Neil Thackray’s Business Media Blog

Just another WordPress.com weblog

Making Online Advertising More Effective

Image representing Phorm as depicted in CrunchBase
Image via CrunchBase

The controversy around the privacy issues that arise from behavioural advertising targetting keeps rearing its head.  According to a report in the Guardian, many Internet companies are considering a boycott of the solution being offered by Phorm, including Google.  Some might say that Google has a vested interest in the failure of Phorm as it butts straight into their own revenue model and approach to improving advertising effectiveness.

Let’s think for a moment about what this might mean for business media.  First, can we agree that there is something wrong with the existing online advertising model? Put to one side how successful or otherwise we are in selling our inventory, a quick look at your results from Google Adsense will tell you all you need to know.  Back in the late nineties, the dot com bubble burst, not least because the extravagent valuations of Internet start ups could not be sustained by the slow rate of growth in revenues – much of it advertising revenues.  Click through rates were horribly low.  Google changed all that.  The ability to deliver ads which were relevant to the content of the page suddenly created a whole new business.

For a while we deluded ourselves that all was solved.  But things are beginning to look like 1999 all over again.  Three years ago, oneof the websites I ran was achieving an ecpm from Google of around $10.  That same site today is struggling to achieve $2.  I have no way of knowing whether this is typical, but we also observe that average ctr for all kinds of online display advertising is falling.  When we were developing our vertical search engine, we learned from the technology market advertisers what the average CTR was, across all their online campaigns.  In the interests of commercial sensitivity I can’t tell you the numbers, but let’s just say they are not large.

The implications of this are serious, and we have touched on them before.  To achieve advertising breakthrough clients are resorting to buying networks.  If the CTR is low, then you need to buy a lot of traffic.  As a result the rates are low.  It is an empirical fact that none of us can make a living from the rates achieved from backfill and ad networks.

 

So what to do about it?  Behavioural targeting is certainly part of the answer and I suspect the privacy issues are less sginificant in b2b than they are in b2c.  The key to the success of this will be the ability to make the advertising experience relevant to the user and that requires a taxonomy which allows the ad server to understand the relevance of the document.  Key words won’t do it.  The approach has to be smarter than that.  The next generation of advertising will be closer to the holy grail of semantic search than the last.  It will require an understanding of the meaning of the content, the ability to understand related concepts and facets.  

Understanding the meaning of documents is at the heart of what the enterprise search companies try to do – most notably Fast and Autonomy.  But there are real issues for these technologies in the kind of applications we are talking about.  In the enterprise search world, most documents are fixed, with only documents at the edges changing.  In web sites, most of the current accessed pages change all the time.  What is more, each business vertical requires a unique taxonomy otherwise users experience too many false positive results.

So let’s imagine we can solve these issues.  If we could understand the meaning of a document, and the possible related concepts and then uses this data to call the ads, is it possible that our customers will see a meaningful improvement in their CTR.  Could we meld with that the click stream information from the session and parse both data sets through the ad server?  Wouldn’t that be cool?

Apart from Google, there is a growing band of specialists thinking about this problem.  Think Wunderloop or Grapeshot as just two interesting examples.

Reblog this post [with Zemanta]

April 3, 2009 Posted by | Advertising Sales, business media strategy, Uncategorized | , , , , , , , | Leave a comment