The New York Times had an interesting article last weekend about
click fraud.
The article points out that click fraud seems to be a growing problem and one which is not well addressed by the market (although the article hints at the various efforts being made to combat it by different groups); however it seems to me that it misses a couple of key points, points that are also missed by some of the other blog commenters, and for that matter by the Business Week article on the same topic.
The critical advantage online advertising has over its printed alternative is that it is accountable and that you can really measure reponse rates. This is the reason why click fraud exists online and no equivalent really exists in the traditional publishing world (although inflated circulation figures may have a somewhat similar effect). Quite simply, until click it was extremely difficult for advertisers to determine the effectiveness of their advertising spend except in fairly cumbersome ways that involved fairly long delays. Hence a small amount of overpayment is probably an acceptable tradeoff for almost instant feedback about the effectiveness of the campaign. This is no different to the petty losses that many businesses suffer (from employee pilfering for example) and which they turn a blind eye to because the cost of stopping it is greater than the saving that results. So long as click fraud is a small amount of the overall spend, and, critically, so long as the conversion ratio to actual purchases is high enough to pay for the campaign, click fraud is something that businesses will accept. Of course they will (and must) look to reduce click fraud and look to recoup losses when fraud is discovered, but it is something that is manageable and an acceptable trade off.
For different businesses the level of aceptable fraud may vary because the effectiveness of the online advertising will vary. Buzzmachine noted a WSJ report that showed how careful choice of keyword advertising could make an enormous difference to a small business (the relevant part of the WSJ report comes after a piece on blogging vs PR which is also worth a read). Online advertising can help businesses reach customers which simply could not have been reached using any other method - or at least not without enormously greater expenditure - and hence tolerance for click fraud may turn out to be equally large.
If (for example) a company sells widgets for $10, makes a $1 net profit per widget and sells an average of 1000 a day before advertising on line (i.e. daily net profit of $1000) and, as is far from uncommon, sells double or triple that afterwards, then it will still benefit from online ads, so long as it makes more than $1001 net profit (i.e. after advertising and other costs are taken into account) a day it sees a benefit from its on line campaign. If sales more than double then click fraud rates that result in up to $1000 per day advertising spend still make the company money even if the genuine advertising spend were $500 / day. Now if click fraud does result in such a hypothetical doubling of advertising spend then that would be worth chasing, but given that the company is clearing over $20,000 per day insetad of the $10,000 it was before identifying whether the true spend should be $600 or $700 may not be worth it since the difference is just 1% of the increased daily turnover.
Finally it would seem to me that an operation that scanned logs looking for fraud and were paid a significant percentage of any repayments identified could be very very successful - perhaps I'd better brush up my programming skills....