Monthly active users is a good proxy for census. Tencent’s been doing something like that quarterly for years, the number of people who signed in to QQ in the last n weeks of the quarter (I think six, if I recall).
So this tells you the number of nodes that might light up.
You also want to know how much those nodes are available for conversation. For instance, early Skype didn’t have any way to ring you if you had your client turned off. So they focused hard on building dialtone: using push messaging on mobile to get you to answer a call, preinstalling Skype on desktops with the default to start at startup, etc.
The capacity of your OTT network is the census * dialtone; minutes of availability. You aspire to having every user connected 24 hours out of every 24 hours . If only for 6 hours a day, then your network has 25% dialtone. Each network has a different dialtone tipping point where users feel synchronous communication works well, at least for their part of the network and for strangers they are likely to reach..
It’s not enough that each user is connected to the network. They really need to be connected to their own social set, the people they want to talk with, the communities to which they belong. You see “add your contacts?” as part of customer onboarding for a reason. So you could compare networks for how many contacts on those networks do people know. For example, in Skype’s early days, they were doing well if users had a dozen contacts in their directory.
Now that people often have hundreds of contacts dragged in from mobile directories and facebook friends, perhaps it’s time to start measuring relationship strength; how well does this network help you repeatedly engage your friends, coworkers, customers, teammates over time?
But the whole point of a communications network is talk. How much are people talking? In landline days you might have counted total minutes served per period or minutes per subscriber. Today’s users substitute and switch modes, freely jumping from SMS to in-app texting, to voice messages, to live calls and back. We need measures of engagement, of how much conversation is happening across the network per person.
This all becomes interesting in a different way as WebRTC enables talk embedding in every medium, from blogs to banner ads, from MMOs to robotics. Human patterns of interaction won’t be bound to a handful of branded clients and devices. We need new social capacity metrics for an era where networks connect people through the Internet of Things, wearables, and dancing robots.