Peak advertising

14 October 2013

Peak advertising

This article by Tim Hwang and Adi Kamdar argues that online advertising will become less efficient over time; the argument is that

  • younger people do not react to advertising as much as older people
  • saturation: lots of adds everywhere, so nobody gives a damn;
  • add blocking is a result of that.
  • click fraud: the opportunity makes the thieve.

As a result they see the following trends

  • advertising is concentrating around a few monopolies (also known as ‘add platforms’)
  • invasions of privacy by add platforms: add platforms want to know more about us in order to make more efficient add placements; they create a model of our interest and want to match the adds to this model of our ‘online self’

They forecast that we will see more adds as part of the online material (payed for articles as adds).

Is that really so?

It may be all true, but the competition is also bleeding: Cable TV companies seem to be suffering from falling subscriptions (here ) So the audience for online content will be growing, as a result of the problems that traditional media is facing.

What seems to be true that

  • lately Google is pushing much more adds in youtube.
  • Search results are very different when the Google search engine detects an intention to buy something; in this case there are much more adds; Compare search for smartphone and buy smartphone - the second search is heavier on adds.
  • So far everything is fine for Google; the share price is up ;

Web adds might not be very effective, but it may well take a lot of time until the advertisers have realized that. The perceived value may outlast the real value of the tool.

Is there another way?

Is there a better way to advertise, one that does not require these companies to gather all scraps of data that it can get hold off. Is it really that important to spy after the users?

  • There is the possibility of merging online content and advertising: Look at Amazon, while looking at book reviews, one also gets an add for related books “People who bought this book also bought …” Amazon is making a lot of money around that offer.
  • It may be enough to understand the online content, and to pair in adds that are related to online content.
  • A better understanding of the served content vs served adds would require less intrusive user data gathering on the part of the ‘add platforms’
  • Also there seems to be at least some degree of consumer backslash: The search engine Duck duck go promises more privacy than its competitors; So thanks to the Snowden revelations it has seen some remarkable growth So far that may be a “nerdy” trend, but over times “nerdy” trends tend to be accepted by the larger public.
  • Placement of adds based on the served online content is quite a technological challenge, this would open up opportunities for newer players.

Well, I think it is hard to tell, when tech is involved you have a constantly changing picture…

Isn’t that too hard?

Currently the adds are separate from the content – I think its a matter of feasibility.

  • Placement of adds can be made on popular pages, the relevantly paired content may not get any views at all!
  • Current system of paring adds onto a constructed user profile is a task of modern practical data driven AI. Now merging of adds and content would require machines that are able to adjust to the meaning of a text; the machine would have to somehow understand what is written in the text in order to put in a relevant add; Now that’s really hard.

Maybe the current ways will reach a limit (don’t know when); it seems that at some stage modern data driven AI will reach its limit; when this happens we may need real machine intelligence;

However the road to real machine intelligence is a long one, so far only Douglas Hofstadter is looking at that problem. Also this article describes the different concepts of AI well.