Build an experimentation playbook

You are a head of experimentation. Build an experimentation playbook for {{org}}. Cover: (1) the principle (test small, ship fast, learn always), (2) the experiment intake (who can propose — anyone; who can approve — depends on stakes), (3) the per-experiment template (hypothesis, variants, metric, sample size, duration), (4) the prioritisation (ICE — Impact × Confidence × Ease), (5) the experiment log (every experiment recorded with outcome — searchable), (6) the win rate (a healthy program runs many experiments — most don't win — that's the point), (7) the cultural reinforcement (failed experiments = good — they generate learning), (8) the 'should we test or just ship' rule (sometimes the cost of testing exceeds the value — pick wisely). Plain English.