Our self-improving, plan-prediction app can forecast how different categories of people will rate the desirability levels of alternative plans in any situation. We used it to help plan the rural-urban fringe of Melbourne, Australia, to eventually conclude, with 95% confidence, that everybody would regard a “do nothing” plan as significantly worse than three suggested action plans - regulated suburban sprawl, a green belt and horticulture. But although it came close, our app did not statistically split these three action plans in terms of their predicted desirability. Nevertheless, one of the app’s spin off benefits was its myriad non-significant suggestions as to why certain types of people might, and might not prefer certain plans. The latter prompted the authors to consider a plethora of potential plan improvements which they probably would not have otherwise thought of. Hence right now our app outputs a tentative, but still mind-expanding and novel type of planning support - plan predictions. It will become more statistically precise, and hence even more useful, after it has “learned” from more users in the future.