GTM-1807: TJ × Everflow Tracking Audit
90-day end-to-end audit of paid (TrafficJunky) and affiliate (Everflow) tracking. Period: Feb 6 to May 7, 2026.
Scope: TrafficJunky and Everflow only. Aggregated data from TJ + Everflow APIs, subject to ongoing accuracy verification. Test affiliate (id=2) is shown separately from production wherever it appears.
Headline KPIs (90 days)
TJ Spend
$24,208
across 65 campaigns
TJ Impressions
317M
eCPM ~$0.07
TJ Clicks Served
111,480
CTR 0.035%
EF Gross Clicks
105,861
96.8% of TJ eligible
EF Valid Clicks
95,989
9,872 invalid filtered
EF Events Fired
1,411
attribution chain working
Production Revenue
$931
83 real conversions
Test Revenue (excl)
$1,269
128 internal test conv
Funnel: Impression → Conversion (90d)
TJ Impressions
100%
317,082,486
baseline
TJ Clicks Served
0.035%
111,480
CTR 0.035%
EF Gross Clicks
96.8%
105,861
3.2% lost (browser/network)
EF Valid Clicks
90.7%
95,989
9,872 antifraud rejected
Events Fired
1.47%
1,411
click→event
Real Conversions (prod)
5.88%
83
event→conv
Production Revenue
$11.22 / conv
$931
excl. test affid
Bar widths are illustrative (clamped for visibility). See absolute numbers and % column for precise values.
Per-Campaign Tracking Status (65 campaigns, 90d)
| Status | Campaigns | 90d Spend | 90d Clicks |
|---|
| OK tracking + uid | 52 | $16,386 | 70,506 |
| NO_UID no campaign attribution | 5 | $6,986 | 37,053 |
| DIRECT direct URL, verified working as intended | 1 | $331 | 1,560 |
| EXTERNAL third-party tracker, by design | 1 | $459 | 2,150 |
| NO_TRAFFIC never ran | 6 | $0 | 0 |
Why are EF clicks fewer than TJ clicks?
Decomposition of the 15,491 click gap (TJ 111,480, EF valid 95,989):
| Source | Approx Clicks | % of Gap | Status |
|---|
| Antifraud rejection (Everflow filter). Bots/proxies from Turkey, Indonesia, Switzerland etc, correctly rejected. | 9,872 | 63.7% | Working as intended |
| Browser/network drop (structural). Safari ITP, slow mobile network, ad blockers, browser privacy extensions. Industry-normal for adult vertical. | ~3,469 | 22.4% | Industry-normal |
External tracker (CrakRevenue campaign). Routes through t.vlmai-1.com, never reaches our Everflow. | 2,150 | 13.9% | By design (3rd party) |
Bigger problem hidden by the totals: 38,429 EF clicks (40% of valid traffic) land on Default URL #0 with no link to specific TJ campaign. This is not "lost" but is "unattributable", we cannot tell which creative test drove which clicks for the 5 NO_UID campaigns.
What's working / What's broken
✅ Working well
- Offer 19 (TJ Tracking) URL hygiene: 102 of 102 active URLs use full template with
{affiliate_id} placeholder
- Transaction ID integrity: 0 empty IDs across 4,593 events, 3,677 unique transactions identified
- Antifraud filter correctly blocks 9,872 invalid clicks (mostly VPN/proxy from non-US/CA geos)
- 97.3% of events approved (only 126 rejected from 4,593)
- 52 of 65 TJ campaigns are properly configured with tracking domain + uid
- TJ to EF gross loss is only 3.2%, well within industry norm
🔴 Broken / Needs fix
- 5 campaigns route to EF Default URL without uid: 38,429 clicks ($6,986 spend) cannot be attributed to specific campaign. Affects creative testing (G-SocNewCreatsDeskTest1, GoSocial Feb Creative Test variants)
Top 3 Recommendations
1Pre-launch URL hygiene check on every TJ campaign
Before activating any TJ campaign, validate that ad target_urls include both the tracking domain (yk4vn1sp.com) AND a uid= parameter pointing to a specific Everflow Offer URL. Skipping this step is what caused 5 campaigns ($7K spend, 37K clicks) to lose per-campaign attribution. A one-line check before "Save & Run".
2Backfill missing uids on 5 in-flight campaigns or accept they cannot be measured per-creative
For G-SocNewCreatsDeskTest1, GoSocial Feb Creative Test (1-20) default, GoSocial Feb Creative Test default lander, G-SocNewCreatsMobile300x100Test, Initial Static Banner Test_New: either pause and rebuild ad URLs with uid, or document them as "aggregate-only" (no per-creative comparison possible). Without action, future creative tests with these will keep producing un-segmentable data.
3Standardize a campaign launch template: 10+ creative variants, each with its own uid and dedicated Offer URL
Best-performing campaigns over 90d had 14 creative variants (Sophie/Lena PV) and routed each one through its own Offer URL via unique uid= parameter. Under-creatived campaigns (4 in audit had <5 creatives) suffered creative fatigue within days and produced thin data. The combination of creative count (rotation health) plus per-creative uid (measurability) determines whether a "creative test" yields actionable insight or just spend. New rule for launches: minimum 10 creatives, 1 uid per creative, EF Offer URL pre-created and validated before campaign goes active.