Project Valentine: TrafficJunky × Everflow Performance
Period: Apr 19 to Apr 26, 2026 · 20 paid in-App call campaigns (Sophie Dee + Lena the Plug × 5 versions × 2 domains)
Disclaimer: Aggregated data pulled via TJ + Everflow APIs. Numbers are subject to ongoing accuracy verification. Click-loss explanations and per-creative breakdowns require deeper investigation before strategic decisions. Use as directional, not as source of truth for finance reconciliation.
Headline KPIs
TJ Spend
$6,342
across 20 campaigns
Revenue Tracked
$55
from 3 conversions
ROAS
0.87%
brand campaign, not direct response
TJ Impressions
90.8M
eCPM ~$0.07
TJ Clicks Served
29,094
CTR 0.032% · eCPC $0.22
Everflow Clicks Recorded
23,224
−5,870 (20.2% loss)
Events Fired
261
$24 per event
Real-money Conversions
3
$2,114 per conversion
Funnel: TJ Impression → Everflow Revenue
TJ Impressions
100%
90,840,013
baseline
TJ Clicks Served
0.032%
29,094
CTR 0.032%
Everflow Clicks
79.8%
23,224
−20.2% drop
Events Fired
1.12%
261
click→event
Paid Conversions
1.15%
3
event→conv
Revenue Tracked
$18 / conv
$55.00
ROAS 0.87%
Note: bar widths are illustrative (clamped for visibility); see absolute numbers and % column for precise values. Bars are NOT proportional to actual funnel drops.
Click Loss per Campaign (TJ → Everflow)
Per-Campaign Detail
| Campaign |
uid |
TJ Imp |
TJ Clk |
Spend |
EF Clk |
Loss % |
Events |
Conv |
URL Hygiene |
Issues Found
🟡 Investigate: Sophie V1 justsext has 31.2% loss with clean URLs
Among campaigns with no URL hygiene issues, "Sophie V1 justsext" stands out at 31.2% loss versus ~17-20% baseline elsewhere. All 1,000 ads use proper tracking URL. Cause unclear.
Hypotheses (untested): geo, spot, or device-mix dropping more during yk4vn1sp redirect, or bot filter on Everflow side. Requires deeper drill into spot-level data.
⚪ Context: Project Valentine is brand campaign, not direct-response
$2,114 CPA and $24 CPL look high but are consistent with billboard + amplifier brand-awareness goals. Real ROI of this campaign should be measured by lift in branded search, organic signups, and creator-ecosystem effects (not by direct conversions in the TJ window).
Loss Investigation: Where did 5,870 clicks go?
The 20.2% TJ → Everflow gap (5,870 clicks) decomposes into three sources:
| Source | Approx Clicks | Status |
|---|
| Everflow invalid filter (antifraud rejected: bots, proxies, suspicious patterns from Turkey, Indonesia, Switzerland etc) | ~310 | Working as intended |
| Structural drop (Safari ITP blocking redirect cookies, slow mobile network, ad blockers, browser privacy extensions) | ~5,560 | Industry-normal for adult vertical |
Geo Distribution (offer 19, period)
| Country | Clicks | % of Total | Events |
|---|
| 🇺🇸 United States | 21,716 | 75.1% | 257 |
| 🇨🇦 Canada | 7,193 | 24.9% | 101 |
| 🇬🇧 United Kingdom | 10 | 0.0% | 0 |
| 🇦🇺 Australia | 5 | 0.0% | 1 |
| Plus ~310 invalid clicks from Turkey, Indonesia, Switzerland, Taiwan, France, Hong Kong (100% rejected by Everflow antifraud, likely VPN/proxy traffic). |
Device + Browser Distribution
| Device | Clicks | Share |
|---|
| Mobile | 16,945 | 58.6% |
| PC | 9,911 | 34.3% |
| Tablet | 1,951 | 6.7% |
| Other (TV/Gaming) | 117 | 0.4% |
| Browser | Clicks | Share |
|---|
| Chrome | 15,757 | 54.5% |
| Safari (iOS WebKit) | 7,361 | 25.4% |
| Edge | 1,632 | 5.6% |
| Google for iOS (iOS WebKit) | 1,095 | 3.8% |
| Samsung Browser (Android) | 1,054 | 3.6% |
| DuckDuckGo for iOS (iOS WebKit) | 668 | 2.4% |
| Other (Kindle-Silk, Opera, Firefox) | 1,357 | 4.7% |
🚨 Critical: iOS WebKit Funnel Drop
Splitting the funnel by browser engine reveals iOS Safari users convert 2.5x worse than everyone else:
| Segment | EF Clicks | % of total | Events | Click → Event CR |
|---|
| Non-iOS (Chrome, Edge, Samsung, Firefox, etc) | 20,095 | 69.5% | 309 | 1.54% |
| iOS WebKit (Safari, Google for iOS, DuckDuckGo for iOS, Opera Mini) | 8,829 | 30.5% | 55 | 0.62% |
| Best browser: | Edge (PC desktop): 1,593 clicks → 36 events = 2.26% CR |
| Worst browser: | Safari (iOS): 7,361 clicks → 35 events = 0.48% CR (3.2x worse than Chrome) |
Likely cause: Safari Intelligent Tracking Prevention (ITP) blocks third-party cookies set by yk4vn1sp.com. iOS WebKit users lose attribution both during the redirect (showing up in our 20.2% loss) and again on the landing page (postbacks fail to fire because cookies are blocked). The same effect applies to all iOS browsers because they all use WebKit by Apple's mandate.
Recommendations for Future Campaigns
Tiered priority. T1 = critical, T2 = important.
- [T1] Split TJ targeting by iOS vs Android. Run separate campaigns with separate budgets and separate KPIs. Expect iOS funnel to be 2-3x worse than Android by design. Don't average them into one number.
- [T1] Switch to Direct Linking for iOS-heavy creatives. Everflow supports it (Click Tracking Type, Direct Linking + EF.click() JS pixel). Bypasses third-party cookie issue. Requires landing page to host the EF JS snippet.
- [T2] Establish a 1st-party tracking domain. Long-term fix for ITP. Server-side cookie sync via a domain that matches the landing page domain (justsext.com or stxtai.com sub).
What's NOT yet in this dashboard
Honest scope note. These dimensions exist in source data and can be added on request. Listed in priority order for a paid-traffic specialist:
- Spot/placement breakdown (Tube8 vs Pornhub Mobile Footer vs Redtube etc). Critical for identifying where 20.2% click loss concentrates and where eCPC efficiency varies. Can be pulled per campaign from TJ
spots field.
- Daily trend (impressions/clicks/spend per day for each campaign). Useful for understanding pacing and creative fatigue. Requires per-day TJ stats query.
- Event type granularity. All 261 events under offer 19 are recorded as
(default) in Everflow's conversion_event_name. We cannot today separate Sign Up vs First Purchase vs Token Spend at the event level. This is an Everflow configuration limitation, not a data limit.
- Geo / device split (TJ side has it, Everflow side has it). Useful for ROAS analysis if certain countries/devices dominate.
- Full revenue picture. $55 reflects only what Everflow's offer 19 captured. Real revenue from same users may exist in Mixpanel under
subscriptionPaid with _ef_transaction_id matching one of these clicks. Cross-system reconciliation pending.
- Brand-side metrics (lift in branded search, organic signups, creator ecosystem effects). Requires GSC + Mixpanel queries and a baseline period to compare against.