retti.aiTeardowns › I Tested 100 Years of World Cup Footballs
Predicted Retention Teardown

I Tested 100 Years of World Cup Footballs

By SIM2 · Sports · 17.8K views · 25:03

I Tested 100 Years of World Cup Footballs

The teardown in brief

What's working

What's costing attention

The first 30 seconds

Every time I score a goal, the World Cup ball gets older. Going all the way back to the first World Cup ball ever made. First up, the brand new 2026 Triion, where our challenge is to score a top bins goal. What? Like this? Surely not. Come on. What the flip have I just seen? Swiftly moving on to 2022, we have the Al Ri

Hook fires immediately — concept stated in the first sentence, rapid ball montage matches the title promise exactly, and the high energy delivery is correct for the format. YouTube's own data confirms 80% at 0:30 which is above typical, validating this as a Tier 1 hook. The only gap is that stakes (prize + forfeit) are absent from this window, meaning the viewer knows WHAT they're watching but not WHY it matters.

Where viewers drop

0:00 — Steep Commitment-Window Drop (critical)

In the first three minutes, you're juggling three different balls (2026 Triion, 2022 Al Rihla, 2018 Telstar), announcing the stakes, and simultaneously showing attempts. Viewers who clicked for a clean 'oldest to newest ball challenge' have no mental anchor yet. The stakes (Lego prize vs old boots) don't land until 1:27, which means the first 87 seconds of actual competition have no consequence. Half your audience is gone before you've told them what they're rooting for.

Why it matters — The real graph shows you drop from 100% to roughly 54% in the first three minutes. That's 46 points of audience lost before ball four even begins. You confirmed 80% at 0:30 which is great, but the subsequent freefall between 0:30 and 3:00 is the single biggest damage event in this video.

4:07 — Mid-Challenge Sponsor Break (moderate)

The Surfshark sponsor fires at 4:06 — right in the middle of the 2014 Bazooka challenge. Seb has just scored a great goal, Sim still has attempts remaining, and you cut away for a 90-second product read. You return with 'Seb getting closer, I needed to step up my game' — which means the sponsor interrupted active game tension. The audio data confirms the delivery around 4:48-5:06 stays at VERY_LOUD (-11.2dB), which is unusual for a sponsor read and suggests the gameplay continued during it, but the transcript shows a clean stop-and-sell structure.

Why it matters — The graph shows the 40% mark landing right around the 5-6 minute window. While this is partly baseline decay, placing the sponsor mid-challenge gives curious viewers an off-ramp. They can leave now without missing a goal because the content has already paused for them.

10:13 — Structural Repetition Fatigue in the Middle Third (moderate)

Between the 2002 Fever Nova (10:12) and the 1978 Tango (15:44), you run through five consecutive balls in roughly 28 minutes of real video time (10:12-18:02). Each follows an identical pattern: ball introduction (30 seconds of history), challenge stated, five attempts each, someone wins or no point awarded, score updated. There's no variation in this structure across five consecutive rounds. The viewer's brain locks into a predict-and-confirm loop — 'intro, attempts, result, next' — and the reason to stay becomes weaker with each correctly-predicted cycle.

Why it matters — The graph holds remarkably steady from minute 6 onwards (approximately 1% per minute decay), which is actually impressive for a 25-minute video. But the repetition index is the reason you're not retaining MORE people through the middle — viewers aren't leaving in spikes, they're trickling away at a near-constant rate because the format is predictable.

24:28 — Forfeit Payoff Rushes Through in Under 45 Seconds (mild)

The 24-hour boot forfeit — one of the two stakes you set up 23 minutes ago — is delivered in approximately 35 seconds of screen time at 24:37-25:03. Starting the boots, wearing them to sleep, waking up, making food, doing editing, playing football — all of it gets one sentence. This is the consequence that was supposed to hurt. It barely registers.

Why it matters — Stakes only work if the payoff feels real. A viewer who watched 25 minutes partly to see someone suffer in horrible boots gets a 35-second montage summary. The graph drops steeply in the final minute (as expected for all outros), but a proper forfeit payoff could have converted some of those fence-sitters into genuine subscribers who felt the whole video was worth it.

How the video is built

What any creator can steal

More teardowns from SIM2

Want this on your own video?

Paste any YouTube URL and Retti maps every drop, spike and plateau to the moment that caused it.

Analyse a video free