bookings (474 days)
51 (6.8%)
FB Ad-Attributed Bookings
696 (93.2%)
Organic Bookings
FB ads directly generated only 6.8% of all bookings. 93% of bookings come from organic sources (direct, search, SNS, etc.). Current ads primarily drive awareness and installs, not bookings.
3. Core Analysis: FB Ad Bookings x Organic Bookings
Period
r (Pearson)
n
FB Bookings
Organic
Interpretation
2025 Q1 (Jan-Mar)
0.00
90
0
51
No FB bookings
2025 Q2 (Apr-Jun)
0.00
91
0
71
No FB bookings
2025 Q3 (Jul-Sep)
0.00
92
0
72
No FB bookings
2025 Q4 (Oct-Dec)
+0.22
92
14
234
Weak positive
2026 Q1 (Jan-Mar)
+0.02
90
31
237
No correlation
2026 Apr (1-19)
+0.16
19
6
31
No correlation
Past 30 days
-0.02
31
15
60
No correlation
Past 14 days
+0.14
15
5
28
No correlation
Across all periods, there is no significant correlation between FB ad bookings and organic bookings.
4. FB Ad Spend x Total Bookings: Period Comparison
Period
r
n
Avg FB Spend/day
Avg BK/day
Interpretation
2025 Q1 (Jan-Mar)
-0.40
90
60,545
0.6
Inverse
2025 Q2 (Apr-Jun)
-0.14
91
74,554
0.8
No correlation
2025 Q3 (Jul-Sep)
-0.03
92
59,326
0.8
No correlation
2025 Q4 (Oct-Dec)
+0.43
92
92,341
2.7
Moderate positive (peak season)
2026 Q1 (Jan-Mar)
+0.13
90
101,138
3.0
No correlation
2026 Apr (1-19)
-0.27
19
90,331
1.9
Weak inverse
Only Q4 2025 (peak season) showed a positive correlation (r=+0.43), likely driven by seasonal booking increases coinciding with higher ad spend. All other periods show no correlation or inverse correlation.
Weekly: FB Ad Spend x Bookings
Period
FB Cost vs Total BK (r)
FB Cost vs Organic (r)
Interpretation
All time
+0.43
+0.39
Seasonal bias
Last 16w
+0.21
+0.08
No correlation
Last 12w
-0.03
-0.18
No correlation
Last 8w
-0.56
-0.60
Strong inverse
In the last 8 weeks, FB ad spend and organic bookings show r=-0.60 inverse correlation. Weeks with higher spend had fewer organic bookings.
5. Lag Analysis: Delayed Effects
FB Ad Bookings(t) → Organic Bookings(t+lag)
"Does an ad booking on day t predict organic bookings N days later?"
Lag
Past 90d (r)
Past 30d (r)
Interpretation
Same day
+0.03
-0.02
No correlation
+1 day
+0.03
+0.27
Weak signal
+2 days
+0.05
+0.14
No correlation
+3 days
+0.20
+0.37
Weak signal
+5 days
+0.03
+0.18
No correlation
+7 days
-0.21
-0.11
No correlation
The Lag+3 signal (r=+0.37) in the past 30 days is noise.
The 6 organic bookings on 4/17 were from a single user making multiple reservations. On a unique-user basis, this is effectively 1 booking. This single user's bulk purchase inflated the lag correlation and does not represent a structural pattern.
FB Ad Spend(t) → Organic Bookings(t+lag)
Lag
Past 90d (r)
Past 30d (r)
Interpretation
Same day
-0.08
-0.17
No correlation
+1 day
-0.03
-0.13
No correlation
+2 days
-0.05
-0.31
Inverse
+3 days
+0.03
-0.21
Weak inverse
+5 days
-0.04
-0.30
Inverse
+7 days
-0.12
-0.49
Moderate inverse (30d)
All lags show negative correlation for ad spend. At Lag+7 days in the past 30 days, r=-0.49. This is the opposite of the SNS lag pattern (r=+0.6 at 7 days), suggesting ad spend and SNS have fundamentally different booking conversion dynamics.
6. Impact of Install Campaign Scale-Down
Install-Heavy (W04-W08, 5w)
Install-Light (W09-W15, 7w)
Change
Avg FB Installs/week
710
42
-94%
Avg FB Ad Spend/week
948,992
832,497
-12%
Avg Bookings/week
23.8
19.7
-17%
Installs dropped 94% but bookings only dropped 17%. Install campaigns have very low booking conversion efficiency. Note: seasonality (Jan-Feb vs Mar-Apr) may also contribute.
7. April Daily Breakdown
Date
FB BK
Organic
Total
FB Spend
Installs
4/1
0
1
1
127,966
3
4/2
1
1
2
124,585
2
4/3
0
0
0
112,687
4
4/4
0
1
1
121,707
3
4/5
0
0
0
136,564
1
4/6
1
0
1
147,928
9
4/7
0
1
1
124,828
6
4/8
0
1
1
77,798
4
4/9
0
3
3
46,341
4
4/10
0
3
3
46,967
4
4/11
0
2
2
48,681
2
4/12
0
0
0
63,157
4
4/13
0
3
3
64,539
0
4/14
0
0
0
54,851
3
4/15
0
2
2
47,413
2
4/16
3
2
5
81,567
6
4/17
1
6*
7
86,520
3
4/18
0
2
2
96,701
3
4/19
0
3
3
105,484
4
*4/17 organic count of 6 includes multiple bookings from a single user. Unique user increase was minimal.
4/9-11: Lowest FB spend days (46K-49K) had the most organic bookings (8 total). 4/3-6: Highest FB spend days (113K-148K) produced only 1 booking.
Ad spend level does not determine booking occurrence.
8. Weekly Trend (Last 16 Weeks)
Week
FB Spend
FB BK
Organic
Total
Installs
W00 (late Dec)
319,142
1
13
14
497
W01
415,486
1
8
9
1,006
W02
551,238
0
12
12
1,332
W03
851,479
3
10
13
1,352
W04
1,013,274
10
27
37
936
W05
874,389
1
15
16
805
W06
935,921
1
15
16
596
W07
841,282
1
22
23
591
W08
597,818
0
27
27
433
W09
478,822
1
31
32
93
W10
538,767
1
19
20
29
W11
677,204
8
15
23
47
W12
743,561
2
17
19
35
W13
887,512
2
9
11
30
W14
555,700
1
10
11
33
W15
537,075
4
18
22
21
W04 (FB spend over 1M) had 37 bookings total, but only 10 from FB ads — 27 were organic.
W09 (FB spend 479K) had 32 bookings with organic at its peak of 31. Doubling ad spend does not proportionally increase bookings.
9. Final Conclusion
Previous "no correlation" conclusion remains unchanged
Test
Result
FB ad bookings x organic bookings (daily)
r = -0.02 (past 30d). No correlation.
FB ad spend x organic bookings (weekly, last 8w)
r = -0.60. Inverse correlation.
Lag effect (FB bookings → organic increase N days later)
Lag+3 r=+0.37 signal was noise from 1 user's multiple bookings.
Install scale-down impact
Installs -94% but bookings only -17%. Low install-to-booking efficiency.
FB ad direct booking contribution
51 of 747 total bookings (6.8%). 93% are organic.
10. Next Actions
Priority
Action
Rationale
1
Focus on LP / Web CVR improvement
Increasing ad spend does not increase bookings. Fixing the landing experience (Web CVR 0.1%) is the priority
2
Fix UTM tracking (broken since 1/16)
Without proper attribution, accurate ad effectiveness measurement is impossible
3
Redefine ad role: awareness/install, not booking acquisition
Ads contribute only 6.8% of bookings. Evaluate as awareness investment with adjusted KPIs
4
Compare with SNS lag correlation
Ad lag is negative; SNS lag is positive (r=+0.6 at 7 days). SNS may be more effective for booking conversion
5
Introduce unique-user based booking analysis
Multi-booking by single users distorts data. UU-based metrics needed for accurate decision-making
Assumptions: Ad data filtered to Facebook Ads only (ServiceName = "Facebook Ads") from AD_RAW. Booking events: MobileAppPurchases (col N) + WebsitePurchases (col P). Organic bookings = Total - FB bookings (approximation; other platform bookings included but Google/TikTok had 0 CV in April). Correlation coefficients for small-sample periods should be treated as indicative.