Just to sum up everything that has pretty much already been said.
First off, establish what hand and arm signals you are going to use and what they mean exactly. When spotting, the driver solely relies on the spotter and essentially trusts him or her to get him through a specific piece of a challenging terrain.
Second, the hand and arm signals need to be precise and not mimic sign language. Corrections, stops, etc. need to be clearly demonstrated to ensure the driver is doing what the spotter wants him to do. If for some reason the driver does his own thing, the spotter needs to stop him right then and there and provide necessary corrections. For casual offroading speed is not of the essence, instead safety and avoiding damage to the vehicle/terrain are paramount IMO.
Third, spotter needs to be visible to the driver at all times and be cognizant of potential danger areas, i.e. truck rolling over him, falling off the cliff himself, etc.
Also, if the driver is relying on the spotter, he is to maintain a given line by the spotter, unless there is another correction made. Too many times you see a driver starting to do his own thing, although no additional corrections have been issued by the spotter.
The biggest issue in this scenario that I see is the total lack of clear cut communication between the driver and spotter, lack of understanding of what hand and arm signals mean, etc.