The Game Theory of Online Advertising.....

This post is an expansion of a comment I posted on Nic Brisbourne's blog,

Here is the conundrum - many (most) web 2.0 businesses have a nearly totally Ad-supported business model, but users continue to find ways to avoid Ads.

And if we won't watch the ads, advertisers won't pay up, and we won't get those delightful services

Ignoring the rights and wrongs of it all, it is fairly easy to model this as a Prisoners dilemma game.

Advertiser can either go for:

- Immoderate usage - punt as many obtrusive ads as they can, payoff is this maximises revenue, but also pisses off user
- Moderate ad usage, reduces revenue but attracts user (or at least does not turn them off)

User can either:

- turn off ads, maximising pleasure but driving Advertiser to wall
- watch all ads, thus minimising pleasure but ensuring that advertisers continue to fund services.

A moment's examination will make you realise this game has two different optimal positions for Advertisers, depending on whether the Advertiser expects very few transactions or very many.

Taking the many transactions first - if the user comes back again and again, then the incentive for the Advertiser to not piss off the user is higher, because the user's lifecycle benefit is higher than the Ad benefit from any one hit. In this case, the Advertiser is best motivated to moderate the use of Ads.

If there are very few transactions - the user visits that site once or a few times only - then the Advertiser's optimal strategy is to pump as many ads as they can at the user, as this will be the higher lifecycle benefit.

The user is happy to deal with sites with moderate Ad usage, but after a few experiences of the Immoderate users, the User adopts Ad blocking software, and from then on no Ads are seen

This then sets up a classic tragedy of the commons effect - the impact of the immoderate Advertisers makes all users adopt Ad-blocking, thus killing the game for all the moderate Advertisers too.

So, by default the game reverts to the "Block Ads" quadrant after a few cycles, to the detriment of all.

Well, not quite - it would seem that we are very capable of discriminating - for example, nearly everyone has popup blocking software, but few people bother with display ad blocking or classifieds blocking.

So the empirical message for the Advertiser is this - be moderate, and you will get users colluding with your advertising - be Immoderate, and you will not. Find the Advertising "Laffer Curve" by staying on the good side of it.

The issue moderate Advertising sites face ongoing is how to stop Immoderate ones from chasing away users as soon as a new consensus re-emerges, because the temptation to cheat in a commons environment is always too strong

It is in practice probably impossible to punish the Immoderate advertisers, and probably impossible to use Coasian approaches and pay them to be moderate.

The longer game is that Moderate Advertisers must find a way of signalling to users that they are moderate, and that the user can therefore collude in switching off the Ad blockers safely, for mutual long term benefit.

One can imagine voluntary codes of conduct, kitemarks for moderate users, educating brands not to advertise on Immoderate sites etc.

But given the tragedy of these commons, I suspect that the endgame will be some form of direct transfer of benefit to the user to switch Ad blocking off - in other words the Ad Blocker will negotiate with the Advertiser to get a benefit - for example, an Immoderate site may find that it gets hit with a "pay per click" cost every time it serves up a popup onto a page.

Clearly the more attractive the content, the less the Advertiser will pay to the user...but the probable outcome for "me too" sites is that they will have to pay over the odds for user attention in some way or another.

The 'Web is a point to point medium, not a broadcast medium and at some point the value of each link will be traded.