This is the number of messages sent / received. Any serious SMS API should provide a handset delivery confirmation. The ratio itself is important, but what are more meaningful are its variations. Some carriers don’t provide delivery confirmation, or provide them sporadically. Therefore, it is more important to track variations for a given network rather than an absolute figure.
Another ratio that could be viewed as meaningful is the number of messages sent that receive a delivery status (be it delivered or failed). This ratio measures how well the route is performing. A healthy route provides statuses for over 97% of messages.
With handset delivery confirmation comes information about delivery latency. Again, the important thing to measure here is the variation in latency, rather than the absolute figure. The reality is that not all carriers have the same quality of infrastructure, and some prioritize outbound traffic on a delivery confirmation path. Plus, some handsets take more time to acknowledge the reception of the message than others. And finally median is a better aggregate measure than mean because one or two phones being out of reach will have an disproportionate effect on the mean. Track changes, not absolutes!
The ultimate measure of performance is your application-specific metrics. For instance, if you are sending passwords to verify phone numbers, you should track the ratio of conversion between SMS sent and users registered. For marketing campaigns, you could measure returns. If your application is two-way, then track replies, etc.