Often, one of the earliest design decisions is what configuration or style is appropriate. Sometimes, the design – whether by vision or constraints – is very specific. The rest of the time, multiple options will be possible and must be assessed by performance, cost, etc.
For instance, a skylight over a rectangular space could be configured as a gable, a hip, a vault, a ridge, or a single slope (if narrow enough). A skylight over a square space is likely a standard pyramid, but could be a many-sided polygonal pyramid or even a segmented dome. Complexity increases cost. A single slope is simplest and cheapest, a ridge skylight adds another pitched half, a gable adds end walls, a true or segmented vault requires either bent materials or many additional joints, and a dome compounds that with another radius, each getting increasingly complex and expensive.
On a more granular level, 4-sided orthogonal profiles are cheaper than multi-sided and/or curved ones, 90° angles are cheaper than others, planar profiles are cheaper than curved/segmented ones, etc.
Larger is obviously more expensive than smaller. Less obviously, it’s not a linear relationship – though a 4’-0” square pyramid requires less material than an 8’-0” square pyramid, the manufacturing process and time is the same.
Speaking of size, an attached greenhouse or solarium eliminates one or more glazed walls and will be less expensive than a freestanding one.
Choice of glazing greatly impacts both cost and performance. Plastic dome/pyramid unit skylights are the cheapest option for small openings (up to 8’x8’), but have the lowest performance and product life. Glass is available with many options affecting performance, appearance, and cost. It has the highest upfront cost, but has a very long lifespan. Translucent, structured polycarbonate panel glazing can be used in lieu of glass – it tends to have better insulating properties than glass and costs less upfront, but has a shorter lifespan. (See “Glazing” below for more details).
Beyond what you design, how you design it can have a big effect on final cost. It’s important to have clear project documents, require an appropriate level of quality, and to invite competitive bidding. If drawings/specs conflict or are vague, manufacturers/installers have to put a cost on the uncertainties, usually “worst-case”. You want enough specificity/detailing/testing to ensure quality, but excessively-picky/demanding standards, testing, etc. can inflate costs with little or no additional benefit. This ties into competitive bidding, as many proprietary product specifications will impose “features”, extreme standards, or unnecessary requirements specifically to exclude competing products. If you start with a proprietary spec, spending a little time to identify and eliminate this and “open the spec up” will likely lead to savings from the competition inherent in having multiple options.