Basically, archaeologists know that something is a tool because they know what a tool is supposed to look like. Archaeologists have done a lot of practicing with making stone tools of their own. Because they have done this, they know what the distinguishing characteristics of such things are. In particular, they know what it looks like when you hit a stone with another stone to shape the first stone. This leaves characteristic marks from the percussive blows.
By making their own stone tools, archaeologists have learned what marks are left by the intentional shaping of stone. That is how they can tell when a stone has been shaped on purpose rather than simply having been split by natural processes.
Archaeologists use three main criteria for distinguishing between stone tools made by humans and those created by natural processes.
They consider the Principles of Conchoidal Fracture:
1) In order to make a stone tool you must have the right material—a hard, homogeneous material—one that fractures in a uniform and specific way to produce a conchoidal shaped flake, e.g., obsidian, chalcedony, flint, etc.
2) You must hit the stone with the right amount of controlled force at an acute angle.
The principles of conchoidal fracture produce characteristic markers on stone tools, e.g., a striking platform, bulb of percussion, flake scares that can be used to compare a purposefully made stone tool from a naturally made rock.
Archaeologists also consider context. Is the stone an isolated find? Are there other artifacts in association with the stone tool? Is it found in a context where you would not expect raw material of that kind?
Degree of modification is also important in making a determination—has the stone been modified in some way, i.e., unifacially flaked, bifacially flaked, or retouched.