Digital advertising: November 2007 Archives

About a week ago I spoke at Netimperative's Online Publishing and Media Summit, and since my talk has been discussed by Andrew Girdwood and David Cushman I thought it was time I got in on the act.

I was talking about how publishers need to adapt different business models to cope with the threat of search. My take on this is that for many online purchases the very last click before purchase will often be via a search engine.  And that last click is very often credited with 100% of the value of a sale.  So the CPA on search looks artificially low - and any other activity elsewhere in the customer journey, like banners, will look correspondingly expensive.

One way of looking at this is at the halo effect research conducted by Atlas. This research found that where a customer has been previously exposed to online display activity, conversion rates through search are 22% higher than someone coming through search alone. 

That 22% figure is an average across eleven advertisers.  In fact the highest reported uplift by using display and search together was around 65% (and, in the interests of fairness, one unlucky advertiser managed to depress response rates slightly with its display activity).

This isn't really an anti-Google argument, as many people seemed to think at the event.  In fact I've seen Google present the same Atlas research to justify why customers should spread spend into display activity - like Google's own advertising network for instance.

But it is an argument that publishers need to get their heads around.  Unless they can persuade clients and agencies to take a more sophisticated view of online conversion attibution models, budgets will continue to drift towards search and away from display advertising.

Where does Mahalo come in? Basically search is so important to online advertising that any new models - however experimental - have to be of interest.  I like Mahalo because it promises the best of both worlds - user crafted search pages for top searches, but defaulting to Google for the long tail of searches where Google excels.

And as the founder of Mahalo, Jason Calacanis, points out in a comment to Andrew Girdwood's blog, another advantage of a review based engine is that publishers can challenge their SERPs ranking and get it changed if their case has merit.  Wow!  

We had a meeting with our rep from Facebook a week or so ago, and she mentioned something to me that is probably screamingly obvious, but had never occured to me.

The average person on Facebook has around 160 friends, and each of those friends are doing between five and ten actions each day that could potentially go into a news feed.  So clearly Facebook is doing some clever filtering to make sure that the twenty or so items in your news feed are actually interesting to you.

Their algorithm is looking for people in your network that you seem to have the closest connection with.  It will be looking at cross-posting on each other's pages, tagging of photos, numbers of friends in common etc.

The Tipping Point talks about mavens, who are often the first to pick up on new trends, and connectors, who help mavens to spread their message.  And the definition of connectors sounds a lot like these 'hot nodes' on Facebook: this definition from Wikipedia:

"...people who have wide network of casual acquaintances by whom they are trusted, often a network that crosses many social boundaries and groups."

So if you want to get actions into the news feed - as many applications do - you particularly need to be appealing to these 'Facebook connectors'.  Naturally there's already a Facebook application, socialistics, that will tell you who in your network is most relevant to you (in case you can't work it out for yourself!)