How Google Inflates AdWords Bids

However, if they're bidding on the search network, Google basically pays itself.

If an ad costs Google $0.01 to serve, but provides an advertiser $1.00 worth of benefit, then possible prices for that placement are anywhere from $0.02 (providing Google with $0.01 profit and the advertiser $0.98 benefit) all the way to $0.99 (providing Google with $0.98 profit and the advertiser $0.01 benefit). A price any lower and a rational Google won't bother to offer the service, a price any higher and a rational advertiser won't bother to buy it (modulo all sorts of approximation/measurement issues, of course).

Whether a negotiation converges closer to the $0.02 or $0.99 price depends on many things, including hard factors like the number of buyers/sellers/alternatives and soft factors like traditional expectations.

The situation Google faces – a near monopoly seller, and many competitive buyers – is already pretty close to ideal for them to drive prices up to the very edge of buyers' willingness to pay. It's hard for those buyers to coordinate to keep their bids down, though over time they might via an iterative 'flocking' process try to stop donating every extra dollar (or in economic terms, the 'surplus from trade') to Google.

For example, if there are 5 highly competitive bidders, they might converge on bids $0.99, $0.98, $0.97, $0.96, $0.95 to reflect their relative desire for #1-#5 placement. But then starting with the person in 5th, they may iterative experiment if they can maintain relative placement with lower bids. Bids of $0.06, $0.05, $0.04, $0.03, $0.02 maintain the same relative order, leave Google a tiny profit, but reassign most of the benefit to the bidders.

If the buyers were to meet to try to accelerate this process, they might run afoul of antitrust law. If Google wants to have a meeting of its CS, economics, and systems-research PhDs to brainstorm ways to drive up the effective prices paid, they just need to reserve a conference room at the Googleplex.

So there are a lot of features and policies of Google AdWords that serve to nudge bids higher and prevent any flocking downward. 'Broad match' brings in competitors from nearby markets. Opaque ad/page fitness scoring keeps unsophisticated bidders guessing about the direct effects of their bid changes. Google communications encourage buyers to state their maximum, start high and work down, and donate to Google all the analytics/conversion data that will help Google discover the maximum value a buyer is receiving.

Even throwing in a few unsophisticated/irrational/novice bidders from time to time can also help: their overbids, until they realize their mistake, may help push up the bids of other more efficient businesses, who were previously collecting more of the benefit that Google could be capturing for itself.

Adding more people bidding with 'house money' (especially novice businesses increasing the number of competitors), and the amount of blindly obedient high-bidding and broad-match-bidding, helps on all these dimensions. Even if Google AdWords reps counsel these new buyers against irrational bids, so that the new bidders still receive some marginal benefit from participation, the added value-extraction from prior bidders, now paying more, could offset some or all of the promotional costs. And that's even before considering any persistent effect via new customers or nudged-higher-bid-levels.

Any of the bidders are of course always free to drop out if they're no longer gaining some benefit over not advertising at all. But the relative division of benefits is changed in Google's favor: the promotional programs can be a crafty, roundabout way of increasing prices.

That's what monopolies are good at.