Wednesday, August 14, 2013

Should we talk about sociometric influence again?

There's one particular firm that has reduced influence down to one single number.

A single, one size fits all, simple, number.
  • Some people use that number to augment their ego. 
  • Some make it a game.
  • Some dismiss it.
It's simple.

And simple wins.

And it's in part pretty much chilled most of the serious discussion we have in, in public, about segmenting populations by the impact they have on the demand curve.

The other part is the hype cycle.

Sociometric influence

The relationship between post frequency, post content, and reach is relatively well understood. Posting consistently and frequently about the same topic causes incremental reach to accrue over time. And, content, applied frequently to an audience (reach), is measurable on many social networks. It's less visible in email networks, forums, blog posts, reviews, and on Facebook, but, it's certainly there. It's observable by somebody. In the general, the relationship is known, and it's not uncommon for different people to have coefficients on that strength.

The relationship between post frequency, post content, and reach, against causing the demand curve  to shift, is not as well understood. There have been statements made about differences-in-differences effects, but there has been fewer data points on how the demand curve is affected.

The relationship between the social structure of a network, and the diffusion of a product through it, over time, is not nearly as well understood as the other two. There are a few practices that have access to that underlining data. There are fewer commercial groups that are actively hacking the graph.

If we understand these relationships, then we can improve them.

People are so much more than a bundle of transaction records. They're actual people. And it's worth understanding people through the lens of how they cause changes in each other.

The value of a single number

A single number wins every time because it's easy.

Simplicity wins every time.

You can create an ordered list, an ordered segment, of people who can predictably cause the demand curve to shift.

But for that abstraction to be an effective diagnostic tool, for it to be more useful, it has to decompose into parts. It's components, ideally, should be independent of one another. And that decomposition should have an intuition that drives it.

Should we talk about it?

When even the taxi cab driver tells you that social analytics sucks, you know that you're deep in the trough.

There is one number that dominates it all. And it's proprietary. It doesn't decompose.

I don't think that that number actually means what people want it to mean.

At what point do we build the foundations of linking actual sociometric influence back to the demand curve, and, expressing it in such a way that it is accessible?

And, moreover, what alternative do we have to the proprietary black box?

Is it time to start talking about it again?


I'm Christopher Berry