There’s more than one way to make software user interface decisions. Depending on the model a product team is most comfortable using, expectations of design professionals and their output can vary. As a result, it may be useful for interface designers (be they visual, interaction, or information focused) to consider the prevalent model on their team and act accordingly to either shift or meet expectations.
Each of these models can make design decisions and thereby “move pixels” on the screen. Not all are explicitly design-driven which might ultimately frustrate people who most closely align with the "designing" model of making decisions.
Designing: decisions are evaluated by how well they contribute to an integrated “human-centric” experience. This is the model most designers crave because it leverages their ability to empathize with their target audience and think holistically. Designing focuses on understanding the fundamental purpose of an application and bringing it to life in a way people can understand and use. An experience that “makes sense” to your audience is your yardstick in this model. This approach is probably most common when developing new products.
Optimizing: decisions are made based on explicit testing of isolated variables to drive very specific behaviors. Designers create variations of a control that are evaluated systematically. The elements that perform best likely become part of the user interface. In this model, performance is your yardstick for decisions. Mature products (especially cash cows) frequently employ optimizing models and designers on these teams spend most of their time creating (lots of) iterations for the key elements of the product.
Managing: decisions are reached through discussion or debate. In this model, designers represent the collective decisions of groups within the product by laying out what everyone agreed to. Consensus and buy-in are the yardsticks by which people judge success.
It’s important to note that the processes that define each model can also be inputs to other models. For example, explicit testing can help inform a holistic design or healthy debate can result in variations to test in order to optimize a product. The thing that distinguishes each model is what is used to make design decisions most of the time: holistic customer experience, optimization, or consensus.
While all of these models may be in use at the same organization, perhaps even on the same product, I think it is useful for designers to be aware of the distinction so they know how to operate effectively. Teams comfortable with and focused on optimization may require a considerable amount of convincing to embrace a holistic product design as their decision-making criteria. Similarly, teams used to managing toward decisions may resist data-driven evaluations of their work.
Most organizations skew pretty strongly to one style of decision making or the other. However, as Tom Chi recently pointed out a healthy integration is where the magic happens.