There are a number of business cases for telepresence but they all reduce to supplying an alternative to physical presence. A good telepresence strategy therefore has to start by asking just what "physical meetings" are targets for a telepresence alternative, and why. Surprised that I'm not suggesting that voice relationships be considered? There's a reason.
Back in the old days, all this presence and collaboration stuff was called "computer-supported cooperative work" or CSCW. Early studies focused, not surprisingly, on whether something like video links really helped workers, and to find out they focused on the relative effectiveness of decisions made cooperatively, with vs. without video support. Interestingly, these studies showed that where the decisions were supported by two or three people, there was no measurable benefit to a video link. People in pairwise or trio relationships did fine based on audio cues and communications. Thus, you probably won't get a lot of benefit from telepresence connections between pairs of people or very small groups.
The second thing that research has shown, in this case my own included, is that where telepresence involves more than about four people, there is an issue with managing how their images are displayed, unless they are somehow collected into small groups (no more than three "windows" on a screen). Everyone has their own meeting strategy. Do you watch the speaker, your boss, the decision-maker, the opposition? Very few systems that support per-person videoconference connections can manage the meeting strategies of the users optimally. Given this, most telepresence applications should be thought of as being between meeting rooms and not between people.
Thinking of telepresence as a pure meeting aid is consistent with the current applications, which are overwhelmingly fixed facilities in meeting rooms. But here we also find interesting dynamics in play. Meeting size and style have a lot to do with the success of even these kinds of telepresence applications.
Users tell me that when meetings are highly interactive, and in particular when they use a white board for extemporaneous illustration of points, telepresence value falls off sharply. The problem is that the people on the other end can't read the white board and often can't follow the quick exchanges. Here the quality of the system has significant impact, though. An HD system with large viewing screens and with a very small compression-induced delay isn't as disruptive as a system with small monitors and with compression so slow that when an arm moves it disintegrates into a shower of pixels.
One point users often make in interviews about telepresence is the difficulty lining up compatible facilities. This is an indication of the importance of figuring out just what face-to-face meetings a telepresence application is supposed to displace. Do these meeting partners have a system? Is it of sufficient quality that it generates a satisfactory experience? You cannot count on building new applications for a telepresence system; your current relationships are likely going to have to justify the cost.
The attitudes of the workers involved in telepresence applications are also a factor. Virtually every organization reports early issues with telepresence, arising out of the fact that virtual meetings are socially quite different than real meetings, and work practices must be adapted to the new format. About a third of organizations I've talked to on the topic indicated that they should have phased in the telepresence program over a longer time (the median time these users believed necessary was six month). The people happiest with their telepresence programs were those who introduced it on a phased basis, with early applications focusing where the resistance to or disruption of a new approach would be minimal. Payback on telepresence should not be measured in at least the first six months of the program and the cost estimates should reflect a gradual introduction of telepresence.
The final point emerging from my own telepresence survey is that most companies will "settle for less" with telepresence technology, a decision that is likely to have truly tragic consequences. Over two-thirds of the users of "minimal" telepresence systems reported dissatisfaction with their decision, while no one we talked to with a very high-end system felt that way. It's tempting to ascribe some of this to the greater planning attention that a larger purchase is likely to involve, but I think it's clear that in all too many cases the hope of controlling costs ends up limiting benefits even more.
Telepresence isn't going to solve all your problems or eliminate all your travel, but if it's planned properly it can be a significant tool in improving communications.