As pointed out by Ross Eckler in the Nov 2008 Word Ways ("Mnemonics for Number Sequences", p. 297), the well-known type of mnemonic which uses the length of successive words to represent a sequence of decimal digits (with a ten-letter word for each occurrence of zero) is not particularly efficient. If the sequence of digits of π and e, for example) then in the long run this scheme will have an "inefficiency ratio", defined as (total number of letters used)/(number of digits represented), of 5.5. This is pretty far away from 1, which could be considered ideal in some sense.
"Toward More Efficient Number Mnemonics,"
Word Ways: Vol. 42
, Article 22.
Available at: https://digitalcommons.butler.edu/wordways/vol42/iss3/22