Posting a brief follow up to yesterdays post.
Just got back from the MRANE event. Right after my talk which ended on some of the work we’ve been doing with Social Network Research, John from LinkedIn spoke about using LinkedIn for B-B Sample. He mentioned how LinkedIn sample has helped their clients “cut down on the sample sizes through micro targeting”.
I thought this was rather interesting. Depending on the client, very little is known about sample size. I have clients who are perfectly happy with n=250, and some that want n=2500 for the same type of project. Many companies usually have no problems with selling more sample and charging more. I prefer to add value in design and analysis.
LinkedIn however is also very careful about sample size, because they don’t want to spam their user base with surveys. Better to sell less sample with higher margins right.
Many marketers are used to working with larger sample sizes than they need, building the sampling lan from top up rather than bottom down. LinkedIn asks clients to stop and think about it for a moment. The client is going to get to talk with VP’s at 1,000 large companies. Do they really need 1,000? According to John, some clients are accustomed to purge as much as 30% of the sample as bad (speeders etc.). While this %age sounds a little extreme to me, I do agree, that sample sizes are sometimes bigger than they actually need to be. If the client can be assured that the sample is of high quality, this may be an easier argument for smaller sample sizes than a statistical or even a cost argument.
Finally, one other thing I found interesting from John’s talk was how LinkedIn views themselves vis-à-vis the other networks (MySpace and Facebook). They see “MySpace as the Bar, LinkedIn as the Office, and Facebook as the Backyard BBQ”. Interesting way of looking at it. We’re currently doing quite a bit of research on social media. I’ll check with some of my contacts at MySpace and Facebook to see if they have similar ways of thinking about the competition?
Tom
Link to original postTom H. C. Anderson – Anderson Analytics
