The first ingredient is the ordering of the letters. Greek, Roman, Cyrillic, and Hebrew go something like ABC. . . I imagine this was a conscious decision. For example Ancient Roman teachers were aware of the Greek language and alphabet. Arabic seems to be its own thing. Farsi seems to be related. I imagine other languages in the same group use the same order. Futhark is different again.
If you make up your own alphabet you get to decide the order. It is up to your imagination.
On the first step we assign each letter to a number and then add up the numbers. If that number is too big we reduce it again. The most common rule reduces $$1349 \to 1+3+4+9 = 17 \to 1+7=8$$
and we stop when we get a single digit.
This reduction rule depends entirely on how we write the number. It only makes sense to people who write things in base 10. If we wrote in binary or octal or hexadecimal we'd get a different reduction rule, and a different range of final values rather than just 0,1,2,3,4,5,6,7,8,9.
To understand this the decimal reduction can be described as
Lower the number by 1000 and add a bean to a pile. Keep lowering the number by 1000, and adding another bean, until the number is less than 1000. Then repeat the process with 100, 10 and 1. Each time the number decreases add an extra bean. When the number is reduced to zero, count the beans. This is the reduction.
In a different system the numbers 1000,100,10,1 might be different. For example the Ancient Mayans used 360,20,5,1.
The dots are 1s and the bars are 5s. One example of a reduction works like this.
$$37 \to 32 \to 27 \to 22 \to 17 \to 12 \to 7 \to 2 \to 1 \to 0$$
We pull out 5s until we cannot any more. Then we pull out 1s. There are 9 steps.
Homework: Reduce 9 using the same procedure.