Duncan Watts is a Yahoo researcher who has done detailed work looking at the best strategy to launch a viral campaign - the kind of campaign which is aimed at getting the consumer to pass the idea along. Watts compares campaigns which targeted viral communications via “influentials” in the social network with campaigns which simply seeded the viral in the maximum number of places in the network. Watts found that the latter was a more effective way of launching a campaign. Watts’ work - it’s argued - casts doubt on the hypothesis of Malcolm Gladwell in The Tipping Point that “influentials” are the key to successful marketing. The canonical example is that the craze for hush puppies in the 1990s was sparked by them being worn by people who were trendsetters.

Watts’ ideas are explored in an excellent article by Clive Thompson of Fast Company. Since much of our business comes from identifying who is influential and measuring how influential they are, this is a fascinating topic for us. Instinctively one feels Watts has a point. Viruses spread when a critical number of carriers are reached.

This is a common understanding amongst epidemiologists. Someone first died of AIDS in the late fifties, but the disease took 25 years to become widespread.

I think the problem here is that the marketeers who have been inspired by Malcolm Gladwell have taken several ideas and conflated them.

The first idea is that some people are “hubs” - they are well connected. (True, as far as we can tell). The second idea is that some people are influencers. (Also true, as far as we can tell). The third idea is to spread an idea - any idea which is “sticky” - you target the influencers, who are gatekeepers to the mass market. (This is an idea which is false, in our experience).

The third idea does not follow from the first two. The reasons for this are to do with how networks assign authority. Authority is - in our metrics - topic specific, it is the characteristic of being disproportionately linked by other authorities on that topic. Authorities are, by their nature, hard to target. A communicator wishing to influence an authority must tailor their message, sometimes at great pain, to make it relevant to that authority. Once it is relevant to the authority, the authority will further shape it (they are, after all, authorities) and pass it on to their network, but in their own time and manner.

This does not make influencers suitable for targeting with “viral” ideas (ideas which are clever, but to an extent generic). The marketeers notion - that influencers have the characteristic of reaching large numbers of people based simply on who they are - is simplistic. It leads to the idea that influencers are simply going to shill any idea or product - if that product is cool enough. Influencers are the last people likely to adopt anyone else’s agenda wholesale. Care must be taken to frame a topic so as to fit their agenda. Trying to apply a single marketing approach to this group does not work.

There are of course, people in the network who do carry information disproportionately - in the jargon these are the group with high “betweenness” values. These people may more closely approximate to the people Ed Keller (author of “The Influentials”) describes as “fonts of word of mouth”.

These people perhaps more closely approximate to Malcolm Gladwell’s idea of “mavens”. They are sometimes authorities, sometimes not. They tend to have more links. They tend to have a higher proportion of leaks from weaker authorities. In communications terms these folks can be just as hard to target as “influentials”, but for a different reason. Because they tend to communicate more on a topic they are less well tracked by real influencers. Their profligacy with messages makes this so. They are likely to be responsive to viral messages, but only in their key subject areas.

A good strategy for a viral marketer would be to target as many people as possible (mass communication works!) but to make sure these hubs are amongst those targeted. For this kind of communication, unless the intention is to radically change thinking in subject area, actual “authorities” can probably be ignored.

The Overture search inventory tool is one of the great little secrets of the internet. For those who have not discovered it, you can find out who is searching for what around your topic and use the information to drive your paid search choices.

We have been reviewing a particularly sulphurous quantity of consumer-generated feedback on behalf of a company which has incurred the wrath of consumers, and trying (on their behalf) to figure out what advice we should give them about managing the situation. The volume of complaints made us think it was worth looking at traffic on Overture, which captures Yahoo! searches. Fortunately it hasn’t quite got to that pitch yet.

But, intriguingly, several other companies figure in the context of the term “complaint”. Here they are in order of the most complaints. We have linked to the page you reach if you follow the link.

There are some observations one should make about this list. Big companies are more likely to figure on it than small ones. Yahoo! is going to be there, because it is Yahoo!’s website. Tech companies are more likely to figure than non tech companies. The two tech companies eBay and Yahoo! have even thought about this search and ensured that they have pages with information for customers there. eLoan complaint appears to be such a popular term that it is spammed, judging from the search results page. Pay Pal, despite being part of eBay, seem less well prepared and their number one result is Pay Pal warning which is a forum for people with cautionary tales of dealing with the company.

Vonage, Wal-mart, Cingular, Herbalife, Bank of America and McDonald’s would be well-advised to think about this search in the context of their web presence and provide customised landing pages answering the issues that consumers and journalists may wish to raise.

The Overture results on the word “complaint” point again to the undoubted lesson that if you are in customer service the search engines (Yahoo! in this case) are doing the job for you.