Whenever we undertake the (re)design or (re)development of a process, product, system, layout, tool, visual control, etc. (you get the point), it's usually a good idea to define the criteria for the future state FIRST. The definition must extend to measurable performance or outputs - like cost, quality, lead time, etc. Similarly, the criteria should also extend to the characteristics, the "whats" and "hows," of the future state. For example, we may determine that the new layout has to, among other things, facilitate visual management and natural work team co-location.
Without this clarity around performance and characteristics, it is difficult to understand "what good looks like." We need to start with the end in mind. It's part of the P, within PDCA. Absent clarity, an individual, team or organization is at risk of ginning up some options and then justifying later why one or more is good. This approach is not acceptable...unless of course it's around something very trivial, like ordering lunch.
Furthermore, especially in a team environment, if the criteria are not articulated in a public and visual way (flipcharts, Post-It notes, whiteboards, etc.), there is no way for the team to discuss, test, debate and reach consensus on those criteria and ultimately share and own the vision. Just think if we asked a team to go ahead and design a dream house without a shared vision. Without any definition, one person would be envisioning a mountain top retreat, another a beach side mansion, another a richly appointed brownstone in the city...
So, how do we go about defining what good looks like? If we're talking new products, there are a host of lean design tools that can be applied individually or systematically, including: quality function deployment, must/should/could prioritization and the "seven-alternatives" process (a technique of 3P). Ron Mascitelli's work, The Lean Design Guidebook is an outstanding reference in this area.
Keep in mind that the level of effort we invest in the process of articulating design criteria should match the importance of the task at hand, related risk and how pragmatically we can take something subjective and make it quantitative. So, for example, if we are doing a quick seven different ways application for the design of a pacemaker scheduling system in a mixed model environment with demand coming from both a supermarket (make-to-stock replenishment) and make-to-order kanban, the criteria may include: visually controlled, kanban cards as visual artifacts, maintained by group leader, reflect status of required changeovers, etc. This criteria will probably be sufficient for a team to pursue the seven different ways, make trade-offs, down-select to three or so for trystorming and eventually and quickly converge on one best way (for now).
We can get fancier with Pugh Methods, weighted averages and the like. The important thing is to match the intensity to the challenge and to never violate the principle of first articulating the criteria. If we don't follow that principle, we're doomed to unthinkingly creating something and then putting lipstick on it later. Heck, then we could move to D.C.!
Related post: Model Lines – Federal Government Take Note