On Measuring Influence

Yesterday’s word of the day, apparently, was “influence.” At least that’s what was on the mind of a few big bloggers. Steve Rubel started the conversation by declaring dead the notion that link count equals influence. He argues that counting inbound links is irrelevant outside the blogosphere – in Facebook or Twitter for example – where many conversations go down. His bottom line is that using inbound links as the barometer of influence misses too much.

I disagree with Rubel’s suggestion that link count is or has ever been synonymous with influence. I’ve always seen it as just part of the formula. Maybe a disproportionately large part, but only because there’s so little else that can be quantified. Traffic data can tell you the reach of a source – which is another mark of influence – but publicly-available traffic data is pretty unreliable. Right now, Rubel is right to point out that the formula for measuring influence is oversimplified, but I don’t think it’s due to a lack of sophistication on the part of those who seek to do so. The problem is a lack of measurable data, and as that improves over time, so will influence metrics.

A companion piece by Rubel’s colleague, David Brain, lays out the beginnings of a more nuanced formula for measuring influence that starts with the top 30 bloggers (using rankings published by various sources) and attempts to factor in a variety of social networking activities carried out by said bloggers, as follows:

  • Blog – analysed Google Rank, inbound links, subscribers, alexa rank, content focus, frequency of updates, number of comments
  • Multi-format – analysed Facebook – number of friends
  • Mini-updates – analysed Twitter – number of friends, followers and updates
  • Business cards – analysed LinkedIn – number of contacts
  • Visual – analysed Flickr – number of photos uploaded from the person/s or about the person/s
  • Favourites – analysed Digg, del.icio.us

This implicitly accounts for some notion of reach, along with some rather ad-hoc and unscientific social media metrics, but since it’s limited to just 30 blogs, it fails to address anything but the most general questions around influence and influencers. Many of the top blogs are technology or marketing focused, so while they might be highly influential to a few key audiences, they are totally irrelevant to, say, urban teens or middle-class parents of toddlers.

Rubel and David Brain also touch on the “Facebook phenomenon,” and there’s no question that Facebook, MySpace, Twitter and the other buzzing social services deserve a place in any discussion about influence, but what place? Because these services enjoy such a high profile right now, it’s tempting to overestimate or misunderstand how they factor in. Facebook user profiles are not accessible to the general public or even to other Facebook members who are not connected as friends – unlike the top bloggers, who enjoy large reader audiences and a lot of visitor traffic. So the only way influence can exist on Facebook is virally – with memes spreading gradually across overlapping circles of friends, through direct contact between individuals. Basically like the real world, and probably just as difficult to predict, detect and measure.

Jeff Jarvis chimes in and pokes fun at Rubel, calling him a sort of “grim reaper of measurements” who “likes declaring things dead.” On the subject of influence, Jarvis discusses some of the complexity involved in quantifying it. He points out that there are different and dynamic spheres of influence that have to do with things like an individual person’s reputation, subject matter expertise and credentials, nature and reach of their forum (e.g. traditional media vs. blog), and nature of the audience.

This brings us to a more fundamental problem that underlies the challenge of measuring influence. Namely, that there’s no consensus on what “influence” actually means. What do we want to measure? What exactly are these mysterious influencers influencing? Brand perception? Purchase decisions? And whom do they influence? Furthermore, a lot of the work on this problem is focused on identifying influential sources – people or publications that tend to influence other people. But a single well-written product or movie review, or one disastrous customer service call can be hugely influential, regardless of the author’s credentials.

It’s clear that no one has influence figured out, which makes it an endlessly fascinating problem area. We’ll keep tinkering with our own formula, and as we’re able to factor in more data, our influence metrics will get better and better.