How can I optimize token distribution to a diverse group efficiently?