Just a simple example to clarify this:
suppose you have a table with two columns: country names and city names.
suppose that you define two indexes, one per column.
finally suppose that your table data is physically sorted on the city names, and then (for equal cities) on the country name.
The cluster ratio of the "city names" index will then be 100%, while it would be very unlikely that the cluster ratio of the other index is higher than 10% (unless there are very few different cities or countries): country names will appear in seemingly random order.
The other way around, and assuming several cities per country, the cluster ratio of the "cities" index *could* be around 50% or even higher, e.g. when rows would be physically sorted on city for equal countries. But you'll never get both indexes have cluster ratio 100%, unless there's just a single country, or in the very unlikely event when the first few characters of city names equal their country name.