n=18.2K
Day-7 CR review of a 'winning' push variant
Panel CTR beat control by 23%. Day-7 CR underperformed by 0.04pp. p=0.04, Q3 2024.
A selection of campaigns and monetization stories where I (or a client) had the panel access to publish the full diagnostic. Every case below carries vertical, GEO, time window, and the number that matters.

Each case below reports vertical, GEO cohort, time window, and the number that decides the outcome. Cases without sample size or significance threshold don't get published — that's the test I apply to my own work too.
n=18.2K
Panel CTR beat control by 23%. Day-7 CR underperformed by 0.04pp. p=0.04, Q3 2024.
n=12.4M
Going 5/day → 3/day improved CR by 0.04pp. 3/day → 1/day improved CR by 0.02pp at 4x the impression cost. Q1 2024.
4.8% CR
14 days of 4.8% CR, day-to-day variance under 1%. Real human behaviour has variance. Suspended on day 15.
n=4.2M
Chrome desktop converted at 1.4–1.8x Edge desktop, holding GEO and vertical constant. Q3 2024.
200 conv.
Smart CPA pre-200 conv: CPA volatility ±48%. Post-200 conv: ±12%. Matches Google's published threshold.
n=8.4M
In-page CTR ran 1.2–2.8x classic. Day-7 CR ran 0.6–0.9x classic. Full-funnel within plus/minus 10%.
$0.18–0.65
BR/MX/CO/AR push iGaming. Tier-1 ran $1.20–3.40. Q4 lift added 28–42%. 2024 data.
17%
24-hour attribution window captured roughly 60% of what a 30-day window would. Why day-7 minimum matters.
$200K
Week-three audience-fatigue collapse. n=12.4M, p less than 0.01. The meeting that ended my Mobidea tenure.
Start here
Advertisers launch the first campaign in under 10 minutes. Publishers see the first eCPM data the same day. The day-7 cohort takes — predictably — seven days.