The most succint process flow is this:

More UTXOs -> more strings of data to be included to block -> more bytes in your TX data -> higher number being multiplied by more the (sats/byte) * (number of bytes in TX data) fee calculation