New here?
Unclear definitions for click fraud
John Battelle comments on a recent click fraud article in WaPo: WaPo Does the Click Fraud Piece, I Scratch My Head….
I’ve been following the click fraud discussion for a while now, and it’s a emotionally charged topic because advertisers feel like they’re being cheated. One problem that’s complicated the discussion is that people simply don’t agree on the definition of click fraud – in fact, there’s really a huge spectrum of different practices that would or would not be considered click fraud.
Here are a couple definitions of click fraud, from really clear to cloudy:
- Building a bot to click on your own ads automatically
- Clicking on your own ads to drive revenue
- Encouraging users to click on your ads
- Users double-clicking on ads
- Having a confusing user interface to drive fake clicks
- Accidental clicks from ads being too close to content
- Placing high-value ads on unrelated content in hope of clicks
I think most advertisers would consider most, if not all, of these practices click fraud. The fact is, each one of these bring in users who may be uninterested in the content behind the ad. These "unqualified" leads result in lower conversion rates, where advertisers end up footing the bill.
Publishers and ad networks, on the other hand, probably view everything after double clicks as fair game. Their definition of click fraud is much more technical in nature – as long as it’s not someone consciously committing an act of fraud, from their perspective, nothing has happened.
The truth of all of this, however, is that advertisers don’t want low converting ad spend. So if that’s caused by a ad-clicking bot, or by clicks from confusing UI, it’s all considered bad. In the long run, the only way to solve this is to implement the sort of "smart pricing" that Google does, to drive a more consistent cost-per-action. This means that, over time, ad networks will start to gravitate towards this CPA model since it gives more consistency for advertisers.
The potential issue with a CPA model versus a CPM or CPC model is that it places much more of the risk on the publisher.
Theoretically, the rewards ($) paid by the advertiser to the publisher should then increase.
I feel that the difficulty comes when you start to look at the how the numbers play out.
For examples, imagine an advertiser is paying $1 for a click (not an extreme price in today’s environment). And let us assume that the advertiser converts 1% of their traffic to their website (ignoring lifetime customer value and brand building value) to immediate sales. So essentially, they’re buying a sale for $100 via CPC.
Theoretically, you should be able convince that advertiser to now pay that $100 to the publisher. But will they? My gut says that there is going to be some significant sticker shock in the transition.
Also, because the advertiser should benefit in additional ways from gaining a visitor if not a long-term customer they should actually pay more than that $100. Do you think advertisers are going to be willing to pay for the value of those additional benefits to a total of say $125?
I’m not saying that CPA is not where the market is going. I’m just saying that it’s going to be VERY bumpy and VERY slow.
I think that looking at the affiliate industry can give you some insight into the challenges of moving to a higher publisher risk, higher publisher reward model.
Eric Mattson
26 Oct 06 at 10:26 am