A rule of thumb that can tell us something about the spread of the data by pointing out the relationship between the % of data and std as follows:

The % of the data that lies within "K" standard deviations is atleast: 1 - 1/k^2 of K> 1.

Using the definition discussed in Mathtutor videos:

The % of the data that lies within "K" standard deviations is atleast: 1 - 1/k^2 of K> 1.

K is the number of standard deviations. So if K = 2, then the % of data that lies within 2 standard deviations is atleast:
1- 1/2^2.
So that is 1 - 1/4 = 3/4 which is 75%

If k = 3, then the % of the data that lies within 3 standard deviations is:
1 - 1/3^2 = 1 - 1/9 = 8/9 = 88.9%

1) Not bell shaped data
In the same way that the empirical rule can be viewed as a rule of thumb for data that is bell shaped, chebyshev's theorem is similar but for data that is not bell-shaped.

2) Where the empirical rule can tell us that a % of the data falls within x stdeviations of the mean, Chebyshev can only give us a minimum amount, like  "atleast x percent is gonna fall between one standard deviation".