The actions, owners and dates for moving named stakeholders from their current engagement level to the level the work requires.
An engagement plan exists to close gaps, so it is only as good as the assessment that produced them: a stakeholder already at the level the work needs belongs in no plan at all. The useful form names an owner and a date per gap rather than listing communication channels, because a channel is not an action and nobody is accountable for a newsletter. Plans decay quickly, since the positions they were built against drift with reorganisations, budget rounds and changes of leadership, which is why the assessment behind the plan is re-run at stage gates rather than filed at initiation.
See also
Where a stakeholder sits on the scale from unaware to leading, judged on observed behaviour rather than on stated position.
Identifying who holds a claim on a piece of work and classifying those claims, so that attention is allocated deliberately rather than by whoever shouts loudest.
Where this comes up
One row per stakeholder, one column per engagement level, and two marks per row. The distance between the two marks is the only thing on the grid you can actually act on.
Three attributes, seven classes, one question: whose claim gets attention first. Hold all three and you are a definitive stakeholder, and definitive stakeholders do not wait.
A reward is set in advance and conditional. Recognition is retrospective, unexpected and free. Most project managers control the second and none of the first, which decides where the effort goes.