I'd guess that most people start with some idea of what colors/size/season of bloom, etc they desire to see in the new bed. You'll also of course want to factor in suitability to your exposure, climate, soil type, degree of maintenance desired, height relationships, etc. Drawing a schematic of circles to fill an area before thinking of actual plant species doesn't necessarily reflect these sorts of factors. I myself like to start the process thinking of what sorts of plants I'd like to use that will like the situation, and will also tie in with or contrast the rest of the garden. Regardless of how you approach the layout and choices; it will always work better if you learn as much about the plant's needs and ultimate size as possible, before you make the purchase. Getting inspiration from actual gardens with live plants in your locality is also a useful shortcut to getting up to speed.

Is this a decorative freestanding bed plopped out in the middle of a lawn or is it a bed adjacent to a building, other structure or group of trees or something else? Is it a bed that's to be mainly an attractive collection of plants? Or is it a bed in which plants will exist, not because of their individual worth, but because of how they can fit together to create something greater than themselves... like extending the form of a building...or making an outdoor room or something like that?

My process involves looking at the site as a whole composition rather than as a single bed.
I think of and layer in the contraints and the opportunities of the site such as climate, drainage, topography, soil conditions, sun, wildlife .... basic site analysis .
Often the site analysis will inform the shape of the beds, other times it is pure design.
Once I have the shape of the landscape beds and hardscape laid out I sketch and visualize in 3 D ( by hand) . I rough out basic foliage shapes , texture and colors , all the while thinking in the 4th dimension of time (seasonal changes).
Eventually it all gets laid out to a tight scale so that I can calculate / design the irrigation system and provide specifications to the installation contractor.

Calling Wellspring... Here's your discussion without photos, each post creating a vision in the mind of the reader. This is an excellent example of how well this forum works, intelligent responses to a deceptively simple question.

Thanks for the responses. I am in a design class, so that is where I am coming from with this question.

@Yardvark, it is all of those things. I was just curious what is the most logical way to go about it, but I guess there is no 'right' way to do it. Everyone seems to have their own method to the madness.

"I was reading a thread here the other day that suggested choosing plants first then going about making the bedlines to fit that plant" "I am in a design class, so that is where I am coming from"

I don't think of specific types of plants when I'm creating bed lines. I let the site analysis and features tell me the basic shapes of beds and why there should be beds. I would first be asking myself what 3-d shapes are necessary to accomplish certain goals. For example, does a building have an ugly foundation and need the equivalent of a baseboard? Is there a blank space on a building wall with no interesting feature (like you might hang a picture on if it was on the inside?) Is there a group of trees that would be a pain to mow around individually and would be better to "collect" in an island? Is there an ugly view to be screened? Are there places that I want people to see and go, that shouldn't get any plants, but instead have plants outside of those areas in order to direct view and/or circulation? Next, I would be thinking in terms of the basic plant shapes: cushion, clipped or unclipped hedge, bosque, cone, dome, mat, wall, ceiling, floor... the very basic shapes and purposes and ask myself how can I solve the problems I've identified in the analysis using these shapes? Once I begin to apply the shapes (and I would be visualizing them at a size that would actually solve the problem, then I can begin to see how they'll fit together. As this forms up, I can begin to see, in plan view, how things begin to relate to one another and to the site conditions and features... especially buildings and hardscapes. At this point, there are usually rough edges. In the plan view shapes can geometrically look clunky next to other shapes which are solving different problems. It's necessary to experiment with the 2-d geometry until you can fit things together in a way that will make them hum like a piece of art. But usually, I'm only tweaking (sometimes heavily!) what the site has already dictated. The bed line serves the purpose of linking together all the various "objects" that are in place solving their various individual problems. Over time, one develops their rules of do's and don'ts and fitting those into the equation becomes automatic. For example, to me, bed lines that meet other objects (walls and paving) don't look good at odd angles. I tie them in at 90*. After I'm content with the plan geometry of the bed line, I begin to figure out what specific plants will be happy in the conditions the site is offering and create the shape I've deemed best to solve a particular problem.

Hope that helps.