Creating Playbooks Creating Playbooks - overview Present Creating Playbooks Codifying practice so it transfers, improves, and retires Most playbooks die one of two deaths The museum piece: written once and never revised against evidence. Stale plays carry false authority. The tell: every play has the same date on it. The enforcement tool: made mandatory, with the when-not-to deleted. The tell: teams comply in public and ignore it in private. Both deaths have one cause. Nobody resolved the tension between standardising and improving. The standard is the baseline that makes improvement visible Codification standardises. Improvement requires variation. The resolution is Toyota's. The standard is not the ceiling. You cannot see a local improvement without a standard to see it against. This holds only if every play is provisional and a local improvement can become the new standard cheaply. A play is a situated action Entry conditions The situation the reader is standing in The action What to do. Usually the smallest part Signals How you would know it is working, within weeks When not to Where the play does not apply. Mandatory When you see these conditions, this move tends to work, and here is how you would know. Most things called playbooks are something else Does it break into plays? Does each play say when it applies, when it does not, and how you would know it is working? Does it still need judgement? Any no, and it is a neighbour: Checklist: items to verify, with no situation Runbook: deterministic steps, with judgement minimised Methodology: adopted wholesale, with the when-not-to deleted Values statement: aspirations without moves or signals Something else is fine. Mislabelling it is not. Every stage pairs its plays with an antipattern Sense Wait until a move recurs a third time Extract Write down what actually happened Codify Put entry conditions first and name the forces Publish Publish before you are confident Govern Make improvements standard and retire in the open The antipatterns: premature codification, the espoused playbook, the context-free recipe, the broadcast, and the museum piece. Extract the play from what actually happened People describe their espoused theory. What they actually do is their theory in use. The gap is widest where practice is most skilled, because skilled practice has gone tacit. Write the instances out raw first, then extract the play from them. Keep the failure in. If it took three attempts, that becomes a force in the play. The tell: a real play contains at least one detail you would rather leave out. A local improvement becomes the new standard Vary A team varies a play locally Prove The variation survives three uses and beats the standard on its signals Decide The owner adopts, adapts, or declines with reasons Shift Other teams' behaviour changes too The change path must cost less than the workaround. Otherwise the playbook forks into an official version and a real one. Nobody should need wizardry to read or change a play Default to public. Make exceptions play by play. Reading needs only a browser. Keep the source under version control and the reading path at a stable URL. Proposing needs no tooling. A form, an email, or a comment counts. The tell that this has failed: the people closest to the work are not the people proposing changes. Three plays that transfer beat fifteen that do not Start a candidate list today, with one line per repeated move. Write one play from a real instance, using the template. Put the entry conditions first. Publish before you are confident. The first three uses are the test. Read the full playbook profound.fyi/playbooks/creating-playbooks These slides are a projection. The plays live, and change, at the source. Licensed under CC BY-SA 4.0.