10 ug acceleration noise has large consequences in displacement error.
To get an estimate of the size of the effect, recall that for uniform acceleration, x(t) = 1/2 * a * t^2 . So, for 10 ug = 9.8e-5 m/s^2, this works out to ~600 m in an hour or 360 km in a day.
The real effect isn't quite this bad (it's random, not continuous, error), but it illustrates the point. Without an absolute reference for occasional comparison, inertial measurement requires exquisite precision and low-frequency stability in order to be useful for most positioning applications.