Every so often a site has soil that cannot support what the owner wants to build on it. It is too soft, too compressible, too loose, or it would liquefy in an earthquake. When that happens you have three honest options: drive deep foundations down through the bad material to something competent, change the structure so it tolerates the soil, or improve the soil itself so an ordinary shallow foundation works. The third option is ground improvement, and it is often the most economical answer when people assume their only choice is expensive piles.
What it actually is
Ground improvement is engineering the soil before you build, so that a volume of marginal ground behaves like better ground. It changes the stiffness, strength, or drainage of the soil mass in a way you can analyze and design for, the same way you would design any other structural element.
The common methods, in plain terms
Stone columns and aggregate piers are the most common methods. You install closely spaced columns of compacted crushed stone into the soft soil, by vibro methods or by ramming, and the result is a stiffer, stronger composite mass. They increase bearing capacity, they cut settlement by carrying a share of the load on the much stiffer stone, and because stone drains freely they can also help with liquefaction by relieving pore pressure. Proprietary rammed aggregate pier systems are a refined version of the same idea.
Geosynthetic reinforcement uses layers of geogrid or geotextile inside engineered fill to spread load and add tensile strength the soil does not have on its own. It is what lets you build mechanically stabilized earth walls, bridge soft spots in a subgrade, and create reinforced soil zones beneath foundations and embankments.
Beyond those, deep soil mixing blends binder into the ground to create stronger columns or panels, and surcharge preloading simply puts the future settlement on the site early, with temporary fill, so the building does not have to absorb it later. Which one fits depends on whether your problem is bearing, settlement, time, or liquefaction, and on the cost of each system against the deep-foundation alternative.
How it gets designed
The key idea is that you are designing a composite: a volume of native soil reinforced by stiffer elements, with a combined behavior you can compute. For straightforward cases there are established analytical methods and a good deal of Federal Highway Administration guidance, and for transportation work it connects to AASHTO. For more involved geometry or staged loading, the improved ground gets modeled directly in a finite element analysis so you can see the settlements and stresses rather than estimate them.
Why this sits next to my buried-structure work
A lot of ground improvement appears on exactly the kinds of sites where I am already working below grade. If you have read about my buried-structure and CANDE work, this is the adjacent problem: the behavior of a buried structure depends on the soil around it, and sometimes the answer to a soft or settlement-prone site is to improve that soil instead of oversizing everything bearing on it.
If a soils report has come back worse than you hoped and someone has quoted you an extensive pile foundation, it is worth a second opinion. Improving the ground is frequently cheaper, and figuring out whether it applies to your site is a short conversation. Ground improvement design is part of my geotechnical services.