The voiced aspirated palatal is close to absent from the inherited vocabulary of Old Indo-Aryan. The few Sanskrit words that carry it are mostly imitative or of uncertain origin, and a classical text can run for a very long stretch without any occasion to write the letter, which puts it among the least attested signs in the whole consonant section of the order.
Its frequency rose sharply in the modern languages. Middle and New Indo-Aryan generated the sound in quantity out of older clusters and out of borrowed vocabulary, so the letter that a Sanskrit text barely needs is entirely ordinary in Hindi, Bengali, Marathi and their neighbours, where it opens common words for a shower of rain, a quarrel and a flag among many others. A sign preserved for two thousand years on the strength of a systematic table turned out to be exactly what the descendants of the language would need, which is not an argument for building tables that way but is a pleasing outcome. It is also a caution against reading letter frequency in one period as a measure of a letter's worth, since the language a script serves does not stay still.
There is a lesson in that for how a script should be judged. An inventory is not a description of what a language needs now but a claim about what it might need, and a table built on features rather than on a word count survives changes in the language that a leaner inventory would not. The Indian phoneticians did not foresee the modern languages, and they did not need to: a system that generates every combination of its features will happen to contain whatever the language later develops, provided the features themselves stay stable.