I nearly quit the sponsor talk

What happened

The residual reward

Raw views mostly measure how big the channel already was. The baseline predicts what this channel “should” have earned given the confounds above; the residual is how the post actually landed relative to that prediction — measured in standard deviations.

Checkpoints

What changed in the posterior

• Post beat its baseline by 0.33σ. Moved: topic_cluster:topic_3 -0.049, format:interview -0.035, hook_type:story_cold_open -0.034.

  • topic_30.357 → 0.308 Δ-0.049±0.28
  • interview0.144 → 0.109 Δ-0.035±0.27
  • story_cold_open0.203 → 0.169 Δ-0.034±0.27
  • weekend_pm0.417 → 0.388 Δ-0.029±0.29
  • 1_4m0.005 → -0.023 Δ-0.028±0.26
  • topic_1-0.058 → -0.039 Δ+0.019±0.27
  • text_dominant0.116 → 0.097 Δ-0.019±0.26
  • topic_6-0.532 → -0.516 Δ+0.016±0.33
  • vlog-0.171 → -0.157 Δ+0.014±0.27
  • 10_20m0.013 → 0.024 Δ+0.011±0.26
  • shorts0.153 → 0.164 Δ+0.011±0.28
  • weekday_am-0.127 → -0.116 Δ+0.011±0.27
  • weekend_am0.104 → 0.115 Δ+0.010±0.30
  • tutorial-0.025 → -0.015 Δ+0.010±0.26
  • number_list-0.162 → -0.152 Δ+0.010±0.27
  • 4_10m-0.242 → -0.234 Δ+0.009±0.28
  • topic_0-0.083 → -0.075 Δ+0.008±0.27
  • face_reaction0.048 → 0.056 Δ+0.008±0.26
  • question0.101 → 0.108 Δ+0.007±0.28
  • under_60s0.119 → 0.124 Δ+0.005±0.28
  • commentary-0.167 → -0.172 Δ-0.005±0.28
  • demo_first-0.029 → -0.025 Δ+0.005±0.27
  • contrarian-0.037 → -0.032 Δ+0.005±0.28
  • topic_40.167 → 0.162 Δ-0.005±0.27
  • claim-0.060 → -0.056 Δ+0.004±0.29
  • object_hero-0.315 → -0.310 Δ+0.004±0.28
  • topic_2-0.176 → -0.172 Δ+0.004±0.28
  • weekday_late-0.304 → -0.300 Δ+0.003±0.29
  • topic_50.104 → 0.108 Δ+0.003±0.30
  • list0.082 → 0.084 Δ+0.002±0.30
  • none0.103 → 0.105 Δ+0.002±0.28
  • before_after0.064 → 0.065 Δ+0.001±0.30
  • weekday_pm-0.075 → -0.074 Δ+0.001±0.25
  • 20m_plus0.121 → 0.122 Δ+0.001±0.29