• by Bill Wise , October 2, 2006

One conclusion I've reached on the click fraud problem--the dark side of online advertising," as last week's Business Week aptly dubbed it--is that the key to fighting click fraud lies in knowing how to target your customers.

I say this because click fraud is really just a subset of a much larger problem: the problem of the non-converting searcher. And whether non-converting searchers are malicious  click-frauders, or innocent searchers who've just decided that your site's not relevant, they all wreak the same havoc: they leave you with a click cost, but offer you nothing in return. And the first step to avoiding non-converting searchers is to know who your best prospects are, and to target only to those best prospects.

After all, if you target only to conversion-likely searchers, then the searchers who aren't likely to convert--for whatever reason--will inevitably get weeded out.

I'll give you an example. Let's say you've found that your best conversions come from the Northeastern U.S., between 12 and 2 p.m. on weekdays. Now let's say that at 3 a.m. on a Sunday, someone from Nepal searches on your keyword. Based on your target-customer profile, you'll know that the Nepali searcher isn't a likely prospect, and so you won't advertise to him.

Is that Nepali searcher a would-be click frauder, hired as part of a "click farm" to damage your campaign? Or is he an innocent searcher who just happens to lie outside of your target customer group? It's impossible for you to know. And what's more, it doesn't matter. Either way, it's unlikely that he would convert for you; and by not advertising to him, you've prevented any damage he might have caused.

Of course, the example of the Nepali searcher is a fairly obvious case. And while many poor prospects are just as easily noticeable as that one is, many other cases are a lot more ambiguous. Searchers from one zip code might be great prospects, while poor prospects live just one zip code over; a Yahoo searcher might behave very differently from a Google searcher; and a Mozilla Firefox user might convert differently than an Internet Explorer user would. In other words, there are a lot of parameters that can go into identifying your target customer; and because of that, it often takes very subtle analytics to flesh out who your poor prospects are, and who will convert best.

Which is to say that if you want to avoid click fraud, you need a search firm with the analytics capability to pinpoint your best customers.

And it's useful to keep in mind that, if you do become a click-fraud victim and seek a refund from the engines, you'll need that same analytics capability to prove your case. "Proving your case" means explaining to the engines why a given searcher is so erratic, that it's reasonable to assume that something's afoul. And the better picture you have of your standard searcher, the easier it will be to make the case that your suspected click-frauder is outside of the norm. Again, your ability to profile your searchers is key to doing that, and your ability to profile is only as good as your analytics are.

Of course, profiling isn't the only defense against click fraud, and it isn't even the only one that matters. You need a system that acts quickly enough to respond to new click-frauders that appear out of the blue. You need proactive campaign management that's vigilant for any problems, and that knows how to present a solid case to the engines. You need strategists and technology who are smart enough to weigh all the factors--to know when not to deliver an ad at all, and when to keep an ad up but just lower your bid; and who can tell the difference between a click fraud problem and a landing page problem that looks like a click fraud problem to an unskilled analyst.

If all of this sounds complicated to you, that's because it is. Which is why my other piece of advice is this: if you want to avoid click fraud, your best bet is working with the smartest people. Which is good advice for any issue in search marketing that you might face.