Evaluation of academic research plays a significant role in government efforts to steer public universities. The scope of such evaluation is now being extended to include the "relevance" or "impact" of academic research outside the academy. We address how evaluation of non-academic research impact can promote more such impact without undermining academic freedom and research excellence. Five questions on evaluation design are considered: (1) What should be the object of measurement? (2) What should be the timeframe? (3) How should non-academic users of research inform evaluation processes? (4) How should controversial impacts be managed? (5) When in funding cycles should impact evaluation occur? We conclude that non-academic impact should be selectively promoted and evaluated. This is how greater gains from research might be best captured without imposing misguided and onerous reporting requirements on individuals and institutions.