Get out your pitchforks, I'm about to commit Enterprise 2.0 heresy.
There's an orthodoxy in Enterprise 2.0 circles about how you're supposed to run an implementation. The orthodoxy goes something like this: Start with small-scale pilots, define your business objectives, watch the pilots closely, evaluate their success, make a go/no-go decision. (A good recent articulation of this view is in Chris McGrath's recent post on 8 Tips for a Successful Social Intranet Pilot.)
As far as I can tell it's what everyone thinks. In fact, it's what I used to think. Unfortunately, it's dead wrong. The orthodoxy is wrong for a very simple reason: Size matters. By constraining the size of your pilot, you significantly alter the way your company can and will use the tools.
I'm not opposed to pilots for most enterprise IT solutions. Companies like to pilot new technologies with small populations before they roll them out enterprise-wide. That approach makes a lot of sense for transactional systems like order management, project management, purchasing, ERP, and so on. By piloting with a small group, you reduce implementation risk. You get a read on the value of the solution, and you get feedback which you can use to make modifications while those modifications are still relatively easy and inexpensive.
But social software is different from traditional IT. Traditional IT enables individuals to carry out well-defined, highly standardized transactions. Users go into the system to process transactions--to transfer funds, purchase supplies, track inventory, etc. The nature of these transactions, and the system's ability to enable them, doe not vary much according to the number of people using the system. Whether 100 people are entering orders or 10,000 people are entering orders, the transactions themselves doesn't really change. What that means is that a representative small-scale sample is an accurate predictor of adoption and value at full scale.
Interactions and transactions have completely different scale economics. When we use Enterprise 2.0, we're not transacting with a system; we're interacting with other people. An interaction is a connection between two or more nodes in a network. And as Metcalf's Law famously states, the more people there are in that network, the more interactions each individual can have with his or her peers, and the more value that individual derives from participation in the network.
When companies pilot Enterprise 2.0 tools with small subsets of their organization, they're not testing collaboration with a representative sample. An artificially constrained pilot is always a poor representation of post-pilot collaboration, because the range of potential interactions is so limited. The value delivered to each individual participant is exponentially smaller than it would be at full scale, and the ways that people will use the tool are different.
That doesn't mean small-scale Enterprise 2.0 pilots can't succeed. They can, and many do. But even when pilots succeed, they have limited ability to predict how the organization goes on to use the capabilities once they are rolled out enterprise-wide. Pilots typically fall into the lower left-hand corner of the Social Software Value Matrix: improve existing interactions within existing silos. That includes things like project team workspaces, departmental workspaces, and technical knowledgebases. When organizations really embrace Enterprise 2.0, however, they almost always play in multiple sections of the Value Matrix, launching solutions like collaborative intranets, ideation portals, private extranets, Those solutions, almost by definition, require scale.
Scale is the oxygen that feeds collaboration. That's why collaborative tools like Facebook, and Twitter have taken off so spectacularly on the public web: With over a billion people on the Internet, the opportunities for interpersonal interaction are unbelievably high.
From a practical standpoint, this has a counter-intuitive implication: If your E 2.0 pilot is struggling, don't shut it down. Make it bigger. Open it up. Invite more people. Tell them to invite even more people. That's the only way you're going to find out the real behavior and the real value.
If you're just starting to launch Enterprise 2.0 tools, throw the doors open wide. Invite your entire company, or as much of your company as you're allowed to invite without waiting for an act of God or Congress. Give them fun, easy ways to start contributing. (My personal favorite is microblogging.) Ease them into deeper forms of collaboration like wiki workspaces. And support those users who come forward with good ideas on how to use the tools.
Just to be clear, I'm not saying you shouldn't have clear business objectives. I'm not saying you shouldn't subject your Enterprise 2.0 implementation to critical evaluation. I'm not saying you shouldn't learn from your users and incorporate those learnings into your plans. Those are all good and important things to do. But you can only do them if you're piloting the tools in a way that you'll really learn from: at enterprise scale.